add sdpa with sinks

* faster contiguous gather for indices in the first axis

* work per thread > 1

* angelos suggestion for scales / biases

.circleci/config.yml

@@ -7,15 +7,9 @@ parameters:
   nightly_build:
     type: boolean
     default: false
-  weekly_build:
-    type: boolean
-    default: false
   test_release:
     type: boolean
     default: false
-  linux_release:
-    type: boolean
-    default: false
 
 jobs:
   build_documentation:
@@ -24,13 +18,14 @@ jobs:
         type: boolean
         default: false
     macos:
-      xcode: "15.2.0"
-    resource_class: macos.m1.medium.gen1
+      xcode: "26.0.0"
+    resource_class: m4pro.medium
     steps:
       - checkout
       - run:
           name: Install
           command: |
+            xcodebuild -downloadComponent MetalToolchain
             brew install python@3.9
             brew install doxygen
             python3.9 -m venv env
@@ -38,7 +33,7 @@ jobs:
             pip install --upgrade pip
             pip install --upgrade cmake
             pip install -r docs/requirements.txt
-            CMAKE_BUILD_PARALLEL_LEVEL=`sysctl -n hw.ncpu` pip install . -v
+            pip install . -v
       - when:
           condition:
             not: << parameters.upload-docs >>
@@ -70,9 +65,9 @@ jobs:
                  git push -f origin gh-pages
 
   linux_build_and_test:
-    docker:
-      - image: cimg/python:3.9
-
+    machine:
+      image: ubuntu-2204:current
+      resource_class: large
     steps:
       - checkout
       - run:
@@ -84,34 +79,37 @@ jobs:
       - run:
           name: Install dependencies
           command: |
-            pip install --upgrade cmake
-            pip install nanobind==2.4.0
-            pip install numpy
+            export DEBIAN_FRONTEND=noninteractive
+            export NEEDRESTART_MODE=a
             sudo apt-get update
-            sudo apt-get install libblas-dev liblapack-dev liblapacke-dev
+            sudo apt-get install -y libblas-dev liblapack-dev liblapacke-dev
+            sudo apt-get install openmpi-bin openmpi-common libopenmpi-dev
+            curl -LsSf https://astral.sh/uv/install.sh | sh
       - run:
           name: Install Python package
           command: |
-            CMAKE_ARGS="-DMLX_BUILD_METAL=OFF" \
-              CMAKE_BUILD_PARALLEL_LEVEL=`nproc` \
-              python3 setup.py build_ext --inplace
-            CMAKE_ARGS="-DMLX_BUILD_METAL=OFF" \
-              CMAKE_BUILD_PARALLEL_LEVEL=`nproc` \
-              python3 setup.py develop
+            uv venv
+            uv pip install cmake
+            DEBUG=1 CMAKE_ARGS="-DCMAKE_COMPILE_WARNING_AS_ERROR=ON" \
+              uv pip install -e ".[dev]" -v
       - run:
           name: Generate package stubs
           command: |
-            echo "stubs"
-            pip install typing_extensions
-            python setup.py generate_stubs 
+            uv pip install typing_extensions
+            uv run --no-project setup.py generate_stubs
       - run:
           name: Run Python tests
           command: |
-            python3 -m unittest discover python/tests -v
+            source .venv/bin/activate
+            python -m unittest discover python/tests -v
+            mpirun --bind-to none -host localhost:8 -np 8 python python/tests/mpi_test_distributed.py
+            mlx.launch --verbose -n 8 python/tests/ring_test_distributed.py -v 2> >(tee -a stderr.log >&2)
+            if $(grep "\[WARN\]" stderr.log); then echo "Distributed ring test failed"; exit 1; fi
       - run:
           name: Build CPP only
           command: |
-            mkdir -p build && cd build 
+            source .venv/bin/activate
+            mkdir -p build && cd build
             cmake .. -DMLX_BUILD_METAL=OFF -DCMAKE_BUILD_TYPE=DEBUG
             make -j `nproc`
       - run:
@@ -122,58 +120,64 @@ jobs:
     parameters:
       xcode_version:
         type: string
-        default: "15.2.0"
+        default: "26.0.0"
+      macosx_deployment_target:
+        type: string
+        default: ""
     macos:
       xcode: << parameters.xcode_version >>
-    resource_class: macos.m1.medium.gen1
+    environment:
+      MACOSX_DEPLOYMENT_TARGET: << parameters.macosx_deployment_target >>
+    resource_class: m4pro.medium
     steps:
       - checkout
       - run:
           name: Install dependencies
           command: |
-            brew install python@3.9
-            brew install openmpi
-            python3.9 -m venv env
-            source env/bin/activate
-            pip install --upgrade pip
-            pip install --upgrade cmake
-            pip install nanobind==2.4.0
-            pip install numpy
-            pip install torch
-            pip install tensorflow
-            pip install unittest-xml-reporting
+            xcodebuild -downloadComponent MetalToolchain
+            HOMEBREW_NO_AUTO_UPDATE=1 HOMEBREW_NO_INSTALL_CLEANUP=1 \
+              brew install openmpi uv
       - run:
           name: Install Python package
           command: |
-            source env/bin/activate
-            DEBUG=1 CMAKE_BUILD_PARALLEL_LEVEL=`sysctl -n hw.ncpu` pip install -e . -v
+            uv venv --python 3.9
+            uv pip install \
+              nanobind==2.4.0 \
+              cmake \
+              numpy \
+              torch \
+              tensorflow \
+              unittest-xml-reporting
+            DEBUG=1 CMAKE_ARGS="-DCMAKE_COMPILE_WARNING_AS_ERROR=ON" \
+              uv pip install -e . -v
       - run:
           name: Generate package stubs
           command: |
-            source env/bin/activate
-            pip install typing_extensions
-            python setup.py generate_stubs 
+            uv pip install typing_extensions
+            uv run --no-project setup.py generate_stubs
       - run:
           name: Run Python tests
           command: |
-            source env/bin/activate
+            source .venv/bin/activate
             LOW_MEMORY=1 DEVICE=cpu python -m xmlrunner discover -v python/tests -o test-results/cpu
             LOW_MEMORY=1 DEVICE=gpu METAL_DEVICE_WRAPPER_TYPE=1 METAL_DEBUG_ERROR_MODE=0 python -m xmlrunner discover -v python/tests -o test-results/gpu
             mpirun --bind-to none -host localhost:8 -np 8 -x DYLD_LIBRARY_PATH=/opt/homebrew/lib/ python python/tests/mpi_test_distributed.py
-            mlx.launch --verbose -n 8 python/tests/ring_test_distributed.py
+            mlx.launch --verbose -n 8 python/tests/ring_test_distributed.py -v 2> >(tee -a stderr.log >&2)
+            if $(grep "\[WARN\]" stderr.log); then echo "Distributed ring test failed"; exit 1; fi
       - run:
           name: Build example extension
           command: |
-            source env/bin/activate
+            source .venv/bin/activate
             cd examples/extensions
-            pip install -r requirements.txt
-            python setup.py build_ext -j8
+            uv pip install -r requirements.txt
+            uv run --no-project setup.py build_ext --inplace
+            uv run --no-project python test.py
       - store_test_results:
           path: test-results
       - run:
           name: Build CPP only
           command: |
-            source env/bin/activate
+            source .venv/bin/activate
             mkdir -p build && cd build && cmake .. && make -j `sysctl -n hw.ncpu`
       - run:
           name: Run CPP tests
@@ -182,7 +186,7 @@ jobs:
       - run:
           name: Build small binary
           command: |
-            source env/bin/activate
+            source .venv/bin/activate
             cd build/
             cmake .. -DCMAKE_BUILD_TYPE=MinSizeRel \
               -DBUILD_SHARED_LIBS=ON \
@@ -194,13 +198,74 @@ jobs:
       - run:
           name: Run Python tests with JIT
           command: |
-            source env/bin/activate
-            CMAKE_BUILD_PARALLEL_LEVEL=`sysctl -n hw.ncpu` \
-              CMAKE_ARGS="-DMLX_METAL_JIT=ON" \
-              pip install -e . -v
+            CMAKE_ARGS="-DMLX_METAL_JIT=ON" \
+              uv pip install -e . -v
             LOW_MEMORY=1 DEVICE=gpu METAL_DEVICE_WRAPPER_TYPE=1 \
               METAL_DEBUG_ERROR_MODE=0 \
-              python -m xmlrunner discover -v python/tests -o test-results/gpu_jit
+              uv run --no-project python -m xmlrunner discover \
+                -v python/tests \
+                -o test-results/gpu_jit
+
+  cuda_build_and_test:
+    parameters:
+      image_date:
+        type: string
+        default: "2023.11.1"
+    machine:
+      image: "linux-cuda-12:<< parameters.image_date >>"
+      resource_class: gpu.nvidia.small.gen2
+    steps:
+      - checkout
+      - restore_cache:
+          keys:
+            - cuda-<< parameters.image_date >>-{{ arch }}-
+      - run:
+          name: Install dependencies
+          command: |
+            sudo apt-get update
+            sudo apt-get install libcudnn9-dev-cuda-12
+            sudo apt-get install libblas-dev liblapack-dev liblapacke-dev
+            sudo apt-get install libnccl2 libnccl-dev
+            curl -sL https://github.com/ccache/ccache/releases/download/v4.11.3/ccache-4.11.3-linux-x86_64.tar.xz | tar xJf -
+            sudo mv ccache-4.11.3-linux-x86_64/ccache /usr/bin/ccache
+            rm -rf ccache-4.11.3-linux-x86_64
+            curl -LsSf https://astral.sh/uv/install.sh | sh
+      - run:
+          name: Install Python package
+          command: |
+            uv venv
+            uv pip install cmake
+            DEBUG=1 CMAKE_ARGS="-DMLX_BUILD_CUDA=ON -DCMAKE_COMPILE_WARNING_AS_ERROR=ON -DCMAKE_CUDA_COMPILER=`which nvcc`" \
+              uv pip install -e ".[dev]" -v
+      - run:
+          name: Run Python tests
+          command: |
+            source .venv/bin/activate
+            LOW_MEMORY=1 DEVICE=cpu python -m unittest discover python/tests -v
+            LOW_MEMORY=1 DEVICE=gpu python -m tests discover python/tests -v
+      - run:
+          name: Build CPP only
+          command: |
+            source .venv/bin/activate
+            cmake . -B build \
+              -DMLX_BUILD_CUDA=ON \
+              -DCMAKE_CUDA_COMPILER=`which nvcc` \
+              -DCMAKE_BUILD_TYPE=DEBUG
+            cmake --build build -j `nproc`
+      - run:
+          name: Run CPP tests
+          command: ./build/tests/tests -sfe="*fft_tests.cpp,*linalg_tests.cpp"
+      - run:
+          name: CCache report
+          command: |
+            ccache --show-stats
+            ccache --zero-stats
+            ccache --max-size 400MB
+            ccache --cleanup
+      - save_cache:
+          key: cuda-<< parameters.image_date >>-{{ arch }}-{{ epoch }}
+          paths:
+            - /home/circleci/.cache/ccache
 
   build_release:
     parameters:
@@ -209,23 +274,32 @@ jobs:
         default: "3.9"
       xcode_version:
         type: string
-        default: "15.2.0"
+        default: "26.0.0"
       build_env:
         type: string
         default: ""
+      macosx_deployment_target:
+        type: string
+        default: ""
     macos:
       xcode: << parameters.xcode_version >>
-    resource_class: macos.m1.medium.gen1
+    resource_class: m4pro.medium
+    environment:
+      MACOSX_DEPLOYMENT_TARGET: << parameters.macosx_deployment_target >>
     steps:
       - checkout
       - run:
           name: Install dependencies
           command: |
-            brew install python@<< parameters.python_version >>
-            brew install openmpi
-            python<< parameters.python_version >> -m venv env
-            source env/bin/activate
-            pip install --upgrade pip
+            xcodebuild -downloadComponent MetalToolchain
+            mkdir -p ~/miniconda3
+            curl https://repo.anaconda.com/miniconda/Miniconda3-latest-MacOSX-arm64.sh -o ~/miniconda3/miniconda.sh
+            bash ~/miniconda3/miniconda.sh -b -u -p ~/miniconda3
+            rm ~/miniconda3/miniconda.sh
+            source ~/miniconda3/bin/activate
+            conda init --all
+            conda create -n env python=<< parameters.python_version >> -y
+            conda activate env
             pip install --upgrade cmake
             pip install nanobind==2.4.0
             pip install --upgrade setuptools
@@ -235,30 +309,38 @@ jobs:
       - run:
           name: Install Python package
           command: |
-            source env/bin/activate
-            DEV_RELEASE=1 \
-              CMAKE_BUILD_PARALLEL_LEVEL=`sysctl -n hw.ncpu` \
+            conda activate env
+            env -u MACOSX_DEPLOYMENT_TARGET DEV_RELEASE=1 \
               pip install . -v
       - run:
           name: Generate package stubs
           command: |
-            source env/bin/activate
+            conda activate env
             pip install typing_extensions
-            python setup.py generate_stubs 
+            python setup.py generate_stubs
       - run:
           name: Build Python package
           command: |
-            source env/bin/activate
-            << parameters.build_env >> \
-              CMAKE_BUILD_PARALLEL_LEVEL=`sysctl -n hw.ncpu` \
-              python -m build -w
+            conda activate env
+            python setup.py clean --all
+            << parameters.build_env >> MLX_BUILD_STAGE=1 python -m build -w
+      - when:
+          condition:
+            equal: ["3.9", << parameters.python_version >>]
+          steps:
+            - run:
+                name: Build common package
+                command: |
+                  conda activate env
+                  python setup.py clean --all
+                  << parameters.build_env >> MLX_BUILD_STAGE=2 python -m build -w
       - when:
           condition: << parameters.build_env >>
           steps:
             - run:
                 name: Upload package
                 command: |
-                  source env/bin/activate
+                  conda activate env
                   twine upload dist/*
       - store_artifacts:
           path: dist/
@@ -268,52 +350,100 @@ jobs:
       python_version:
         type: string
         default: "3.9"
-      extra_env:
+      build_env:
         type: string
-        default: "DEV_RELEASE=1"
-    docker:
-      - image: ubuntu:20.04
+        default: ""
+    machine:
+      image: ubuntu-2204:current
+      resource_class: large
     steps:
       - checkout
       - run:
           name: Build wheel
           command: |
             PYTHON=python<< parameters.python_version >>
-            apt-get update
-            apt-get upgrade -y
-            DEBIAN_FRONTEND=noninteractive TZ=Etc/UTC apt-get -y install tzdata
-            apt-get install -y apt-utils
-            apt-get install -y software-properties-common
-            add-apt-repository -y ppa:deadsnakes/ppa
-            apt-get install -y $PYTHON $PYTHON-dev $PYTHON-full
-            apt-get install -y libblas-dev liblapack-dev liblapacke-dev
-            apt-get install -y build-essential git
+            export DEBIAN_FRONTEND=noninteractive
+            export NEEDRESTART_MODE=a
+            sudo apt-get update
+            TZ=Etc/UTC sudo apt-get -y install tzdata
+            sudo add-apt-repository -y ppa:deadsnakes/ppa
+            sudo apt-get install -y $PYTHON $PYTHON-dev $PYTHON-full
+            sudo apt-get install -y libblas-dev liblapack-dev liblapacke-dev
             $PYTHON -m venv env
             source env/bin/activate
             pip install --upgrade pip
             pip install --upgrade cmake
-            pip install nanobind==2.4.0
-            pip install --upgrade setuptools
-            pip install numpy
             pip install auditwheel
             pip install patchelf
             pip install build
             pip install twine
-            << parameters.extra_env >> \
-              CMAKE_BUILD_PARALLEL_LEVEL=`nproc` \
-              pip install . -v
+            << parameters.build_env >> pip install ".[dev]" -v
             pip install typing_extensions
-            python setup.py generate_stubs 
-            << parameters.extra_env >> \
-              CMAKE_BUILD_PARALLEL_LEVEL=`nproc` \
-              python -m build --wheel
-            auditwheel show dist/*
-            auditwheel repair dist/* --plat manylinux_2_31_x86_64
+            python setup.py generate_stubs
+            python setup.py clean --all
+            MLX_BUILD_STAGE=1 << parameters.build_env >> python -m build -w
+            bash python/scripts/repair_linux.sh
+      - when:
+          condition:
+            equal: ["3.9", << parameters.python_version >>]
+          steps:
+            - run:
+                name: Build common package
+                command: |
+                  source env/bin/activate
+                  python setup.py clean --all
+                  << parameters.build_env >> MLX_BUILD_STAGE=2 \
+                    python -m build -w
+                  auditwheel repair dist/mlx_cpu*.whl --plat manylinux_2_35_x86_64
+      - when:
+          condition: << parameters.build_env >>
+          steps:
+            - run:
+                name: Upload packages
+                command: |
+                  source env/bin/activate
+                  twine upload wheelhouse/*.whl
+      - store_artifacts:
+          path: wheelhouse/
+
+  build_cuda_release:
+    parameters:
+      build_env:
+        type: string
+        default: ""
+    machine:
+      image: ubuntu-2204:current
+      resource_class: xlarge
+    steps:
+      - checkout
       - run:
-          name: Upload package
+          name: Build wheel
           command: |
-            source env/bin/activate
-            twine upload wheelhouse/*
+            export DEBIAN_FRONTEND=noninteractive
+            export NEEDRESTART_MODE=a
+            wget https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2404/x86_64/cuda-keyring_1.1-1_all.deb
+            sudo dpkg -i cuda-keyring_1.1-1_all.deb
+            sudo apt-get update
+            sudo apt-get install cuda-toolkit-12-9 libcudnn9-dev-cuda-12
+            sudo apt-get install libblas-dev liblapack-dev liblapacke-dev
+            sudo apt-get install zip
+            pip install auditwheel
+            pip install patchelf
+            pip install build
+            pip install twine
+            export PATH=/usr/local/cuda/bin${PATH:+:${PATH}}
+            export LD_LIBRARY_PATH=/usr/local/cuda/lib64${LD_LIBRARY_PATH:+:${LD_LIBRARY_PATH}}
+            << parameters.build_env >> MLX_BUILD_STAGE=2 \
+              CMAKE_ARGS="-DMLX_BUILD_CUDA=ON -DCMAKE_CUDA_COMPILER=`which nvcc`" \
+              python -m build -w
+            bash python/scripts/repair_cuda.sh
+      - when:
+          condition: << parameters.build_env >>
+          steps:
+            - run:
+                name: Upload package
+                command: |
+                  twine upload wheelhouse/*.whl
       - store_artifacts:
           path: wheelhouse/
 
@@ -325,21 +455,23 @@ workflows:
             pattern: "^(?!pull/)[-\\w]+$"
             value: << pipeline.git.branch >>
         - not: << pipeline.parameters.nightly_build >>
-        - not: << pipeline.parameters.weekly_build >>
         - not: << pipeline.parameters.test_release >>
     jobs:
       - mac_build_and_test:
           matrix:
             parameters:
-              xcode_version: ["15.0.0", "15.2.0", "16.0.0"]
+              macosx_deployment_target: ["13.5", "15.0"]
       - linux_build_and_test
+      - cuda_build_and_test:
+          matrix:
+            parameters:
+              image_date: ["2023.11.1", "2025.05.1"]
       - build_documentation 
 
   build_pypi_release:
     when:
       and:
         - not: << pipeline.parameters.nightly_build >>
-        - not: << pipeline.parameters.weekly_build >>
         - not: << pipeline.parameters.test_release >>
     jobs:
       - build_release:
@@ -351,8 +483,9 @@ workflows:
           matrix:
             parameters:
               python_version: ["3.9", "3.10", "3.11", "3.12", "3.13"]
-              xcode_version: ["15.0.0", "15.2.0"]
+              macosx_deployment_target: ["13.5", "14.0", "15.0"]
               build_env: ["PYPI_RELEASE=1"]
+              xcode_version: ["26.0.0"]
       - build_documentation:
           filters:
             tags:
@@ -360,6 +493,25 @@ workflows:
             branches:
               ignore: /.*/
           upload-docs: true
+      - build_linux_release:
+          filters:
+            tags:
+              only: /^v.*/
+            branches:
+              ignore: /.*/
+          matrix:
+            parameters:
+              python_version: ["3.9", "3.10", "3.11", "3.12", "3.13"]
+              build_env: ["PYPI_RELEASE=1"]
+      - build_cuda_release:
+          filters:
+            tags:
+              only: /^v.*/
+            branches:
+              ignore: /.*/
+          matrix:
+            parameters:
+              build_env: ["PYPI_RELEASE=1"]
 
   prb:
     when:
@@ -375,9 +527,14 @@ workflows:
           requires: [ hold ]
           matrix:
             parameters:
-              xcode_version: ["15.0.0", "15.2.0", "16.0.0"]
+              macosx_deployment_target: ["13.5", "15.0"]
       - linux_build_and_test:
           requires: [ hold ]
+      - cuda_build_and_test:
+          requires: [ hold ]
+          matrix:
+            parameters:
+              image_date: ["2023.11.1", "2025.05.1"]
   nightly_build:
     when:
       and:
@@ -388,27 +545,33 @@ workflows:
           matrix:
             parameters:
               python_version: ["3.9", "3.10", "3.11", "3.12", "3.13"]
-              xcode_version: ["15.0.0", "15.2.0"]
-  weekly_build:
+              macosx_deployment_target: ["13.5", "14.0", "15.0"]
+              xcode_version: ["26.0.0"]
+      - build_linux_release:
+          matrix:
+            parameters:
+              python_version: ["3.9", "3.10", "3.11", "3.12", "3.13"]
+      - build_cuda_release
+
+  build_dev_release:
     when:
       and:
         - equal: [ main, << pipeline.git.branch >> ]
-        - << pipeline.parameters.weekly_build >>
+        - << pipeline.parameters.test_release >>
     jobs:
       - build_release:
           matrix:
             parameters:
               python_version: ["3.9", "3.10", "3.11", "3.12", "3.13"]
-              xcode_version: ["15.0.0", "15.2.0", "16.0.0"]
+              macosx_deployment_target: ["13.5", "14.0", "15.0"]
               build_env: ["DEV_RELEASE=1"]
-  linux_test_release:
-    when:
-      and:
-        - equal: [ main, << pipeline.git.branch >> ]
-        - << pipeline.parameters.linux_release >>
-    jobs:
+              xcode_version: ["26.0.0"]
       - build_linux_release:
           matrix:
             parameters:
               python_version: ["3.9", "3.10", "3.11", "3.12", "3.13"]
-              extra_env: ["PYPI_RELEASE=1"]
+              build_env: ["DEV_RELEASE=1"]
+      - build_cuda_release:
+          matrix:
+            parameters:
+              build_env: ["DEV_RELEASE=1"]

.gitignore

@@ -36,6 +36,7 @@ share/python-wheels/
 .installed.cfg
 *.egg
 MANIFEST
+uv.lock
 
 # vim
 *.swp

ACKNOWLEDGMENTS.md

@@ -19,11 +19,17 @@ MLX was developed with contributions from the following individuals:
 - Gleb Pobudzey: Added the `where` primitive, and groups in 1D and 2D convolutions.
 - Paul Paczuski: Improved stability of BCE loss calculation
 - Max-Heinrich Laves: Added `conv_transpose1d`, `conv_transpose2d`, and `conv_transpose3d` ops.
+- Gökdeniz Gülmez: Added the `Muon (MomentUm Orthogonalized by Newton-schulz)` optimizer.
 
 <a href="https://github.com/ml-explore/mlx/graphs/contributors">
   <img class="dark-light" src="https://contrib.rocks/image?repo=ml-explore/mlx&anon=0&columns=20&max=100&r=true" />
 </a>
 
+# Organizations
+
+MLX has received contributions from the following companies:
+- NVIDIA Corporation & Affiliates
+
 # Third-Party Software
 
 MLX leverages several third-party software, listed here together with

CMakeLists.txt

@@ -1,6 +1,24 @@
 cmake_minimum_required(VERSION 3.25)
 
-project(mlx LANGUAGES C CXX)
+if(NOT MLX_VERSION)
+  file(STRINGS "mlx/version.h" _mlx_h_version REGEX "^#define MLX_VERSION_.*$")
+  string(REGEX MATCH "#define MLX_VERSION_MAJOR ([0-9]+)" _ "${_mlx_h_version}")
+  set(_major ${CMAKE_MATCH_1})
+  string(REGEX MATCH "#define MLX_VERSION_MINOR ([0-9]+)" _ "${_mlx_h_version}")
+  set(_minor ${CMAKE_MATCH_1})
+  string(REGEX MATCH "#define MLX_VERSION_PATCH ([0-9]+)" _ "${_mlx_h_version}")
+  set(_patch ${CMAKE_MATCH_1})
+  set(MLX_PROJECT_VERSION "${_major}.${_minor}.${_patch}")
+  set(MLX_VERSION ${MLX_PROJECT_VERSION})
+else()
+  string(REGEX REPLACE "^([0-9]+\.[0-9]+\.[0-9]+).*" "\\1" MLX_PROJECT_VERSION
+                       ${MLX_VERSION})
+endif()
+
+project(
+  mlx
+  LANGUAGES C CXX
+  VERSION ${MLX_PROJECT_VERSION})
 
 # ----------------------------- Setup -----------------------------
 set(CMAKE_MODULE_PATH "${PROJECT_SOURCE_DIR}/cmake")
@@ -16,18 +34,16 @@ option(MLX_BUILD_BENCHMARKS "Build benchmarks for mlx" OFF)
 option(MLX_BUILD_PYTHON_BINDINGS "Build python bindings for mlx" OFF)
 option(MLX_BUILD_METAL "Build metal backend" ON)
 option(MLX_BUILD_CPU "Build cpu backend" ON)
+option(MLX_BUILD_CUDA "Build cuda backend" OFF)
 option(MLX_METAL_DEBUG "Enhance metal debug workflow" OFF)
 option(MLX_ENABLE_X64_MAC "Enable building for x64 macOS" OFF)
 option(MLX_BUILD_GGUF "Include support for GGUF format" ON)
 option(MLX_BUILD_SAFETENSORS "Include support for safetensors format" ON)
 option(MLX_BUILD_BLAS_FROM_SOURCE "Build OpenBLAS from source code" OFF)
 option(MLX_METAL_JIT "Use JIT compilation for Metal kernels" OFF)
+option(MLX_USE_CCACHE "Use CCache for compilation cache when available" ON)
 option(BUILD_SHARED_LIBS "Build mlx as a shared library" OFF)
-
-if(NOT MLX_VERSION)
-  set(MLX_VERSION 0.23.0)
-endif()
-add_compile_definitions("MLX_VERSION=${MLX_VERSION}")
+option(USE_SYSTEM_FMT "Use system's provided fmt library" OFF)
 
 # --------------------- Processor tests -------------------------
 message(
@@ -50,10 +66,17 @@ if(${CMAKE_SYSTEM_NAME} MATCHES "Darwin")
       message(WARNING "Building for x86_64 arch is not officially supported.")
     endif()
   endif()
-
 else()
   set(MLX_BUILD_METAL OFF)
-  message(WARNING "MLX is prioritised for Apple silicon systems using macOS.")
+endif()
+
+if(MLX_USE_CCACHE)
+  find_program(CCACHE_PROGRAM ccache)
+  if(CCACHE_PROGRAM)
+    set(CMAKE_C_COMPILER_LAUNCHER "${CCACHE_PROGRAM}")
+    set(CMAKE_CXX_COMPILER_LAUNCHER "${CCACHE_PROGRAM}")
+    set(CMAKE_CUDA_COMPILER_LAUNCHER "${CCACHE_PROGRAM}")
+  endif()
 endif()
 
 # ----------------------------- Lib -----------------------------
@@ -63,7 +86,6 @@ include(FetchContent)
 cmake_policy(SET CMP0135 NEW)
 
 add_library(mlx)
-set_target_properties(mlx PROPERTIES COMPILE_WARNING_AS_ERROR ON)
 
 if(MLX_BUILD_METAL)
   set(METAL_LIB "-framework Metal")
@@ -71,6 +93,10 @@ if(MLX_BUILD_METAL)
   set(QUARTZ_LIB "-framework QuartzCore")
 endif()
 
+if(MLX_BUILD_CUDA)
+  enable_language(CUDA)
+endif()
+
 if(MLX_BUILD_METAL AND NOT METAL_LIB)
   message(STATUS "Metal not found. Unable to build GPU")
   set(MLX_BUILD_METAL OFF)
@@ -114,6 +140,12 @@ elseif(MLX_BUILD_METAL)
   target_link_libraries(mlx PUBLIC ${METAL_LIB} ${FOUNDATION_LIB} ${QUARTZ_LIB})
 endif()
 
+if(CMAKE_SYSTEM_NAME STREQUAL "Linux")
+  # With newer clang/gcc versions following libs are implicitly linked, but when
+  # building on old distributions they need to be explicitly listed.
+  target_link_libraries(mlx PRIVATE dl pthread)
+endif()
+
 if(WIN32)
   if(MSVC)
     # GGUF does not build with MSVC.
@@ -200,23 +232,13 @@ else()
   set(MLX_BUILD_ACCELERATE OFF)
 endif()
 
-find_package(MPI)
-if(MPI_FOUND)
-  execute_process(
-    COMMAND zsh "-c" "mpirun --version"
-    OUTPUT_VARIABLE MPI_VERSION
-    ERROR_QUIET)
-  if(${MPI_VERSION} MATCHES ".*Open MPI.*")
-    target_include_directories(mlx PRIVATE ${MPI_INCLUDE_PATH})
-  elseif(MPI_VERSION STREQUAL "")
-    set(MPI_FOUND FALSE)
-    message(
-      WARNING "MPI found but mpirun is not available. Building without MPI.")
-  else()
-    set(MPI_FOUND FALSE)
-    message(WARNING "MPI which is not OpenMPI found. Building without MPI.")
-  endif()
-endif()
+message(STATUS "Downloading json")
+FetchContent_Declare(
+  json
+  URL https://github.com/nlohmann/json/releases/download/v3.11.3/json.tar.xz)
+FetchContent_MakeAvailable(json)
+target_include_directories(
+  mlx PRIVATE $<BUILD_INTERFACE:${json_SOURCE_DIR}/single_include/nlohmann>)
 
 add_subdirectory(${CMAKE_CURRENT_LIST_DIR}/mlx)
 
@@ -224,12 +246,19 @@ target_include_directories(
   mlx PUBLIC $<BUILD_INTERFACE:${CMAKE_CURRENT_LIST_DIR}>
              $<INSTALL_INTERFACE:include>)
 
-FetchContent_Declare(
-  fmt
-  GIT_REPOSITORY https://github.com/fmtlib/fmt.git
-  GIT_TAG 10.2.1
-  EXCLUDE_FROM_ALL)
-FetchContent_MakeAvailable(fmt)
+# Do not add mlx_EXPORTS define for shared library.
+set_target_properties(mlx PROPERTIES DEFINE_SYMBOL "")
+
+if(USE_SYSTEM_FMT)
+  find_package(fmt REQUIRED)
+else()
+  FetchContent_Declare(
+    fmt
+    GIT_REPOSITORY https://github.com/fmtlib/fmt.git
+    GIT_TAG 10.2.1
+    EXCLUDE_FROM_ALL)
+  FetchContent_MakeAvailable(fmt)
+endif()
 target_link_libraries(mlx PRIVATE $<BUILD_INTERFACE:fmt::fmt-header-only>)
 
 if(MLX_BUILD_PYTHON_BINDINGS)

CONTRIBUTING.md

@@ -5,26 +5,26 @@ possible.
 
 ## Pull Requests
 
-1. Fork and submit pull requests to the repo. 
+1. Fork and submit pull requests to the repo.
 2. If you've added code that should be tested, add tests.
 3. If a change is likely to impact efficiency, run some of the benchmarks before
    and after the change. Examples of benchmarks can be found in `benchmarks/python/`.
 4. If you've changed APIs, update the documentation.
-5. Every PR should have passing tests and at least one review. 
+5. Every PR should have passing tests and at least one review.
 6. For code formatting install `pre-commit` using something like `pip install pre-commit` and run `pre-commit install`.
    This should install hooks for running `black` and `clang-format` to ensure
    consistent style for C++ and python code.
- 
+
    You can also run the formatters manually as follows:
- 
-     ```
-     clang-format -i file.cpp
-     ```
- 
-     ```
-     black file.py
-     ```
- 
+
+   ```shell
+   clang-format -i file.cpp
+   ```
+
+   ```shell
+   black file.py
+   ```
+
    or run `pre-commit run --all-files` to check all files in the repo.
 
 ## Issues

MANIFEST.in

@@ -1,4 +1,6 @@
 include CMakeLists.txt
+include mlx.pc.in
 recursive-include mlx/ *
+include cmake/*
 include python/src/*
 include python/mlx/py.typed # support type hinting as in PEP-561

README.md

@@ -11,10 +11,10 @@ brought to you by Apple machine learning research.
 
 Some key features of MLX include:
 
- - **Familiar APIs**: MLX has a Python API that closely follows NumPy.  MLX
+ - **Familiar APIs**: MLX has a Python API that closely follows NumPy. MLX
    also has fully featured C++, [C](https://github.com/ml-explore/mlx-c), and
    [Swift](https://github.com/ml-explore/mlx-swift/) APIs, which closely mirror
-   the Python API.  MLX has higher-level packages like `mlx.nn` and
+   the Python API. MLX has higher-level packages like `mlx.nn` and
    `mlx.optimizers` with APIs that closely follow PyTorch to simplify building
    more complex models.
 
@@ -68,18 +68,23 @@ in the documentation.
 
 ## Installation
 
-MLX is available on [PyPI](https://pypi.org/project/mlx/). To install the Python API, run:
+MLX is available on [PyPI](https://pypi.org/project/mlx/). To install MLX on
+macOS, run:
 
-**With `pip`**:
-
-```
+```bash
 pip install mlx
 ```
 
-**With `conda`**:
+To install the CUDA backend on Linux, run:
 
+```bash
+pip install mlx[cuda]
 ```
-conda install -c conda-forge mlx
+
+To install a CPU-only Linux package, run:
+
+```bash
+pip install mlx[cpu]
 ```
 
 Checkout the

benchmarks/cpp/irregular_strides.cpp

@@ -1,5 +1,6 @@
 // Copyright © 2023 Apple Inc.
 
+#include <cstring>
 #include <iostream>
 #include <sstream>
 

benchmarks/cpp/single_ops.cpp

@@ -192,6 +192,22 @@ void time_reductions() {
 
   auto argmin_along_1 = [&a]() { return mx::argmin(a, 1, false); };
   TIME(argmin_along_1);
+
+  auto indices = mx::array({1});
+  auto updates = mx::reshape(mx::array({NAN}), {1, 1, 1});
+  std::vector<int> axes{0};
+  auto b = scatter(a, {indices}, updates, axes);
+  mx::eval(b);
+
+  auto max_along_0 = [&b]() { return mx::max(b, 0, false); };
+  TIME(max_along_0);
+  auto max_along_1 = [&b]() { return mx::max(b, 1, false); };
+  TIME(max_along_1);
+
+  auto min_along_0 = [&b]() { return mx::min(b, 0, false); };
+  TIME(min_along_0);
+  auto min_along_1 = [&b]() { return mx::min(b, 1, false); };
+  TIME(min_along_1);
 }
 
 void time_gather_scatter() {

benchmarks/python/comparative/bench_torch.py

@@ -5,6 +5,7 @@ import os
 import time
 
 import torch
+import torch.cuda
 import torch.mps
 
 
@@ -44,8 +45,10 @@ def bench(f, *args):
 
 
 def sync_if_needed(x):
-    if x.device != torch.device("cpu"):
+    if x.device == torch.device("mps"):
         torch.mps.synchronize()
+    elif x.device == torch.device("cuda"):
+        torch.cuda.synchronize()
 
 
 @torch.no_grad()
@@ -99,6 +102,14 @@ def reduction(op, axis, x):
     sync_if_needed(x)
 
 
+@torch.no_grad()
+def sum_and_add(axis, x, y):
+    z = x.sum(axis=axis, keepdims=True)
+    for i in range(50):
+        z = (z + y).sum(axis=axis, keepdims=True)
+    sync_if_needed(x)
+
+
 @torch.no_grad()
 def softmax(axis, x):
     ys = []
@@ -340,7 +351,11 @@ if __name__ == "__main__":
         args.axis.pop(0)
 
     torch.set_num_threads(1)
-    device = "cpu" if args.cpu else "mps"
+    device = "mps"
+    if torch.cuda.is_available():
+        device = "cuda"
+    if args.cpu:
+        device = "cpu"
 
     types = args.dtype
     if not types:
@@ -460,5 +475,8 @@ if __name__ == "__main__":
     elif args.benchmark == "selu":
         print(bench(selu, x))
 
+    elif args.benchmark == "sum_and_add":
+        print(bench(sum_and_add, axis, *xs))
+
     else:
         raise ValueError(f"Unknown benchmark `{args.benchmark}`.")

benchmarks/python/conv_unaligned_bench.py

@@ -0,0 +1,107 @@
+import math
+import time
+
+import mlx.core as mx
+import numpy as np
+import torch
+
+N_warmup = 10
+N_iter_bench = 100
+N_iter_func = 5
+
+
+def bench(f, a, b):
+    for i in range(N_warmup):
+        f(a, b)
+    torch.mps.synchronize()
+
+    s = time.perf_counter_ns()
+    for i in range(N_iter_bench):
+        f(a, b)
+    e = time.perf_counter_ns()
+    return (e - s) * 1e-9
+
+
+def make_mx_conv_2D(strides=(1, 1), padding=(0, 0), groups=1):
+    def mx_conv_2D(a, b):
+        ys = []
+        for i in range(N_iter_func):
+            y = mx.conv2d(a, b, stride=strides, padding=padding, groups=groups)
+            ys.append(y)
+        mx.eval(ys)
+        return ys
+
+    return mx_conv_2D
+
+
+def make_pt_conv_2D(strides=(1, 1), padding=(0, 0), groups=1):
+    @torch.no_grad()
+    def pt_conv_2D(a, b):
+        ys = []
+        for i in range(N_iter_func):
+            y = torch.conv2d(a, b, stride=strides, padding=padding, groups=groups)
+            ys.append(y)
+        torch.mps.synchronize()
+        return ys
+
+    return pt_conv_2D
+
+
+def bench_shape(N, H, W, C, kH, kW, O, strides, padding, groups, np_dtype):
+    scale = 1.0 / math.sqrt(kH * kH * C)
+    a_np = np.random.uniform(0, 0.5, (N, H, W, C)).astype(np_dtype)
+    b_np = np.random.uniform(-scale, scale, (O, kH, kW, int(C / groups))).astype(
+        np_dtype
+    )
+
+    a_mx = mx.array(a_np)
+    b_mx = mx.array(b_np)
+
+    a_pt = torch.from_numpy(a_np.transpose((0, 3, 1, 2))).to("mps")
+    b_pt = torch.from_numpy(b_np.transpose((0, 3, 1, 2))).to("mps")
+
+    torch.mps.synchronize()
+
+    f_mx = make_mx_conv_2D(strides, padding, groups)
+    f_pt = make_pt_conv_2D(strides, padding, groups)
+
+    time_torch = bench(f_pt, a_pt, b_pt)
+    time_mlx = bench(f_mx, a_mx, b_mx)
+
+    out_mx = mx.conv2d(a_mx, b_mx, stride=strides, padding=padding, groups=groups)
+    out_pt = torch.conv2d(
+        a_pt.to("cpu"), b_pt.to("cpu"), stride=strides, padding=padding, groups=groups
+    )
+    out_pt = torch.permute(out_pt, (0, 2, 3, 1))
+    out_pt = out_pt.numpy(force=True)
+
+    atol = 2e-5 if np_dtype == np.float32 else 1e-4
+
+    if not np.allclose(out_pt, out_mx, atol=atol):
+        print(
+            f"Failed at {(N, H, W, C)}, {(O, kH, kW, C)} [strides = {strides}, padding = {padding}, groups = {groups}] with max(|a - b|) = {np.max(np.abs(out_pt - out_mx))}"
+        )
+
+    return time_mlx, time_torch
+
+
+if __name__ == "__main__":
+    dtype = "float32"
+    shapes = (
+        (4, 32, 32, 21, 3, 3, 128),
+        (4, 32, 32, 21, 3, 3, 37),
+        (4, 32, 32, 370, 3, 3, 370),
+        (4, 32, 32, 370, 7, 7, 128),
+        (2, 320, 640, 21, 7, 7, 21),
+    )
+    for N, H, W, C, kh, kw, O in shapes:
+        time_mlx, time_torch = bench_shape(
+            N, H, W, C, kh, kw, O, (1, 1), (0, 0), 1, dtype
+        )
+        diff = time_torch / time_mlx - 1.0
+
+        print(
+            f"({N}, {H:3d}, {W:3d}, {C:3d}), ({O:3d}, {kh:2d}, {kw:2d}, {C:3d}), {dtype}, {100. * diff:+5.2f}%"
+        )
+        if time_mlx >= 2.0 * time_torch:
+            print("ATTENTION ^^^^^^^")

benchmarks/python/gather_bench.py

@@ -1,7 +1,6 @@
 # Copyright © 2023-2024 Apple Inc.
 
 import argparse
-from time import time
 
 import mlx.core as mx
 import torch

benchmarks/python/gather_mm_bench.py

@@ -0,0 +1,74 @@
+# Copyright © 2025 Apple Inc.
+
+import mlx.core as mx
+from time_utils import time_fn
+
+N = 1024
+D = 1024
+M = 1024
+E = 32
+I = 4
+
+
+def gather_sort(x, indices):
+    N, M = indices.shape
+    indices = indices.flatten()
+    order = mx.argsort(indices)
+    inv_order = mx.argsort(order)
+    return x.flatten(0, -3)[order // M], indices[order], inv_order
+
+
+def scatter_unsort(x, inv_order, shape=None):
+    x = x[inv_order]
+    if shape is not None:
+        x = mx.unflatten(x, 0, shape)
+    return x
+
+
+def gather_mm_simulate(x, w, indices):
+    x, idx, inv_order = gather_sort(x, indices)
+    for i in range(2):
+        y = mx.concatenate([x[i] @ w[j].T for i, j in enumerate(idx.tolist())], axis=0)
+        x = y[:, None]
+    x = scatter_unsort(x, inv_order, indices.shape)
+    return x
+
+
+def time_gather_mm():
+    x = mx.random.normal((N, 1, 1, D)) / 1024**0.5
+    w1 = mx.random.normal((E, M, D)) / 1024**0.5
+    w2 = mx.random.normal((E, D, M)) / 1024**0.5
+    indices = (mx.random.uniform(shape=(N, I)) * E).astype(mx.uint32)
+    sorted_indices = mx.sort(indices.flatten()).reshape(N, I)
+    mx.eval(x, w1, w2, indices, sorted_indices)
+
+    def gather_mm(x, w1, w2, indices, sort):
+        idx = indices
+        inv_order = None
+        if sort:
+            x, idx, inv_order = gather_sort(x, indices)
+        x = mx.gather_mm(x, w1.swapaxes(-1, -2), rhs_indices=idx, sorted_indices=sort)
+        x = mx.gather_mm(x, w2.swapaxes(-1, -2), rhs_indices=idx, sorted_indices=sort)
+        if sort:
+            x = scatter_unsort(x, inv_order, indices.shape)
+        return x
+
+    time_fn(gather_mm, x, w1, w2, indices, False)
+    time_fn(gather_mm, x, w1, w2, sorted_indices, False)
+    time_fn(gather_mm, x, w1, w2, indices, True)
+
+    x = mx.random.normal((N * I, D)) / 1024**0.5
+    w1 = mx.random.normal((M, D)) / 1024**0.5
+    w2 = mx.random.normal((D, M)) / 1024**0.5
+    mx.eval(x, w1, w2)
+
+    def equivalent_matmul(x, w1, w2):
+        x = x @ w1.T
+        x = x @ w2.T
+        return x
+
+    time_fn(equivalent_matmul, x, w1, w2)
+
+
+if __name__ == "__main__":
+    time_gather_mm()

benchmarks/python/gather_qmm_bench.py

@@ -0,0 +1,84 @@
+# Copyright © 2025 Apple Inc.
+
+import mlx.core as mx
+from time_utils import time_fn
+
+N = 1024
+D = 1024
+M = 1024
+E = 32
+I = 4
+
+
+def gather_sort(x, indices):
+    N, M = indices.shape
+    indices = indices.flatten()
+    order = mx.argsort(indices)
+    inv_order = mx.argsort(order)
+    return x.flatten(0, -3)[order // M], indices[order], inv_order
+
+
+def scatter_unsort(x, inv_order, shape=None):
+    x = x[inv_order]
+    if shape is not None:
+        x = mx.unflatten(x, 0, shape)
+    return x
+
+
+def gather_mm_simulate(x, w, indices):
+    x, idx, inv_order = gather_sort(x, indices)
+    for i in range(2):
+        y = mx.concatenate(
+            [
+                mx.quantized_matmul(x[i], w[0][j], w[1][j], w[2][j], transpose=True)
+                for i, j in enumerate(idx.tolist())
+            ],
+            axis=0,
+        )
+        x = y[:, None]
+    x = scatter_unsort(x, inv_order, indices.shape)
+    return x
+
+
+def time_gather_qmm():
+    x = mx.random.normal((N, 1, 1, D)) / 1024**0.5
+    w1 = mx.random.normal((E, M, D)) / 1024**0.5
+    w2 = mx.random.normal((E, D, M)) / 1024**0.5
+    w1 = mx.quantize(w1)
+    w2 = mx.quantize(w2)
+    indices = (mx.random.uniform(shape=(N, I)) * E).astype(mx.uint32)
+    sorted_indices = mx.sort(indices.flatten()).reshape(N, I)
+    mx.eval(x, w1, w2, indices, sorted_indices)
+
+    def gather_mm(x, w1, w2, indices, sort):
+        idx = indices
+        inv_order = None
+        if sort:
+            x, idx, inv_order = gather_sort(x, indices)
+        x = mx.gather_qmm(x, *w1, transpose=True, rhs_indices=idx, sorted_indices=sort)
+        x = mx.gather_qmm(x, *w2, transpose=True, rhs_indices=idx, sorted_indices=sort)
+        if sort:
+            x = scatter_unsort(x, inv_order, indices.shape)
+        return x
+
+    time_fn(gather_mm, x, w1, w2, indices, False)
+    time_fn(gather_mm, x, w1, w2, sorted_indices, False)
+    time_fn(gather_mm, x, w1, w2, indices, True)
+
+    x = mx.random.normal((N * I, D)) / 1024**0.5
+    w1 = mx.random.normal((M, D)) / 1024**0.5
+    w2 = mx.random.normal((D, M)) / 1024**0.5
+    w1 = mx.quantize(w1)
+    w2 = mx.quantize(w2)
+    mx.eval(x, w1, w2)
+
+    def equivalent_matmul(x, w1, w2):
+        x = mx.quantized_matmul(x, *w1, transpose=True)
+        x = mx.quantized_matmul(x, *w2, transpose=True)
+        return x
+
+    time_fn(equivalent_matmul, x, w1, w2)
+
+
+if __name__ == "__main__":
+    time_gather_qmm()

benchmarks/python/layer_norm_bench.py

@@ -1,5 +1,7 @@
 # Copyright © 2023-2024 Apple Inc.
 
+from functools import partial
+
 import mlx.core as mx
 import mlx.nn as nn
 from time_utils import time_fn
@@ -10,32 +12,71 @@ def layer_norm(x, w, b, eps):
     x = x.astype(mx.float32)
     mu = mx.mean(x, -1, keepdims=True)
     v = mx.var(x, -1, keepdims=True)
-    return (x - mu) * mx.rsqrt(v + eps) * w + b
+    y = (x - mu) * mx.rsqrt(v + eps)
+    if w is not None:
+        y = y * w
+    if b is not None:
+        y = y + b
+    return y
 
 
-def time_layer_norm():
+def time_layer_norm(N, dt):
+    L = 1024
     f1 = lambda x, w, b, y: (layer_norm(x, w, b, 1e-5) * y).sum()
     f2 = lambda x, w, b, y: (mx.fast.layer_norm(x, w, b, 1e-5) * y).sum()
     g1 = mx.grad(f1, argnums=(0, 1, 2))
     g2 = mx.grad(f2, argnums=(0, 1, 2))
 
-    x = mx.random.uniform(shape=(8, 1024, 4096)).astype(mx.float16)
-    w = mx.random.uniform(shape=(4096,)).astype(mx.float16)
-    b = mx.random.uniform(shape=(4096,)).astype(mx.float16)
-    y = mx.random.uniform(shape=(8, 1024, 4096)).astype(mx.float16)
+    x = mx.random.uniform(shape=(8, L, N)).astype(dt)
+    w = mx.random.uniform(shape=(N,)).astype(dt)
+    b = mx.random.uniform(shape=(N,)).astype(dt)
+    y = mx.random.uniform(shape=(8, L, N)).astype(dt)
     mx.eval(x, w, b, y)
 
-    def layer_norm_loop(g, x, w, b):
+    def layer_norm_loop(f, x, w, b):
+        for _ in range(32):
+            x = f(x, w, b)
+        return x
+
+    time_fn(layer_norm_loop, partial(layer_norm, eps=1e-5), x, w, b)
+    time_fn(layer_norm_loop, partial(mx.fast.layer_norm, eps=1e-5), x, w, b)
+
+    def layer_norm_grad_loop(g, x, w, b):
         gx, gw, gb = x, w, b
         for _ in range(32):
             gx, gw, gb = g(gx, gw, gb, y)
         return gx, gw, gb
 
-    time_fn(layer_norm_loop, g1, x, w, b)
-    time_fn(layer_norm_loop, g2, x, w, b)
-    time_fn(layer_norm_loop, mx.compile(g1), x, w, b)
-    time_fn(layer_norm_loop, mx.compile(g2), x, w, b)
+    time_fn(layer_norm_grad_loop, g1, x, w, b)
+    time_fn(layer_norm_grad_loop, g2, x, w, b)
+    time_fn(layer_norm_grad_loop, mx.compile(g1), x, w, b)
+    time_fn(layer_norm_grad_loop, mx.compile(g2), x, w, b)
+
+    f1 = lambda x, y: (layer_norm(x, None, None, 1e-5) * y).sum()
+    f2 = lambda x, y: (mx.fast.layer_norm(x, None, None, 1e-5) * y).sum()
+    g1 = mx.grad(f1, argnums=(0,))
+    g2 = mx.grad(f2, argnums=(0,))
+
+    x = mx.random.uniform(shape=(8, L, N)).astype(dt)
+    w = mx.random.uniform(shape=(N,)).astype(dt)
+    b = mx.random.uniform(shape=(N,)).astype(dt)
+    y = mx.random.uniform(shape=(8, L, N)).astype(dt)
+    mx.eval(x, w, b, y)
+
+    def layer_norm_grad_x_loop(g, x):
+        gx = x
+        for _ in range(32):
+            gx = g(gx, y)
+        return gx
+
+    time_fn(layer_norm_grad_x_loop, g1, x)
+    time_fn(layer_norm_grad_x_loop, g2, x)
+    time_fn(layer_norm_grad_x_loop, mx.compile(g1), x)
+    time_fn(layer_norm_grad_x_loop, mx.compile(g2), x)
 
 
 if __name__ == "__main__":
-    time_layer_norm()
+    for dt in [mx.float32, mx.float16, mx.bfloat16]:
+        for n in [1024, 2048, 4096, 8192, 8192 + 1024]:
+            print(dt, n)
+            time_layer_norm(n, dt)

benchmarks/python/rms_norm_bench.py

@@ -9,7 +9,10 @@ def rms_norm(x, w, eps):
     ot = x.dtype
     x = x.astype(mx.float32)
     n = mx.rsqrt(x.square().mean(-1, keepdims=True) + eps)
-    return (x * n).astype(ot) * w
+    y = (x * n).astype(ot)
+    if w is not None:
+        y = y * w
+    return y
 
 
 def time_rms_norm():
@@ -34,6 +37,27 @@ def time_rms_norm():
     time_fn(rms_norm_loop, mx.compile(g1), x, w)
     time_fn(rms_norm_loop, mx.compile(g2), x, w)
 
+    f1 = lambda x, y: (rms_norm(x, None, 1e-5) * y).sum()
+    f2 = lambda x, y: (mx.fast.rms_norm(x, None, 1e-5) * y).sum()
+    g1 = mx.grad(f1, argnums=(0,))
+    g2 = mx.grad(f2, argnums=(0,))
+
+    x = mx.random.uniform(shape=(8, 1024, 4096)).astype(mx.float16)
+    w = mx.random.uniform(shape=(4096,)).astype(mx.float16)
+    y = mx.random.uniform(shape=(8, 1024, 4096)).astype(mx.float16)
+    mx.eval(x, w, y)
+
+    def rms_norm_loop(g, x):
+        gx = x
+        for _ in range(32):
+            gx = g(gx, y)
+        return gx
+
+    time_fn(rms_norm_loop, g1, x)
+    time_fn(rms_norm_loop, g2, x)
+    time_fn(rms_norm_loop, mx.compile(g1), x)
+    time_fn(rms_norm_loop, mx.compile(g2), x)
+
 
 if __name__ == "__main__":
     time_rms_norm()

benchmarks/python/sdpa_bench.py

@@ -28,11 +28,34 @@ def bench(f, *args):
     return (e - s) * 1e-9
 
 
-def mlx_sdpa_fused_inner(q, k, v, scale):
-    return mx.fast.scaled_dot_product_attention(q, k, v, scale=scale, mask=None)
+def prepare_inputs(B, qL, kL, D, qH, kH, mask, transpose, dtype):
+    np_dtype = getattr(np, dtype)
+
+    shape_q = (B, qL, qH, D) if transpose else (B, qH, qL, D)
+    shape_kv = (B, kL, kH, D) if transpose else (B, kH, kL, D)
+
+    scale = 1.0 / math.sqrt(D)
+
+    q_np = np.random.normal(0.0, 1.0, shape_q).astype(np_dtype)
+    k_np = np.random.normal(0.0, scale, shape_kv).astype(np_dtype)
+    v_np = np.random.normal(0.0, scale, shape_kv).astype(np_dtype)
+
+    q_mx = mx.array(q_np)
+    k_mx = mx.array(k_np)
+    v_mx = mx.array(v_np)
+
+    if mask is not None:
+        if mask == "additive":
+            mask_np = np.random.normal(0.0, 1.0, (B, qH, qL, kL)).astype(np_dtype)
+            mask = mx.array(mask_np)
+        elif mask == "bool":
+            mask_np = np.random.uniform(0.0, 1.0, (B, qH, qL, kL)) < 0.5
+            mask = mx.array(mask_np)
+
+    return q_mx, k_mx, v_mx, scale, mask
 
 
-def mlx_sdpa_unfused_inner(q, k, v, scale, f32softmax=False):
+def mlx_ref_attn(q, k, v, scale=1.0, mask=None):
     q_dtype = q.dtype
     q = q * mx.array(scale, q_dtype)
     n_q_heads = q.shape[-3]
@@ -41,6 +64,7 @@ def mlx_sdpa_unfused_inner(q, k, v, scale, f32softmax=False):
 
     B = q.shape[0]
     L = q.shape[2]
+    kL = k.shape[2]
 
     if n_repeats > 1:
         q = mx.reshape(q, [B, n_kv_heads, n_repeats, L, -1])
@@ -48,10 +72,27 @@ def mlx_sdpa_unfused_inner(q, k, v, scale, f32softmax=False):
         v = mx.expand_dims(v, 2)
 
     scores = q @ mx.swapaxes(k, -1, -2)
-    if f32softmax:
-        scores = mx.softmax(scores.astype(mx.float32), axis=-1).astype(q_dtype)
-    else:
-        scores = mx.softmax(scores, axis=-1)
+
+    if mask is not None:
+
+        if mask == "causal":
+            q_offset = max(0, kL - L)
+            q_indices = mx.arange(q_offset, q_offset + L)
+            k_indices = mx.arange(kL)
+            mask = q_indices[:, None] >= k_indices[None]
+
+        if n_repeats > 1 and mask.ndim >= 3:
+            if mask.shape[-3] == 1:
+                mask = mx.expand_dims(mask, -3)
+            else:
+                mask = mx.unflatten(mask, -3, (n_kv_heads, n_repeats))
+
+        if mask.dtype == mx.bool_:
+            scores = mx.where(mask, scores, -np.float32(np.inf))
+        else:
+            scores += mask
+
+    scores = mx.softmax(scores, axis=-1, precise=True)
 
     out = scores @ v
     if n_repeats > 1:
@@ -60,74 +101,55 @@ def mlx_sdpa_unfused_inner(q, k, v, scale, f32softmax=False):
     return out
 
 
-def mlx_spda_unfused(q, k, v, scale, transpose):
-    q_out = q
+def mlx_fused_attn(q, k, v, scale, mask):
+    return mx.fast.scaled_dot_product_attention(q, k, v, scale=scale, mask=mask)
+
+
+def do_attention(f, q, k, v, scale, mask=None, transpose=False):
     if transpose:
-        k = mx.transpose(k, (0, 2, 1, 3))
-        v = mx.transpose(v, (0, 2, 1, 3))
+        q_t = mx.transpose(q, (0, 2, 1, 3))
+        k_t = mx.transpose(k, (0, 2, 1, 3))
+        v_t = mx.transpose(v, (0, 2, 1, 3))
+        o_t = f(q_t, k_t, v_t, scale=scale, mask=mask)
+        return mx.transpose(o_t, (0, 2, 1, 3))
+    else:
+        return f(q, k, v, scale=scale, mask=mask)
+
+
+def do_attention_bench(f, q, k, v, scale, mask=None, transpose=False):
+    q_out = q
 
     for i in range(N_iter_func):
-        if transpose:
-            q_out = mx.transpose(q_out, (0, 2, 1, 3))
-        q_out = mlx_sdpa_unfused_inner(q_out, k, v, scale)
-        if transpose:
-            q_out = mx.transpose(q_out, (0, 2, 1, 3))
+        q_out = do_attention(f, q_out, k, v, scale, mask=mask, transpose=transpose)
 
     mx.eval(q_out)
     return q_out
 
 
-def mlx_spda_fused(q, k, v, scale, transpose):
-    q_out = q
-    if transpose:
-        k = mx.transpose(k, (0, 2, 1, 3))
-        v = mx.transpose(v, (0, 2, 1, 3))
-
-    for i in range(N_iter_func):
-        if transpose:
-            q_out = mx.transpose(q_out, (0, 2, 1, 3))
-        q_out = mlx_sdpa_fused_inner(q_out, k, v, scale)
-        if transpose:
-            q_out = mx.transpose(q_out, (0, 2, 1, 3))
-
-    mx.eval(q_out)
-    return q_out
-
-
-def bench_shape(B, qsl, ksl, head_dim, n_q_heads, n_kv_heads, np_dtype, transpose=True):
-    shape_q = (
-        (B, qsl, n_q_heads, head_dim) if transpose else (B, n_q_heads, qsl, head_dim)
-    )
-    shape_kv = (
-        (B, ksl, n_kv_heads, head_dim) if transpose else (B, n_kv_heads, ksl, head_dim)
+def bench_shape(
+    B, qsl, ksl, head_dim, n_q_heads, n_kv_heads, dtype, transpose=True, mask_in=None
+):
+    q_mx, k_mx, v_mx, scale, mask = prepare_inputs(
+        B, qsl, ksl, head_dim, n_q_heads, n_kv_heads, mask_in, transpose, dtype
     )
 
-    q_np = np.random.normal(0.0, 1.0 / math.sqrt(head_dim), shape_q).astype(np_dtype)
-    k_np = np.random.normal(0.0, 1.0 / math.sqrt(head_dim), shape_kv).astype(np_dtype)
-    v_np = np.random.normal(0.0, 1.0 / math.sqrt(head_dim), shape_kv).astype(np_dtype)
+    time_mlx_unfused = bench(
+        do_attention_bench, mlx_ref_attn, q_mx, k_mx, v_mx, scale, mask, transpose
+    )
+    time_mlx_fused = bench(
+        do_attention_bench, mlx_fused_attn, q_mx, k_mx, v_mx, scale, mask, transpose
+    )
 
-    scale = math.sqrt(1.0 / head_dim)
+    o_mlx_fused = do_attention(mlx_ref_attn, q_mx, k_mx, v_mx, scale, mask, transpose)
+    o_mlx_unfused = do_attention(
+        mlx_fused_attn, q_mx, k_mx, v_mx, scale, mask, transpose
+    )
 
-    q_mx = mx.array(q_np)
-    k_mx = mx.array(k_np)
-    v_mx = mx.array(v_np)
+    atol = 1e-5 if dtype == "float32" else 2e-4
 
-    time_mlx_unfused = bench(mlx_spda_unfused, q_mx, k_mx, v_mx, scale, transpose)
-    time_mlx_fused = bench(mlx_spda_fused, q_mx, k_mx, v_mx, scale, transpose)
-
-    if transpose:
-        q_mx = mx.transpose(q_mx, (0, 2, 1, 3))
-        k_mx = mx.transpose(k_mx, (0, 2, 1, 3))
-        v_mx = mx.transpose(v_mx, (0, 2, 1, 3))
-
-    o_mlx_fused = mlx_sdpa_fused_inner(q_mx, k_mx, v_mx, scale)
-    o_mlx_unfused = mlx_sdpa_unfused_inner(q_mx, k_mx, v_mx, scale, f32softmax=True)
-
-    atol = 1e-5 if np_dtype == np.float32 else 1e-4
-
-    if not mx.allclose(o_mlx_fused, o_mlx_unfused, atol=atol):
+    if not mx.allclose(o_mlx_fused, o_mlx_unfused, atol=atol, rtol=atol):
         print(
-            f"Failed at (B: {B}, qsl: {qsl}, ksl: {ksl}, head_dim: {head_dim}, n_qh: {n_q_heads}, n_kvh: {n_kv_heads}) [tpose = {transpose}] with max(|a - b|) = {mx.max(mx.abs(o_mlx_unfused - o_mlx_fused)):3.2e}"
+            f"Failed at (B: {B}, qsl: {qsl}, ksl: {ksl}, head_dim: {head_dim}, n_qh: {n_q_heads}, n_kvh: {n_kv_heads}, mask: {mask_in}) [tpose = {transpose}] with max(|a - b|) = {mx.max(mx.abs(o_mlx_unfused - o_mlx_fused)):3.2e}"
         )
 
     return time_mlx_fused, time_mlx_unfused
@@ -151,39 +173,51 @@ if __name__ == "__main__":
           (  1,   128,   128,       64,   32,    32),
           (  1,   256,   256,       64,   32,    32),
           (  1,   512,   512,       64,   32,    32),
-          (  1,  1024,  1024,       64,   32,    32),
-          (  1,  2048,  2048,       64,   32,    32),
-          (  1,  4096,  4096,       64,   32,    32),
+          (  1,  1024,  1024,       64,   32,     8),
+          (  1,  2048,  2048,       64,   32,     8),
+          (  1,  4096,  4096,       64,   32,     8),
     )
 
     shapes_80 = (
         # (  B,   qsl,   ksl, head_dim, n_qh, n_kvh)
-          (  1,  1024,  1024,       80,   32,    32),
-          (  1,  2048,  2048,       80,   32,    32),
-          (  1,  4096,  4096,       80,   32,    32),
+          (  1,  1024,  1024,       80,   32,     8),
+          (  1,  2048,  2048,       80,   32,     8),
+          (  1,  4096,  4096,       80,   32,     8),
     )
 
     shapes_128 = (
         # (  B,   qsl,   ksl, head_dim, n_qh, n_kvh)
-          (  1,  1024,  1024,      128,   32,    32),
-          (  1,  2048,  2048,      128,   32,    32),
-          (  1,  4096,  4096,      128,   32,    32),
+          (  1,  1024,  1024,      128,   32,     8),
+          (  1,  2048,  2048,      128,   32,     8),
+          (  1,  4096,  4096,      128,   32,     8),
     )
     # fmt: on
 
     shapes = shapes_64 + shapes_80 + shapes_128
 
-    print("  B,   qsl,   ksl, hdim, n_qh, n_kvh, tpose,   dtype, t_unfs, t_fuse, diff%")
+    masks = [None, "bool", "causal"]
+
+    print(
+        "  B,   qsl,   ksl, hdim, n_qh, n_kvh, t,   dtype,     mask, t_unfs, t_fuse, diff%"
+    )
 
     for dtype in dtypes:
         for transpose in transposes:
             for B, qsl, ksl, head_dim, n_q_heads, n_kv_heads in shapes:
-                np_dtype = getattr(np, dtype)
-                time_mlx_fused, time_mlx_unfused = bench_shape(
-                    B, qsl, ksl, head_dim, n_q_heads, n_kv_heads, np_dtype, transpose
-                )
-                diff = time_mlx_unfused / time_mlx_fused - 1.0
-                t_str = 1 if transpose else 0
-                print(
-                    f"{B:3d}, {qsl:5d}, {ksl:5d}, {head_dim:4d}, {n_q_heads:4d}, {n_kv_heads:5d}, {t_str:5d}, {dtype}, {time_mlx_unfused: 2.3f}, {time_mlx_fused: 2.3f}, {100. * diff:+5.2f}%"
-                )
+                for mask_in in masks:
+                    time_mlx_fused, time_mlx_unfused = bench_shape(
+                        B,
+                        qsl,
+                        ksl,
+                        head_dim,
+                        n_q_heads,
+                        n_kv_heads,
+                        dtype,
+                        transpose,
+                        mask_in,
+                    )
+                    diff = time_mlx_unfused / time_mlx_fused - 1.0
+                    t_str = 1 if transpose else 0
+                    print(
+                        f"{B:3d}, {qsl:5d}, {ksl:5d}, {head_dim:4d}, {n_q_heads:4d}, {n_kv_heads:5d}, {t_str:1d}, {dtype}, {str(mask_in):>8}, {time_mlx_unfused: 2.3f}, {time_mlx_fused: 2.3f}, {100. * diff:+5.2f}%"
+                    )

benchmarks/python/single_ops.py

@@ -51,6 +51,20 @@ def time_maximum():
     time_fn(mx.maximum, a, b)
 
 
+def time_max():
+    a = mx.random.uniform(shape=(32, 1024, 1024))
+    a[1, 1] = mx.nan
+    mx.eval(a)
+    time_fn(mx.max, a, 0)
+
+
+def time_min():
+    a = mx.random.uniform(shape=(32, 1024, 1024))
+    a[1, 1] = mx.nan
+    mx.eval(a)
+    time_fn(mx.min, a, 0)
+
+
 def time_negative():
     a = mx.random.uniform(shape=(10000, 1000))
     mx.eval(a)
@@ -108,6 +122,8 @@ if __name__ == "__main__":
 
     time_add()
     time_matmul()
+    time_min()
+    time_max()
     time_maximum()
     time_exp()
     time_negative()

cmake/FindNCCL.cmake

@@ -0,0 +1,54 @@
+# FindNCCL.cmake This module finds the NVIDIA NCCL library and its include
+# directories.
+
+set(NCCL_ROOT_DIR
+    $ENV{NCCL_ROOT_DIR}
+    CACHE PATH "Folder contains NVIDIA NCCL")
+
+find_path(
+  NCCL_INCLUDE_DIRS
+  NAMES nccl.h
+  HINTS ${NCCL_INCLUDE_DIR} ${NCCL_ROOT_DIR} ${NCCL_ROOT_DIR}/include
+        ${CUDA_TOOLKIT_ROOT_DIR}/include)
+
+if($ENV{USE_STATIC_NCCL})
+  message(
+    STATUS "USE_STATIC_NCCL detected. Linking against static NCCL library")
+  set(NCCL_LIBNAME "libnccl_static.a")
+else()
+  set(NCCL_LIBNAME "nccl")
+endif()
+
+find_library(
+  NCCL_LIBRARIES
+  NAMES ${NCCL_LIBNAME}
+  HINTS ${NCCL_LIB_DIR}
+        ${NCCL_ROOT_DIR}
+        ${NCCL_ROOT_DIR}/lib
+        ${NCCL_ROOT_DIR}/lib/x86_64-linux-gnu
+        ${NCCL_ROOT_DIR}/lib64
+        ${CUDA_TOOLKIT_ROOT_DIR}/lib
+        ${CUDA_TOOLKIT_ROOT_DIR}/lib64)
+
+include(FindPackageHandleStandardArgs)
+find_package_handle_standard_args(NCCL DEFAULT_MSG NCCL_INCLUDE_DIRS
+                                  NCCL_LIBRARIES)
+
+if(NCCL_FOUND)
+  set(NCCL_HEADER_FILE "${NCCL_INCLUDE_DIRS}/nccl.h")
+  message(
+    STATUS "Determining NCCL version from the header file: ${NCCL_HEADER_FILE}")
+  file(
+    STRINGS ${NCCL_HEADER_FILE} NCCL_MAJOR_VERSION_DEFINED
+    REGEX "^[ \t]*#define[ \t]+NCCL_MAJOR[ \t]+[0-9]+.*$"
+    LIMIT_COUNT 1)
+  if(NCCL_MAJOR_VERSION_DEFINED)
+    string(REGEX REPLACE "^[ \t]*#define[ \t]+NCCL_MAJOR[ \t]+" ""
+                         NCCL_MAJOR_VERSION ${NCCL_MAJOR_VERSION_DEFINED})
+    message(STATUS "NCCL_MAJOR_VERSION: ${NCCL_MAJOR_VERSION}")
+  endif()
+  message(
+    STATUS
+      "Found NCCL (include: ${NCCL_INCLUDE_DIRS}, library: ${NCCL_LIBRARIES})")
+  mark_as_advanced(NCCL_ROOT_DIR NCCL_INCLUDE_DIRS NCCL_LIBRARIES)
+endif()

cmake/extension.cmake

@@ -1,5 +1,7 @@
 include(CMakeParseArguments)
 
+# clang format off
+#
 # ##############################################################################
 # Build metal library
 #
@@ -9,11 +11,14 @@ include(CMakeParseArguments)
 # Args: TARGET: Custom target to be added for the metal library TITLE: Name of
 # the .metallib OUTPUT_DIRECTORY: Where to place ${TITLE}.metallib SOURCES: List
 # of source files INCLUDE_DIRS: List of include dirs DEPS: List of dependency
-# files (like headers)
+# files (like headers) DEBUG: Boolean, if true, enables debug compile options
+# for this specific library. If not provided, uses global MLX_METAL_DEBUG.
 #
+# clang format on
+
 macro(mlx_build_metallib)
   # Parse args
-  set(oneValueArgs TARGET TITLE OUTPUT_DIRECTORY)
+  set(oneValueArgs TARGET TITLE OUTPUT_DIRECTORY DEBUG)
   set(multiValueArgs SOURCES INCLUDE_DIRS DEPS)
   cmake_parse_arguments(MTLLIB "" "${oneValueArgs}" "${multiValueArgs}" ${ARGN})
 
@@ -21,7 +26,11 @@ macro(mlx_build_metallib)
   set(MTLLIB_BUILD_TARGET "${MTLLIB_OUTPUT_DIRECTORY}/${MTLLIB_TITLE}.metallib")
 
   # Collect compile options
-  set(MTLLIB_COMPILE_OPTIONS -Wall -Wextra -fno-fast-math)
+  set(MTLLIB_COMPILE_OPTIONS -Wall -Wextra -fno-fast-math -Wno-c++17-extensions)
+  if(MLX_METAL_DEBUG OR MTLLIB_DEBUG)
+    set(MTLLIB_COMPILE_OPTIONS ${MTLLIB_COMPILE_OPTIONS} -gline-tables-only
+                               -frecord-sources)
+  endif()
 
   # Prepare metallib build command
   add_custom_command(

docs/Doxyfile

@@ -13,7 +13,7 @@ EXCLUDE_PATTERNS       = */private/*
 CREATE_SUBDIRS         = NO
 FULL_PATH_NAMES        = YES
 RECURSIVE              = YES
-GENERATE_HTML          = YES
+GENERATE_HTML          = NO
 GENERATE_LATEX         = NO
 GENERATE_XML           = YES
 XML_PROGRAMLISTING     = YES

docs/requirements.txt

@@ -1,4 +1,5 @@
 sphinx
 breathe
 sphinx-book-theme
+sphinx-copybutton
 mlx

docs/src/conf.py

@@ -10,7 +10,7 @@ import mlx.core as mx
 # -- Project information -----------------------------------------------------
 
 project = "MLX"
-copyright = "2023, MLX Contributors"
+copyright = "2023, Apple"
 author = "MLX Contributors"
 version = ".".join(mx.__version__.split(".")[:3])
 release = version
@@ -18,6 +18,7 @@ release = version
 # -- General configuration ---------------------------------------------------
 
 extensions = [
+    "sphinx_copybutton",
     "sphinx.ext.autodoc",
     "sphinx.ext.autosummary",
     "sphinx.ext.intersphinx",

docs/src/dev/custom_metal_kernels.rst

@@ -8,23 +8,26 @@ MLX supports writing custom Metal kernels through the Python and C++ APIs.
 Simple Example
 --------------
 
+.. currentmodule:: mlx.core
+
 Let's write a custom kernel that computes ``exp`` elementwise:
 
 .. code-block:: python
 
-  def exp_elementwise(a: mx.array):
-      source = """
-          uint elem = thread_position_in_grid.x;
-          T tmp = inp[elem];
-          out[elem] = metal::exp(tmp);
-      """
+  source = """
+      uint elem = thread_position_in_grid.x;
+      T tmp = inp[elem];
+      out[elem] = metal::exp(tmp);
+  """
 
-      kernel = mx.fast.metal_kernel(
-          name="myexp",
-          input_names=["inp"],
-          output_names=["out"],
-          source=source,
-      )
+  kernel = mx.fast.metal_kernel(
+      name="myexp",
+      input_names=["inp"],
+      output_names=["out"],
+      source=source,
+  )
+
+  def exp_elementwise(a: mx.array):
       outputs = kernel(
           inputs=[a],
           template=[("T", mx.float32)],
@@ -39,8 +42,13 @@ Let's write a custom kernel that computes ``exp`` elementwise:
   b = exp_elementwise(a)
   assert mx.allclose(b, mx.exp(a))
 
+Every time you make a kernel, a new Metal library is created and possibly
+JIT compiled. To reduce the overhead from that, build the kernel once with
+:func:`fast.metal_kernel` and then use it many times.
+
 .. note::
-    We are only required to pass the body of the Metal kernel in ``source``.
+   Only pass the body of the Metal kernel in ``source``. The function
+   signature is generated automatically.
 
 The full function signature will be generated using:
 
@@ -78,44 +86,52 @@ Putting this all together, the generated function signature for ``myexp`` is as
 
   template [[host_name("custom_kernel_myexp_float")]] [[kernel]] decltype(custom_kernel_myexp_float<float>) custom_kernel_myexp_float<float>;
 
-Note: ``grid`` and ``threadgroup`` are parameters to the Metal `dispatchThreads <https://developer.apple.com/documentation/metal/mtlcomputecommandencoder/2866532-dispatchthreads>`_ function.
-This means we will launch ``mx.prod(grid)`` threads, subdivided into ``threadgroup`` size threadgroups.
-For optimal performance, each thread group dimension should be less than or equal to the corresponding grid dimension.
+Note: ``grid`` and ``threadgroup`` are parameters to the Metal `dispatchThreads
+<https://developer.apple.com/documentation/metal/mtlcomputecommandencoder/2866532-dispatchthreads>`_
+function. This means we will launch ``mx.prod(grid)`` threads, subdivided into
+``threadgroup`` size threadgroups.  For optimal performance, each thread group
+dimension should be less than or equal to the corresponding grid dimension.
 
-Passing ``verbose=True`` to ``mx.fast.metal_kernel.__call__`` will print the generated code for debugging purposes.
+Passing ``verbose=True`` to :func:`ast.metal_kernel.__call__` will print the
+generated code for debugging purposes.
 
 Using Shape/Strides
 -------------------
 
-``mx.fast.metal_kernel`` supports an argument ``ensure_row_contiguous`` which is ``True`` by default.
-This will copy the ``mx.array`` inputs if needed before the kernel is launched to ensure that the memory layout is row contiguous.
-Generally this makes writing the kernel easier, since we don't have to worry about gaps or the ordering of the dims
-when indexing.
+:func:`fast.metal_kernel` supports an argument ``ensure_row_contiguous`` which
+is ``True`` by default. This will copy the array inputs if needed
+before the kernel is launched to ensure that the memory layout is row
+contiguous.  Generally this makes writing the kernel easier, since we don't
+have to worry about gaps or the ordering of the dims when indexing.
 
-If we want to avoid this copy, ``metal_kernel`` automatically passes ``a_shape``, ``a_strides`` and ``a_ndim`` for each
-input array ``a`` if any are present in ``source``.
-We can then use MLX's built in indexing utils to fetch the right elements for each thread.
+If we want to avoid this copy, :func:`fast.metal_kernel` automatically passes
+``a_shape``, ``a_strides`` and ``a_ndim`` for each input array ``a`` if any are
+present in ``source``. We can then use MLX's built in indexing utils to fetch
+the right elements for each thread.
 
-Let's convert ``myexp`` above to support arbitrarily strided arrays without relying on a copy from ``ensure_row_contiguous``:
+Let's convert ``myexp`` above to support arbitrarily strided arrays without
+relying on a copy from ``ensure_row_contiguous``:
 
 .. code-block:: python
+   
+  source = """
+      uint elem = thread_position_in_grid.x;
+      // Utils from `mlx/backend/metal/kernels/utils.h` are automatically included
+      uint loc = elem_to_loc(elem, inp_shape, inp_strides, inp_ndim);
+      T tmp = inp[loc];
+      // Output arrays are always row contiguous
+      out[elem] = metal::exp(tmp);
+  """
+
+  kernel = mx.fast.metal_kernel(
+      name="myexp_strided",
+      input_names=["inp"],
+      output_names=["out"],
+      source=source,
+      ensure_row_contiguous=False,
+  )
 
   def exp_elementwise(a: mx.array):
-      source = """
-          uint elem = thread_position_in_grid.x;
-          // Utils from `mlx/backend/metal/kernels/utils.h` are automatically included
-          uint loc = elem_to_loc(elem, inp_shape, inp_strides, inp_ndim);
-          T tmp = inp[loc];
-          // Output arrays are always row contiguous
-          out[elem] = metal::exp(tmp);
-      """
-
-      kernel = mx.fast.metal_kernel(
-          name="myexp_strided",
-          input_names=["inp"],
-          output_names=["out"],
-          source=source
-      )
       outputs = kernel(
           inputs=[a],
           template=[("T", mx.float32)],
@@ -123,7 +139,6 @@ Let's convert ``myexp`` above to support arbitrarily strided arrays without rely
           threadgroup=(256, 1, 1),
           output_shapes=[a.shape],
           output_dtypes=[a.dtype],
-          ensure_row_contiguous=False,
       )
       return outputs[0]
 
@@ -142,137 +157,139 @@ We'll start with the following MLX implementation using standard ops:
 
 .. code-block:: python
 
-    def grid_sample_ref(x, grid):
-        N, H_in, W_in, _ = x.shape
-        ix = ((grid[..., 0] + 1) * W_in - 1) / 2
-        iy = ((grid[..., 1] + 1) * H_in - 1) / 2
+  def grid_sample_ref(x, grid):
+      N, H_in, W_in, _ = x.shape
+      ix = ((grid[..., 0] + 1) * W_in - 1) / 2
+      iy = ((grid[..., 1] + 1) * H_in - 1) / 2
 
-        ix_nw = mx.floor(ix).astype(mx.int32)
-        iy_nw = mx.floor(iy).astype(mx.int32)
+      ix_nw = mx.floor(ix).astype(mx.int32)
+      iy_nw = mx.floor(iy).astype(mx.int32)
 
-        ix_ne = ix_nw + 1
-        iy_ne = iy_nw
+      ix_ne = ix_nw + 1
+      iy_ne = iy_nw
 
-        ix_sw = ix_nw
-        iy_sw = iy_nw + 1
+      ix_sw = ix_nw
+      iy_sw = iy_nw + 1
 
-        ix_se = ix_nw + 1
-        iy_se = iy_nw + 1
+      ix_se = ix_nw + 1
+      iy_se = iy_nw + 1
 
-        nw = (ix_se - ix)    * (iy_se - iy)
-        ne = (ix    - ix_sw) * (iy_sw - iy)
-        sw = (ix_ne - ix)    * (iy    - iy_ne)
-        se = (ix    - ix_nw) * (iy    - iy_nw)
+      nw = (ix_se - ix)    * (iy_se - iy)
+      ne = (ix    - ix_sw) * (iy_sw - iy)
+      sw = (ix_ne - ix)    * (iy    - iy_ne)
+      se = (ix    - ix_nw) * (iy    - iy_nw)
 
-        I_nw = x[mx.arange(N)[:, None, None], iy_nw, ix_nw, :]
-        I_ne = x[mx.arange(N)[:, None, None], iy_ne, ix_ne, :]
-        I_sw = x[mx.arange(N)[:, None, None], iy_sw, ix_sw, :]
-        I_se = x[mx.arange(N)[:, None, None], iy_se, ix_se, :]
+      I_nw = x[mx.arange(N)[:, None, None], iy_nw, ix_nw, :]
+      I_ne = x[mx.arange(N)[:, None, None], iy_ne, ix_ne, :]
+      I_sw = x[mx.arange(N)[:, None, None], iy_sw, ix_sw, :]
+      I_se = x[mx.arange(N)[:, None, None], iy_se, ix_se, :]
 
-        mask_nw = (iy_nw >= 0) & (iy_nw <= H_in - 1) & (ix_nw >= 0) & (ix_nw <= W_in - 1)
-        mask_ne = (iy_ne >= 0) & (iy_ne <= H_in - 1) & (ix_ne >= 0) & (ix_ne <= W_in - 1)
-        mask_sw = (iy_sw >= 0) & (iy_sw <= H_in - 1) & (ix_sw >= 0) & (ix_sw <= W_in - 1)
-        mask_se = (iy_se >= 0) & (iy_se <= H_in - 1) & (ix_se >= 0) & (ix_se <= W_in - 1)
+      mask_nw = (iy_nw >= 0) & (iy_nw <= H_in - 1) & (ix_nw >= 0) & (ix_nw <= W_in - 1)
+      mask_ne = (iy_ne >= 0) & (iy_ne <= H_in - 1) & (ix_ne >= 0) & (ix_ne <= W_in - 1)
+      mask_sw = (iy_sw >= 0) & (iy_sw <= H_in - 1) & (ix_sw >= 0) & (ix_sw <= W_in - 1)
+      mask_se = (iy_se >= 0) & (iy_se <= H_in - 1) & (ix_se >= 0) & (ix_se <= W_in - 1)
 
-        I_nw *= mask_nw[..., None]
-        I_ne *= mask_ne[..., None]
-        I_sw *= mask_sw[..., None]
-        I_se *= mask_se[..., None]
+      I_nw *= mask_nw[..., None]
+      I_ne *= mask_ne[..., None]
+      I_sw *= mask_sw[..., None]
+      I_se *= mask_se[..., None]
 
-        output = nw[..., None] * I_nw + ne[..., None] * I_ne + sw[..., None] * I_sw + se[..., None] * I_se
+      output = nw[..., None] * I_nw + ne[..., None] * I_ne + sw[..., None] * I_sw + se[..., None] * I_se
 
-        return output
+      return output
 
-Now let's use ``mx.custom_function`` together with ``mx.fast.metal_kernel``
+Now let's use :func:`custom_function` together with :func:`fast.metal_kernel`
 to write a fast GPU kernel for both the forward and backward passes.
 
 First we'll implement the forward pass as a fused kernel:
 
 .. code-block:: python
 
-    @mx.custom_function
-    def grid_sample(x, grid):
+  source = """
+      uint elem = thread_position_in_grid.x;
+      int H = x_shape[1];
+      int W = x_shape[2];
+      int C = x_shape[3];
+      int gH = grid_shape[1];
+      int gW = grid_shape[2];
 
-        assert x.ndim == 4, "`x` must be 4D."
-        assert grid.ndim == 4, "`grid` must be 4D."
+      int w_stride = C;
+      int h_stride = W * w_stride;
+      int b_stride = H * h_stride;
 
-        B, _, _, C = x.shape
-        _, gN, gM, D = grid.shape
-        out_shape = (B, gN, gM, C)
+      uint grid_idx = elem / C * 2;
+      float ix = ((grid[grid_idx] + 1) * W - 1) / 2;
+      float iy = ((grid[grid_idx + 1] + 1) * H - 1) / 2;
 
-        assert D == 2, "Last dim of `grid` must be size 2."
+      int ix_nw = floor(ix);
+      int iy_nw = floor(iy);
 
-        source = """
-            uint elem = thread_position_in_grid.x;
-            int H = x_shape[1];
-            int W = x_shape[2];
-            int C = x_shape[3];
-            int gH = grid_shape[1];
-            int gW = grid_shape[2];
+      int ix_ne = ix_nw + 1;
+      int iy_ne = iy_nw;
 
-            int w_stride = C;
-            int h_stride = W * w_stride;
-            int b_stride = H * h_stride;
+      int ix_sw = ix_nw;
+      int iy_sw = iy_nw + 1;
 
-            uint grid_idx = elem / C * 2;
-            float ix = ((grid[grid_idx] + 1) * W - 1) / 2;
-            float iy = ((grid[grid_idx + 1] + 1) * H - 1) / 2;
+      int ix_se = ix_nw + 1;
+      int iy_se = iy_nw + 1;
 
-            int ix_nw = floor(ix);
-            int iy_nw = floor(iy);
+      T nw = (ix_se - ix)    * (iy_se - iy);
+      T ne = (ix    - ix_sw) * (iy_sw - iy);
+      T sw = (ix_ne - ix)    * (iy    - iy_ne);
+      T se = (ix    - ix_nw) * (iy    - iy_nw);
 
-            int ix_ne = ix_nw + 1;
-            int iy_ne = iy_nw;
+      int batch_idx = elem / C / gH / gW * b_stride;
+      int channel_idx = elem % C;
+      int base_idx = batch_idx + channel_idx;
 
-            int ix_sw = ix_nw;
-            int iy_sw = iy_nw + 1;
+      T I_nw = x[base_idx + iy_nw * h_stride + ix_nw * w_stride];
+      T I_ne = x[base_idx + iy_ne * h_stride + ix_ne * w_stride];
+      T I_sw = x[base_idx + iy_sw * h_stride + ix_sw * w_stride];
+      T I_se = x[base_idx + iy_se * h_stride + ix_se * w_stride];
 
-            int ix_se = ix_nw + 1;
-            int iy_se = iy_nw + 1;
+      I_nw = iy_nw >= 0 && iy_nw <= H - 1 && ix_nw >= 0 && ix_nw <= W - 1 ? I_nw : 0;
+      I_ne = iy_ne >= 0 && iy_ne <= H - 1 && ix_ne >= 0 && ix_ne <= W - 1 ? I_ne : 0;
+      I_sw = iy_sw >= 0 && iy_sw <= H - 1 && ix_sw >= 0 && ix_sw <= W - 1 ? I_sw : 0;
+      I_se = iy_se >= 0 && iy_se <= H - 1 && ix_se >= 0 && ix_se <= W - 1 ? I_se : 0;
 
-            T nw = (ix_se - ix)    * (iy_se - iy);
-            T ne = (ix    - ix_sw) * (iy_sw - iy);
-            T sw = (ix_ne - ix)    * (iy    - iy_ne);
-            T se = (ix    - ix_nw) * (iy    - iy_nw);
+      out[elem] = nw * I_nw + ne * I_ne + sw * I_sw + se * I_se;
+  """
 
-            int batch_idx = elem / C / gH / gW * b_stride;
-            int channel_idx = elem % C;
-            int base_idx = batch_idx + channel_idx;
+  kernel = mx.fast.metal_kernel(
+      name="grid_sample",
+      input_names=["x", "grid"],
+      output_names=["out"],
+      source=source,
+  )
 
-            T I_nw = x[base_idx + iy_nw * h_stride + ix_nw * w_stride];
-            T I_ne = x[base_idx + iy_ne * h_stride + ix_ne * w_stride];
-            T I_sw = x[base_idx + iy_sw * h_stride + ix_sw * w_stride];
-            T I_se = x[base_idx + iy_se * h_stride + ix_se * w_stride];
+  @mx.custom_function
+  def grid_sample(x, grid):
 
-            I_nw = iy_nw >= 0 && iy_nw <= H - 1 && ix_nw >= 0 && ix_nw <= W - 1 ? I_nw : 0;
-            I_ne = iy_ne >= 0 && iy_ne <= H - 1 && ix_ne >= 0 && ix_ne <= W - 1 ? I_ne : 0;
-            I_sw = iy_sw >= 0 && iy_sw <= H - 1 && ix_sw >= 0 && ix_sw <= W - 1 ? I_sw : 0;
-            I_se = iy_se >= 0 && iy_se <= H - 1 && ix_se >= 0 && ix_se <= W - 1 ? I_se : 0;
+      assert x.ndim == 4, "`x` must be 4D."
+      assert grid.ndim == 4, "`grid` must be 4D."
 
-            out[elem] = nw * I_nw + ne * I_ne + sw * I_sw + se * I_se;
-        """
-        kernel = mx.fast.metal_kernel(
-            name="grid_sample",
-            input_names=["x", "grid"],
-            output_names=["out"],
-            source=source,
-        )
-        outputs = kernel(
-            inputs=[x, grid],
-            template=[("T", x.dtype)],
-            output_shapes=[out_shape],
-            output_dtypes=[x.dtype],
-            grid=(np.prod(out_shape), 1, 1),
-            threadgroup=(256, 1, 1),
-        )
-        return outputs[0]
+      B, _, _, C = x.shape
+      _, gN, gM, D = grid.shape
+      out_shape = (B, gN, gM, C)
+
+      assert D == 2, "Last dim of `grid` must be size 2."
+
+      outputs = kernel(
+          inputs=[x, grid],
+          template=[("T", x.dtype)],
+          output_shapes=[out_shape],
+          output_dtypes=[x.dtype],
+          grid=(np.prod(out_shape), 1, 1),
+          threadgroup=(256, 1, 1),
+      )
+      return outputs[0]
 
 For a reasonably sized input such as:
 
 .. code-block:: python
 
-    x.shape = (8, 1024, 1024, 64)
-    grid.shape = (8, 256, 256, 2)
+  x.shape = (8, 1024, 1024, 64)
+  grid.shape = (8, 256, 256, 2)
 
 On an M1 Max, we see a big performance improvement:
 
@@ -281,11 +298,11 @@ On an M1 Max, we see a big performance improvement:
 Grid Sample VJP
 ---------------
 
-Since we decorated ``grid_sample`` with ``mx.custom_function``, we can now define
-its custom vjp transform so MLX can differentiate it.
+Since we decorated ``grid_sample`` with :func:`custom_function`, we can now
+define its custom vjp transform so MLX can differentiate it.
 
 The backwards pass requires atomically updating ``x_grad``/``grid_grad`` and so
-requires a few extra ``mx.fast.metal_kernel`` features:
+requires a few extra :func:`fast.metal_kernel` features:
 
 * ``init_value=0``
     Initialize all of the kernel's outputs to this value before it runs. This allows us to update only part of the output arrays with the kernel.
@@ -299,128 +316,129 @@ We can then implement the backwards pass as follows:
 
 .. code-block:: python
 
-    @grid_sample.vjp
-    def grid_sample_vjp(primals, cotangent, _):
-        x, grid = primals
-        B, _, _, C = x.shape
-        _, gN, gM, D = grid.shape
+  source = """
+      uint elem = thread_position_in_grid.x;
+      int H = x_shape[1];
+      int W = x_shape[2];
+      int C = x_shape[3];
+      // Pad C to the nearest larger simdgroup size multiple
+      int C_padded = ceildiv(C, threads_per_simdgroup) * threads_per_simdgroup;
 
-        assert D == 2, "Last dim of `grid` must be size 2."
+      int gH = grid_shape[1];
+      int gW = grid_shape[2];
 
-        source = """
-            uint elem = thread_position_in_grid.x;
-            int H = x_shape[1];
-            int W = x_shape[2];
-            int C = x_shape[3];
-            // Pad C to the nearest larger simdgroup size multiple
-            int C_padded = ceildiv(C, threads_per_simdgroup) * threads_per_simdgroup;
+      int w_stride = C;
+      int h_stride = W * w_stride;
+      int b_stride = H * h_stride;
 
-            int gH = grid_shape[1];
-            int gW = grid_shape[2];
+      uint grid_idx = elem / C_padded * 2;
+      float ix = ((grid[grid_idx] + 1) * W - 1) / 2;
+      float iy = ((grid[grid_idx + 1] + 1) * H - 1) / 2;
 
-            int w_stride = C;
-            int h_stride = W * w_stride;
-            int b_stride = H * h_stride;
+      int ix_nw = floor(ix);
+      int iy_nw = floor(iy);
 
-            uint grid_idx = elem / C_padded * 2;
-            float ix = ((grid[grid_idx] + 1) * W - 1) / 2;
-            float iy = ((grid[grid_idx + 1] + 1) * H - 1) / 2;
+      int ix_ne = ix_nw + 1;
+      int iy_ne = iy_nw;
 
-            int ix_nw = floor(ix);
-            int iy_nw = floor(iy);
+      int ix_sw = ix_nw;
+      int iy_sw = iy_nw + 1;
 
-            int ix_ne = ix_nw + 1;
-            int iy_ne = iy_nw;
+      int ix_se = ix_nw + 1;
+      int iy_se = iy_nw + 1;
 
-            int ix_sw = ix_nw;
-            int iy_sw = iy_nw + 1;
+      T nw = (ix_se - ix)    * (iy_se - iy);
+      T ne = (ix    - ix_sw) * (iy_sw - iy);
+      T sw = (ix_ne - ix)    * (iy    - iy_ne);
+      T se = (ix    - ix_nw) * (iy    - iy_nw);
 
-            int ix_se = ix_nw + 1;
-            int iy_se = iy_nw + 1;
+      int batch_idx = elem / C_padded / gH / gW * b_stride;
+      int channel_idx = elem % C_padded;
+      int base_idx = batch_idx + channel_idx;
 
-            T nw = (ix_se - ix)    * (iy_se - iy);
-            T ne = (ix    - ix_sw) * (iy_sw - iy);
-            T sw = (ix_ne - ix)    * (iy    - iy_ne);
-            T se = (ix    - ix_nw) * (iy    - iy_nw);
+      T gix = T(0);
+      T giy = T(0);
+      if (channel_idx < C) {
+          int cot_index = elem / C_padded * C + channel_idx;
+          T cot = cotangent[cot_index];
+          if (iy_nw >= 0 && iy_nw <= H - 1 && ix_nw >= 0 && ix_nw <= W - 1) {
+              int offset = base_idx + iy_nw * h_stride + ix_nw * w_stride;
+              atomic_fetch_add_explicit(&x_grad[offset], nw * cot, memory_order_relaxed);
 
-            int batch_idx = elem / C_padded / gH / gW * b_stride;
-            int channel_idx = elem % C_padded;
-            int base_idx = batch_idx + channel_idx;
+              T I_nw = x[offset];
+              gix -= I_nw * (iy_se - iy) * cot;
+              giy -= I_nw * (ix_se - ix) * cot;
+          }
+          if (iy_ne >= 0 && iy_ne <= H - 1 && ix_ne >= 0 && ix_ne <= W - 1) {
+              int offset = base_idx + iy_ne * h_stride + ix_ne * w_stride;
+              atomic_fetch_add_explicit(&x_grad[offset], ne * cot, memory_order_relaxed);
 
-            T gix = T(0);
-            T giy = T(0);
-            if (channel_idx < C) {
-                int cot_index = elem / C_padded * C + channel_idx;
-                T cot = cotangent[cot_index];
-                if (iy_nw >= 0 && iy_nw <= H - 1 && ix_nw >= 0 && ix_nw <= W - 1) {
-                    int offset = base_idx + iy_nw * h_stride + ix_nw * w_stride;
-                    atomic_fetch_add_explicit(&x_grad[offset], nw * cot, memory_order_relaxed);
+              T I_ne = x[offset];
+              gix += I_ne * (iy_sw - iy) * cot;
+              giy -= I_ne * (ix - ix_sw) * cot;
+          }
+          if (iy_sw >= 0 && iy_sw <= H - 1 && ix_sw >= 0 && ix_sw <= W - 1) {
+              int offset = base_idx + iy_sw * h_stride + ix_sw * w_stride;
+              atomic_fetch_add_explicit(&x_grad[offset], sw * cot, memory_order_relaxed);
 
-                    T I_nw = x[offset];
-                    gix -= I_nw * (iy_se - iy) * cot;
-                    giy -= I_nw * (ix_se - ix) * cot;
-                }
-                if (iy_ne >= 0 && iy_ne <= H - 1 && ix_ne >= 0 && ix_ne <= W - 1) {
-                    int offset = base_idx + iy_ne * h_stride + ix_ne * w_stride;
-                    atomic_fetch_add_explicit(&x_grad[offset], ne * cot, memory_order_relaxed);
+              T I_sw = x[offset];
+              gix -= I_sw * (iy - iy_ne) * cot;
+              giy += I_sw * (ix_ne - ix) * cot;
+          }
+          if (iy_se >= 0 && iy_se <= H - 1 && ix_se >= 0 && ix_se <= W - 1) {
+              int offset = base_idx + iy_se * h_stride + ix_se * w_stride;
+              atomic_fetch_add_explicit(&x_grad[offset], se * cot, memory_order_relaxed);
 
-                    T I_ne = x[offset];
-                    gix += I_ne * (iy_sw - iy) * cot;
-                    giy -= I_ne * (ix - ix_sw) * cot;
-                }
-                if (iy_sw >= 0 && iy_sw <= H - 1 && ix_sw >= 0 && ix_sw <= W - 1) {
-                    int offset = base_idx + iy_sw * h_stride + ix_sw * w_stride;
-                    atomic_fetch_add_explicit(&x_grad[offset], sw * cot, memory_order_relaxed);
+              T I_se = x[offset];
+              gix += I_se * (iy - iy_nw) * cot;
+              giy += I_se * (ix - ix_nw) * cot;
+          }
+      }
 
-                    T I_sw = x[offset];
-                    gix -= I_sw * (iy - iy_ne) * cot;
-                    giy += I_sw * (ix_ne - ix) * cot;
-                }
-                if (iy_se >= 0 && iy_se <= H - 1 && ix_se >= 0 && ix_se <= W - 1) {
-                    int offset = base_idx + iy_se * h_stride + ix_se * w_stride;
-                    atomic_fetch_add_explicit(&x_grad[offset], se * cot, memory_order_relaxed);
+      T gix_mult = W / 2;
+      T giy_mult = H / 2;
 
-                    T I_se = x[offset];
-                    gix += I_se * (iy - iy_nw) * cot;
-                    giy += I_se * (ix - ix_nw) * cot;
-                }
-            }
+      // Reduce across each simdgroup first.
+      // This is much faster than relying purely on atomics.
+      gix = simd_sum(gix);
+      giy = simd_sum(giy);
 
-            T gix_mult = W / 2;
-            T giy_mult = H / 2;
+      if (thread_index_in_simdgroup == 0) {
+          atomic_fetch_add_explicit(&grid_grad[grid_idx], gix * gix_mult, memory_order_relaxed);
+          atomic_fetch_add_explicit(&grid_grad[grid_idx + 1], giy * giy_mult, memory_order_relaxed);
+      }
+  """
+  kernel = mx.fast.metal_kernel(
+      name="grid_sample_grad",
+      input_names=["x", "grid", "cotangent"],
+      output_names=["x_grad", "grid_grad"],
+      source=source,
+      atomic_outputs=True,
+  )
 
-            // Reduce across each simdgroup first.
-            // This is much faster than relying purely on atomics.
-            gix = simd_sum(gix);
-            giy = simd_sum(giy);
+  @grid_sample.vjp
+  def grid_sample_vjp(primals, cotangent, _):
+      x, grid = primals
+      B, _, _, C = x.shape
+      _, gN, gM, D = grid.shape
 
-            if (thread_index_in_simdgroup == 0) {
-                atomic_fetch_add_explicit(&grid_grad[grid_idx], gix * gix_mult, memory_order_relaxed);
-                atomic_fetch_add_explicit(&grid_grad[grid_idx + 1], giy * giy_mult, memory_order_relaxed);
-            }
-        """
-        kernel = mx.fast.metal_kernel(
-            name="grid_sample_grad",
-            input_names=["x", "grid", "cotangent"],
-            output_names=["x_grad", "grid_grad"],
-            source=source,
-            atomic_outputs=True,
-        )
-        # pad the output channels to simd group size
-        # so that our `simd_sum`s don't overlap.
-        simdgroup_size = 32
-        C_padded = (C + simdgroup_size - 1) // simdgroup_size * simdgroup_size
-        grid_size = B * gN * gM * C_padded
-        outputs = kernel(
-            inputs=[x, grid, cotangent],
-            template=[("T", x.dtype)],
-            output_shapes=[x.shape, grid.shape],
-            output_dtypes=[x.dtype, x.dtype],
-            grid=(grid_size, 1, 1),
-            threadgroup=(256, 1, 1),
-            init_value=0,
-        )
-        return outputs[0], outputs[1]
+      assert D == 2, "Last dim of `grid` must be size 2."
+
+      # pad the output channels to simd group size
+      # so that our `simd_sum`s don't overlap.
+      simdgroup_size = 32
+      C_padded = (C + simdgroup_size - 1) // simdgroup_size * simdgroup_size
+      grid_size = B * gN * gM * C_padded
+      outputs = kernel(
+          inputs=[x, grid, cotangent],
+          template=[("T", x.dtype)],
+          output_shapes=[x.shape, grid.shape],
+          output_dtypes=[x.dtype, x.dtype],
+          grid=(grid_size, 1, 1),
+          threadgroup=(256, 1, 1),
+          init_value=0,
+      )
+      return outputs[0], outputs[1]
 
 There's an even larger speed up for the vjp:
 

docs/src/dev/extensions.rst

@@ -22,12 +22,12 @@ You can do that in MLX directly:
 This function performs that operation while leaving the implementation and
 function transformations to MLX.
 
-However you may need to customize the underlying implementation, perhaps to
-make it faster or for custom differentiation. In this tutorial we will go
-through adding custom extensions. It will cover:
+However, you may want to customize the underlying implementation, perhaps to
+make it faster. In this tutorial we will go through adding custom extensions.
+It will cover:
 
 * The structure of the MLX library.
-* Implementing a CPU operation that redirects to Accelerate_ when appropriate.
+* Implementing a CPU operation.
 * Implementing a GPU operation using metal.
 * Adding the ``vjp`` and ``jvp`` function transformation.
 * Building a custom extension and binding it to python.
@@ -45,7 +45,7 @@ Operations
 Operations are the front-end functions that operate on arrays. They are defined
 in the C++ API (:ref:`cpp_ops`), and the Python API (:ref:`ops`) binds them.
 
-We would like an operation, :meth:`axpby` that takes in two arrays ``x`` and
+We would like an operation :meth:`axpby` that takes in two arrays, ``x`` and
 ``y``, and two scalars, ``alpha`` and ``beta``. This is how to define it in
 C++:
 
@@ -55,7 +55,7 @@ C++:
     *  Scale and sum two vectors element-wise
     *  z = alpha * x + beta * y
     *
-    *  Follow numpy style broadcasting between x and y
+    *  Use NumPy-style broadcasting between x and y
     *  Inputs are upcasted to floats if needed
     **/
     array axpby(
@@ -66,7 +66,7 @@ C++:
         StreamOrDevice s = {} // Stream on which to schedule the operation
     );
 
-The simplest way to this operation is in terms of existing operations:
+The simplest way to implement this is with existing operations:
 
 .. code-block:: C++
 
@@ -93,9 +93,9 @@ Primitives
 ^^^^^^^^^^^
 
 A :class:`Primitive` is part of the computation graph of an :class:`array`. It
-defines how to create outputs arrays given a input arrays. Further, a
+defines how to create output arrays given input arrays. Further, a
 :class:`Primitive` has methods to run on the CPU or GPU and for function
-transformations such as ``vjp`` and ``jvp``.  Lets go back to our example to be
+transformations such as ``vjp`` and ``jvp``.  Let's go back to our example to be
 more concrete:
 
 .. code-block:: C++
@@ -128,7 +128,7 @@ more concrete:
         /** The vector-Jacobian product. */
         std::vector<array> vjp(
             const std::vector<array>& primals,
-            const array& cotan,
+            const std::vector<array>& cotangents,
             const std::vector<int>& argnums,
             const std::vector<array>& outputs) override;
 
@@ -138,13 +138,13 @@ more concrete:
         * representing the vectorized computation and the axis which
         * corresponds to the output vectorized dimension.
         */
-        virtual std::pair<std::vector<array>, std::vector<int>> vmap(
+        std::pair<std::vector<array>, std::vector<int>> vmap(
             const std::vector<array>& inputs,
             const std::vector<int>& axes) override;
 
-        /** Print the primitive. */
-        void print(std::ostream& os) override {
-            os << "Axpby";
+        /** The name of primitive. */
+        const char* name() const override {
+          return "Axpby";
         }
 
         /** Equivalence check **/
@@ -153,9 +153,6 @@ more concrete:
       private:
         float alpha_;
         float beta_;
-
-        /** Fall back implementation for evaluation on CPU */
-        void eval(const std::vector<array>& inputs, array& out);
     };
 
 The :class:`Axpby` class derives from the base :class:`Primitive` class. The
@@ -188,7 +185,7 @@ Let's reimplement our operation now in terms of our :class:`Axpby` primitive.
         auto promoted_dtype = promote_types(x.dtype(), y.dtype());
 
         // Upcast to float32 for non-floating point inputs x and y
-        auto out_dtype = is_floating_point(promoted_dtype)
+        auto out_dtype = issubdtype(promoted_dtype, float32)
             ? promoted_dtype
             : promote_types(promoted_dtype, float32);
 
@@ -234,49 +231,57 @@ the execution of the computation graph, and calls :meth:`Axpby::eval_cpu` or
 Implementing the CPU Back-end
 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
 
-Let's start by implementing a naive and generic version of
-:meth:`Axpby::eval_cpu`. We declared this as a private member function of
-:class:`Axpby` earlier called :meth:`Axpby::eval`.
+Let's start by implementing :meth:`Axpby::eval_cpu`.
 
-Our naive method will go over each element of the output array, find the
+The method will go over each element of the output array, find the
 corresponding input elements of ``x`` and ``y`` and perform the operation
 point-wise. This is captured in the templated function :meth:`axpby_impl`.
 
 .. code-block:: C++
 
-    template <typename T>
-    void axpby_impl(
-            const array& x,
-            const array& y,
-            array& out,
-            float alpha_,
-            float beta_) {
-        // We only allocate memory when we are ready to fill the output
-        // malloc_or_wait synchronously allocates available memory
-        // There may be a wait executed here if the allocation is requested
-        // under memory-pressured conditions
-        out.set_data(allocator::malloc_or_wait(out.nbytes()));
+  template <typename T>
+  void axpby_impl(
+      const mx::array& x,
+      const mx::array& y,
+      mx::array& out,
+      float alpha_,
+      float beta_,
+      mx::Stream stream) {
+    out.set_data(mx::allocator::malloc(out.nbytes()));
 
-        // Collect input and output data pointers
-        const T* x_ptr = x.data<T>();
-        const T* y_ptr = y.data<T>();
-        T* out_ptr = out.data<T>();
+    // Get the CPU command encoder and register input and output arrays
+    auto& encoder = mx::cpu::get_command_encoder(stream);
+    encoder.set_input_array(x);
+    encoder.set_input_array(y);
+    encoder.set_output_array(out);
 
-        // Cast alpha and beta to the relevant types
-        T alpha = static_cast<T>(alpha_);
-        T beta = static_cast<T>(beta_);
+    // Launch the CPU kernel
+    encoder.dispatch([x_ptr = x.data<T>(),
+                      y_ptr = y.data<T>(),
+                      out_ptr = out.data<T>(),
+                      size = out.size(),
+                      shape = out.shape(),
+                      x_strides = x.strides(),
+                      y_strides = y.strides(),
+                      alpha_,
+                      beta_]() {
 
-        // Do the element-wise operation for each output
-        for (size_t out_idx = 0; out_idx < out.size(); out_idx++) {
-            // Map linear indices to offsets in x and y
-            auto x_offset = elem_to_loc(out_idx, x.shape(), x.strides());
-            auto y_offset = elem_to_loc(out_idx, y.shape(), y.strides());
+      // Cast alpha and beta to the relevant types
+      T alpha = static_cast<T>(alpha_);
+      T beta = static_cast<T>(beta_);
 
-            // We allocate the output to be contiguous and regularly strided
-            // (defaults to row major) and hence it doesn't need additional mapping
-            out_ptr[out_idx] = alpha * x_ptr[x_offset] + beta * y_ptr[y_offset];
-        }
-    }
+      // Do the element-wise operation for each output
+      for (size_t out_idx = 0; out_idx < size; out_idx++) {
+        // Map linear indices to offsets in x and y
+        auto x_offset = mx::elem_to_loc(out_idx, shape, x_strides);
+        auto y_offset = mx::elem_to_loc(out_idx, shape, y_strides);
+
+        // We allocate the output to be contiguous and regularly strided
+        // (defaults to row major) and hence it doesn't need additional mapping
+        out_ptr[out_idx] = alpha * x_ptr[x_offset] + beta * y_ptr[y_offset];
+      }
+    });
+  }
 
 Our implementation should work for all incoming floating point arrays.
 Accordingly, we add dispatches for ``float32``, ``float16``, ``bfloat16`` and
@@ -284,112 +289,32 @@ Accordingly, we add dispatches for ``float32``, ``float16``, ``bfloat16`` and
 
 .. code-block:: C++
 
-    /** Fall back implementation for evaluation on CPU */
-    void Axpby::eval(
-      const std::vector<array>& inputs,
-      const std::vector<array>& outputs) {
-        auto& x = inputs[0];
-        auto& y = inputs[1];
-        auto& out = outputs[0];
-
-        // Dispatch to the correct dtype
-        if (out.dtype() == float32) {
-            return axpby_impl<float>(x, y, out, alpha_, beta_);
-        } else if (out.dtype() == float16) {
-            return axpby_impl<float16_t>(x, y, out, alpha_, beta_);
-        } else if (out.dtype() == bfloat16) {
-            return axpby_impl<bfloat16_t>(x, y, out, alpha_, beta_);
-        } else if (out.dtype() == complex64) {
-            return axpby_impl<complex64_t>(x, y, out, alpha_, beta_);
-        } else {
-            throw std::runtime_error(
-                "[Axpby] Only supports floating point types.");
-        }
-    }
-
-This is good as a fallback implementation. We can use the ``axpby`` routine
-provided by the Accelerate_ framework for a faster implementation in certain
-cases:
-
-#.  Accelerate does not provide implementations of ``axpby`` for half precision
-    floats. We can only use it for ``float32`` types.
-#.  Accelerate assumes the inputs ``x`` and ``y`` are contiguous and all
-    elements have fixed strides between them. We only direct to Accelerate
-    if both ``x`` and ``y`` are row contiguous or column contiguous.
-#.  Accelerate performs the routine ``Y = (alpha * X) + (beta * Y)`` in-place.
-    MLX expects to write the output to a new array. We must copy the elements
-    of ``y`` into the output and use that as an input to ``axpby``.
-
-Let's write an implementation that uses Accelerate in the right conditions.
-It allocates data for the output, copies ``y`` into it, and then calls the
-:func:`catlas_saxpby` from accelerate.
-
-.. code-block:: C++
-
-    template <typename T>
-    void axpby_impl_accelerate(
-            const array& x,
-            const array& y,
-            array& out,
-            float alpha_,
-            float beta_) {
-        // Accelerate library provides catlas_saxpby which does
-        // Y = (alpha * X) + (beta * Y) in place
-        // To use it, we first copy the data in y over to the output array
-        out.set_data(allocator::malloc_or_wait(out.nbytes()));
-
-        // We then copy over the elements using the contiguous vector specialization
-        copy_inplace(y, out, CopyType::Vector);
-
-        // Get x and y pointers for catlas_saxpby
-        const T* x_ptr = x.data<T>();
-        T* y_ptr = out.data<T>();
-
-        T alpha = static_cast<T>(alpha_);
-        T beta = static_cast<T>(beta_);
-
-        // Call the inplace accelerate operator
-        catlas_saxpby(
-            /* N = */ out.size(),
-            /* ALPHA = */ alpha,
-            /* X = */ x_ptr,
-            /* INCX = */ 1,
-            /* BETA = */ beta,
-            /* Y = */ y_ptr,
-            /* INCY = */ 1);
-    }
-
-For inputs that do not fit the criteria for accelerate, we fall back to
-:meth:`Axpby::eval`. With this in mind, let's finish our
-:meth:`Axpby::eval_cpu`.
-
-.. code-block:: C++
-
-    /** Evaluate primitive on CPU using accelerate specializations */
     void Axpby::eval_cpu(
-      const std::vector<array>& inputs,
-      const std::vector<array>& outputs) {
-        assert(inputs.size() == 2);
-        auto& x = inputs[0];
-        auto& y = inputs[1];
-        auto& out = outputs[0];
+        const std::vector<mx::array>& inputs,
+        std::vector<mx::array>& outputs) {
+      auto& x = inputs[0];
+      auto& y = inputs[1];
+      auto& out = outputs[0];
 
-        // Accelerate specialization for contiguous single precision float arrays
-        if (out.dtype() == float32 &&
-            ((x.flags().row_contiguous && y.flags().row_contiguous) ||
-            (x.flags().col_contiguous && y.flags().col_contiguous))) {
-            axpby_impl_accelerate<float>(x, y, out, alpha_, beta_);
-            return;
-        }
-
-        // Fall back to common back-end if specializations are not available
-        eval(inputs, outputs);
+      // Dispatch to the correct dtype
+      if (out.dtype() == mx::float32) {
+        return axpby_impl<float>(x, y, out, alpha_, beta_, stream());
+      } else if (out.dtype() == mx::float16) {
+        return axpby_impl<mx::float16_t>(x, y, out, alpha_, beta_, stream());
+      } else if (out.dtype() == mx::bfloat16) {
+        return axpby_impl<mx::bfloat16_t>(x, y, out, alpha_, beta_, stream());
+      } else if (out.dtype() == mx::complex64) {
+        return axpby_impl<mx::complex64_t>(x, y, out, alpha_, beta_, stream());
+      } else {
+        throw std::runtime_error(
+            "Axpby is only supported for floating point types.");
+      }
     }
 
 Just this much is enough to run the operation :meth:`axpby` on a CPU stream! If
 you do not plan on running the operation on the GPU or using transforms on
 computation graphs that contain :class:`Axpby`, you can stop implementing the
-primitive here and enjoy the speed-ups you get from the Accelerate library.
+primitive here.
 
 Implementing the GPU Back-end
 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
@@ -466,17 +391,17 @@ below.
         auto& d = metal::device(s.device);
 
         // Allocate output memory
-        out.set_data(allocator::malloc_or_wait(out.nbytes()));
+        out.set_data(allocator::malloc(out.nbytes()));
 
         // Resolve name of kernel
-        std::ostringstream kname;
-        kname << "axpby_" << "general_" << type_to_name(out);
+        std::stream kname;
+        kname = "axpby_general_" + type_to_name(out);
 
-        // Make sure the metal library is available
-        d.register_library("mlx_ext");
+        // Load the metal library
+        auto lib = d.get_library("mlx_ext", current_binary_dir());
 
         // Make a kernel from this metal library
-        auto kernel = d.get_kernel(kname.str(), "mlx_ext");
+        auto kernel = d.get_kernel(kname, lib);
 
         // Prepare to encode kernel
         auto& compute_encoder = d.get_command_encoder(s.index);
@@ -544,7 +469,7 @@ one we just defined:
             const std::vector<array>& tangents,
             const std::vector<int>& argnums) {
         // Forward mode diff that pushes along the tangents
-        // The jvp transform on the primitive can built with ops
+        // The jvp transform on the primitive can be built with ops
         // that are scheduled on the same stream as the primitive
 
         // If argnums = {0}, we only push along x in which case the
@@ -556,7 +481,7 @@ one we just defined:
             auto scale_arr = array(scale, tangents[0].dtype());
             return {multiply(scale_arr, tangents[0], stream())};
         }
-        // If, argnums = {0, 1}, we take contributions from both
+        // If argnums = {0, 1}, we take contributions from both
         // which gives us jvp = tangent_x * alpha + tangent_y * beta
         else {
             return {axpby(tangents[0], tangents[1], alpha_, beta_, stream())};
@@ -810,7 +735,7 @@ Let's look at a simple script and its results:
 
     print(f"c shape: {c.shape}")
     print(f"c dtype: {c.dtype}")
-    print(f"c correct: {mx.all(c == 6.0).item()}")
+    print(f"c is correct: {mx.all(c == 6.0).item()}")
 
 Output:
 
@@ -818,13 +743,13 @@ Output:
 
     c shape: [3, 4]
     c dtype: float32
-    c correctness: True
+    c is correct: True
 
 Results
 ^^^^^^^
 
 Let's run a quick benchmark and see how our new ``axpby`` operation compares
-with the naive :meth:`simple_axpby` we first defined on the CPU.
+with the naive :meth:`simple_axpby` we first defined.
 
 .. code-block:: python
 
@@ -832,13 +757,11 @@ with the naive :meth:`simple_axpby` we first defined on the CPU.
     from mlx_sample_extensions import axpby
     import time
 
-    mx.set_default_device(mx.cpu)
-
     def simple_axpby(x: mx.array, y: mx.array, alpha: float, beta: float) -> mx.array:
         return alpha * x + beta * y
 
-    M = 256
-    N = 512
+    M = 4096
+    N = 4096
 
     x = mx.random.normal((M, N))
     y = mx.random.normal((M, N))
@@ -849,24 +772,24 @@ with the naive :meth:`simple_axpby` we first defined on the CPU.
 
     def bench(f):
         # Warm up
-        for i in range(100):
+        for i in range(5):
             z = f(x, y, alpha, beta)
             mx.eval(z)
 
         # Timed run
         s = time.time()
-        for i in range(5000):
+        for i in range(100):
             z = f(x, y, alpha, beta)
             mx.eval(z)
         e = time.time()
-        return e - s
+        return 1000 * (e - s) / 100
 
     simple_time = bench(simple_axpby)
     custom_time = bench(axpby)
 
-    print(f"Simple axpby: {simple_time:.3f} s | Custom axpby: {custom_time:.3f} s")
+    print(f"Simple axpby: {simple_time:.3f} ms | Custom axpby: {custom_time:.3f} ms")
 
-The results are ``Simple axpby: 0.114 s | Custom axpby: 0.109 s``. We see
+The results are ``Simple axpby: 1.559 ms | Custom axpby: 0.774 ms``. We see
 modest improvements right away!
 
 This operation is now good to be used to build other operations, in

docs/src/index.rst

@@ -70,6 +70,8 @@ are the CPU and GPU.
    python/fft
    python/linalg
    python/metal
+   python/cuda
+   python/memory_management
    python/nn
    python/optimizers
    python/distributed

docs/src/install.rst

@@ -13,7 +13,7 @@ silicon computer is
 
     pip install mlx
 
-To install from PyPI you must meet the following requirements:
+To install from PyPI your system must meet the following requirements:
 
 - Using an M series chip (Apple silicon)
 - Using a native Python >= 3.9
@@ -23,12 +23,39 @@ To install from PyPI you must meet the following requirements:
     MLX is only available on devices running macOS >= 13.5
     It is highly recommended to use macOS 14 (Sonoma)
 
+CUDA
+^^^^
 
-MLX is also available on conda-forge. To install MLX with conda do:
+MLX has a CUDA backend which you can install with:
 
 .. code-block:: shell
 
-   conda install conda-forge::mlx
+    pip install mlx[cuda]
+
+To install the CUDA package from PyPi your system must meet the following
+requirements:
+
+- Nvidia architecture >= SM 7.0 (Volta)
+- Nvidia driver >= 550.54.14
+- CUDA toolkit >= 12.0
+- Linux distribution with glibc >= 2.35
+- Python >= 3.9
+
+
+CPU-only (Linux)
+^^^^^^^^^^^^^^^^
+
+For a CPU-only version of MLX that runs on Linux use:
+
+.. code-block:: shell
+
+    pip install mlx[cpu]
+
+To install the CPU-only package from PyPi your system must meet the following
+requirements:
+
+- Linux distribution with glibc >= 2.35
+- Python >= 3.9
 
 
 Troubleshooting
@@ -65,6 +92,8 @@ Build Requirements
 Python API
 ^^^^^^^^^^
 
+.. _python install:
+
 To build and install the MLX python library from source, first, clone MLX from
 `its GitHub repo <https://github.com/ml-explore/mlx>`_:
 
@@ -76,20 +105,20 @@ Then simply build and install MLX using pip:
 
 .. code-block:: shell
 
-  CMAKE_BUILD_PARALLEL_LEVEL=8 pip install .
+  pip install .
 
 For developing, install the package with development dependencies, and use an
 editable install:
 
 .. code-block:: shell
 
-  CMAKE_BUILD_PARALLEL_LEVEL=8 pip install -e ".[dev]"
+  pip install -e ".[dev]"
 
 Once the development dependencies are installed, you can build faster with:
 
 .. code-block:: shell
 
- CMAKE_BUILD_PARALLEL_LEVEL=8 python setup.py build_ext --inplace
+ python setup.py build_ext --inplace
 
 Run the tests with:
 
@@ -107,6 +136,8 @@ IDE:
 C++ API
 ^^^^^^^
 
+.. _cpp install:
+
 Currently, MLX must be built and installed from source.
 
 Similarly to the python library, to build and install the MLX C++ library start
@@ -185,6 +216,7 @@ should point to the path to the built metal library.
 
       xcrun -sdk macosx --show-sdk-version
 
+
 Binary Size Minimization
 ~~~~~~~~~~~~~~~~~~~~~~~~
 
@@ -213,6 +245,50 @@ be anwywhere from a few hundred millisecond to a few seconds depending on the
 application. Once a kernel is compiled, it will be cached by the system. The
 Metal kernel cache persists across reboots.
 
+Linux
+^^^^^
+
+To build from source on Linux (CPU only), install the BLAS and LAPACK headers.
+For example on Ubuntu, run the following:
+
+.. code-block:: shell
+
+   apt-get update -y
+   apt-get install libblas-dev liblapack-dev liblapacke-dev -y
+
+From here follow the instructions to install either the :ref:`Python <python
+install>` or :ref:`C++ <cpp install>` APIs.
+
+CUDA
+^^^^
+
+To build from source on Linux with CUDA, install the BLAS and LAPACK headers
+and the CUDA toolkit. For example on Ubuntu, run the following:
+
+.. code-block:: shell
+
+   wget https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2204/x86_64/cuda-keyring_1.1-1_all.deb
+   dpkg -i cuda-keyring_1.1-1_all.deb
+   apt-get update -y
+   apt-get -y install cuda-toolkit-12-9
+   apt-get install libblas-dev liblapack-dev liblapacke-dev libcudnn9-dev-cuda-12 -y
+
+
+When building either the Python or C++ APIs make sure to pass the cmake flag
+``MLX_BUILD_CUDA=ON``. For example, to build the Python API run:
+
+.. code-block:: shell
+
+  CMAKE_ARGS="-DMLX_BUILD_CUDA=ON" pip install -e ".[dev]"
+
+To build the C++ package run:
+
+.. code-block:: shell
+
+   mkdir -p build && cd build
+   cmake .. -DMLX_BUILD_CUDA=ON && make -j
+
+
 Troubleshooting
 ^^^^^^^^^^^^^^^
 

docs/src/python/array.rst

@@ -19,6 +19,8 @@ Array
     array.ndim
     array.shape
     array.size
+    array.real
+    array.imag
     array.abs
     array.all
     array.any
@@ -38,6 +40,7 @@ Array
     array.log10
     array.log1p
     array.log2
+    array.logcumsumexp
     array.logsumexp
     array.max
     array.mean

docs/src/python/cuda.rst

@@ -0,0 +1,9 @@
+CUDA
+=====
+
+.. currentmodule:: mlx.core.cuda
+
+.. autosummary::
+  :toctree: _autosummary
+
+  is_available

docs/src/python/fast.rst

@@ -13,3 +13,4 @@ Fast
   rope
   scaled_dot_product_attention
   metal_kernel
+  cuda_kernel

docs/src/python/fft.rst

@@ -20,3 +20,5 @@ FFT
   irfft2
   rfftn
   irfftn
+  fftshift
+  ifftshift

docs/src/python/linalg.rst

@@ -16,9 +16,12 @@ Linear Algebra
     cross
     qr
     svd
+    eigvals
+    eig
     eigvalsh
     eigh
     lu
     lu_factor
+    pinv
     solve
     solve_triangular

docs/src/python/memory_management.rst

@@ -0,0 +1,16 @@
+Memory Management
+=================
+
+.. currentmodule:: mlx.core
+
+.. autosummary::
+  :toctree: _autosummary
+
+  get_active_memory
+  get_peak_memory
+  reset_peak_memory
+  get_cache_memory
+  set_memory_limit
+  set_cache_limit
+  set_wired_limit
+  clear_cache

docs/src/python/metal.rst

@@ -8,13 +8,5 @@ Metal
 
   is_available
   device_info
-  get_active_memory
-  get_peak_memory
-  reset_peak_memory
-  get_cache_memory
-  set_memory_limit
-  set_cache_limit
-  set_wired_limit
-  clear_cache
   start_capture
   stop_capture

docs/src/python/nn.rst

@@ -174,6 +174,7 @@ In detail:
 
    value_and_grad
    quantize
+   average_gradients
 
 .. toctree::
 

docs/src/python/ops.rst

@@ -36,10 +36,12 @@ Operations
    bitwise_or
    bitwise_xor
    block_masked_mm
+   broadcast_arrays
    broadcast_to
    ceil
    clip
    concatenate
+   contiguous
    conj
    conjugate
    convolve
@@ -101,6 +103,7 @@ Operations
    log10
    log1p
    logaddexp
+   logcumsumexp
    logical_not
    logical_and
    logical_or

docs/src/python/optimizers.rst

@@ -51,14 +51,14 @@ the saved state. Here's a simple example:
    optimizer.update(model, grads)
 
    # Save the state
-   state = tree_flatten(optimizer.state)
-   mx.save_safetensors("optimizer.safetensors", dict(state))
+   state = tree_flatten(optimizer.state, destination={})
+   mx.save_safetensors("optimizer.safetensors", state)
 
    # Later on, for example when loading from a checkpoint,
    # recreate the optimizer and load the state
    optimizer = optim.Adam(learning_rate=1e-2)
 
-   state = tree_unflatten(list(mx.load("optimizer.safetensors").items()))
+   state = tree_unflatten(mx.load("optimizer.safetensors"))
    optimizer.state = state
 
 Note, not every optimizer configuation parameter is saved in the state. For

docs/src/python/optimizers/common_optimizers.rst

@@ -18,3 +18,5 @@ Common Optimizers
    AdamW
    Adamax
    Lion
+   MultiOptimizer
+   Muon

docs/src/python/transforms.rst

@@ -9,6 +9,7 @@ Transforms
   :toctree: _autosummary
 
    eval
+   async_eval
    compile
    custom_function
    disable_compile

docs/src/usage/compile.rst

@@ -225,7 +225,7 @@ In some cases returning updated state can be pretty inconvenient. Hence,
   def fun(x, y):
       z = x + y
       state.append(z)
-      return mx.exp(z), state
+      return mx.exp(z)
 
   fun(mx.array(1.0), mx.array(2.0))
   # Prints [array(3, dtype=float32)]

docs/src/usage/distributed.rst

@@ -5,21 +5,27 @@ Distributed Communication
 
 .. currentmodule:: mlx.core.distributed
 
-MLX utilizes `MPI <https://en.wikipedia.org/wiki/Message_Passing_Interface>`_ to
-provide distributed communication operations that allow the computational cost
-of training or inference to be shared across many physical machines. You can
-see a list of the supported operations in the :ref:`API docs<distributed>`.
+MLX supports distributed communication operations that allow the computational cost
+of training or inference to be shared across many physical machines. At the
+moment we support two different communication backends:
+
+* `MPI <https://en.wikipedia.org/wiki/Message_Passing_Interface>`_ a
+  full-featured and mature distributed communications library
+* A **ring** backend of our own that uses native TCP sockets and should be
+  faster for thunderbolt connections.
+
+The list of all currently supported operations and their documentation can be
+seen in the :ref:`API docs<distributed>`.
 
 .. note::
-   A lot of operations may not be supported or not as fast as they should be.
+   Some operations may not be supported or not as fast as they should be.
    We are adding more and tuning the ones we have as we are figuring out the
    best way to do distributed computing on Macs using MLX.
 
 Getting Started
 ---------------
 
-MLX already comes with the ability to "talk" to MPI if it is installed on the
-machine. The minimal distributed program in MLX is as simple as:
+A distributed program in MLX is as simple as:
 
 .. code:: python
 
@@ -30,74 +36,79 @@ machine. The minimal distributed program in MLX is as simple as:
     print(world.rank(), x)
 
 The program above sums the array ``mx.ones(10)`` across all
-distributed processes. If simply run with ``python``, however, only one
-process is launched and no distributed communication takes place.
+distributed processes. However, when this script is run with ``python`` only
+one process is launched and no distributed communication takes place. Namely,
+all operations in ``mx.distributed`` are noops when the distributed group has a
+size of one. This property allows us to avoid code that checks if we are in a
+distributed setting similar to the one below:
 
-To launch the program in distributed mode we need to use ``mpirun`` or
-``mpiexec`` depending on the MPI installation. The simplest possible way is the
-following:
+.. code:: python
+
+    import mlx.core as mx
+
+    x = ...
+    world = mx.distributed.init()
+    # No need for the check we can simply do x = mx.distributed.all_sum(x)
+    if world.size() > 1:
+        x = mx.distributed.all_sum(x)
+
+Running Distributed Programs
+^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+
+MLX provides ``mlx.launch`` a helper script to launch distributed programs.
+Continuing with our initial example we can run it on localhost with 4 processes using
 
 .. code:: shell
 
-    $ mpirun -np 2 python test.py
-    1 array([2, 2, 2, ..., 2, 2, 2], dtype=float32)
-    0 array([2, 2, 2, ..., 2, 2, 2], dtype=float32)
+    $ mlx.launch -n 4 my_script.py
+    3 array([4, 4, 4, ..., 4, 4, 4], dtype=float32)
+    2 array([4, 4, 4, ..., 4, 4, 4], dtype=float32)
+    1 array([4, 4, 4, ..., 4, 4, 4], dtype=float32)
+    0 array([4, 4, 4, ..., 4, 4, 4], dtype=float32)
 
-The above launches two processes on the same (local) machine and we can see
-both standard output streams. The processes send the array of 1s to each other
-and compute the sum which is printed. Launching with ``mpirun -np 4 ...`` would
-print 4 etc.
-
-Installing MPI
----------------
-
-MPI can be installed with Homebrew, using the Anaconda package manager or
-compiled from source. Most of our testing is done using ``openmpi`` installed
-with the Anaconda package manager as follows:
+We can also run it on some remote hosts by providing their IPs (provided that
+the script exists on all hosts and they are reachable by ssh)
 
 .. code:: shell
 
-    $ conda install conda-forge::openmpi
+    $ mlx.launch --hosts ip1,ip2,ip3,ip4 my_script.py
+    3 array([4, 4, 4, ..., 4, 4, 4], dtype=float32)
+    2 array([4, 4, 4, ..., 4, 4, 4], dtype=float32)
+    1 array([4, 4, 4, ..., 4, 4, 4], dtype=float32)
+    0 array([4, 4, 4, ..., 4, 4, 4], dtype=float32)
 
-Installing with Homebrew may require specifying the location of ``libmpi.dyld``
-so that MLX can find it and load it at runtime. This can simply be achieved by
-passing the ``DYLD_LIBRARY_PATH`` environment variable to ``mpirun``.
+Consult the dedicated :doc:`usage guide<launching_distributed>` for more
+information on using ``mlx.launch``.
 
-.. code:: shell
+Selecting Backend
+^^^^^^^^^^^^^^^^^
 
-    $ mpirun -np 2 -x DYLD_LIBRARY_PATH=/opt/homebrew/lib/ python test.py
-
-Setting up Remote Hosts
------------------------
-
-MPI can automatically connect to remote hosts and set up the communication over
-the network if the remote hosts can be accessed via ssh. A good checklist to
-debug connectivity issues is the following:
-
-* ``ssh hostname`` works from all machines to all machines without asking for
-  password or host confirmation
-* ``mpirun`` is accessible on all machines. You can call ``mpirun`` using its
-  full path to force all machines to use a specific path.
-* Ensure that the ``hostname`` used by MPI is the one that you have configured
-  in the ``.ssh/config`` files on all machines.
+You can select the backend you want to use when calling :func:`init` by passing
+one of ``{'any', 'ring', 'mpi'}``. When passing ``any``, MLX will try to
+initialize the ``ring`` backend and if it fails the ``mpi`` backend. If they
+both fail then a singleton group is created.
 
 .. note::
-  For an example hostname ``foo.bar.com`` MPI can use only ``foo`` as
-  the hostname passed to ssh if the current hostname matches ``*.bar.com``.
+   After a distributed backend is successfully initialized :func:`init` will
+   return **the same backend** if called without arguments or with backend set to
+   ``any``.
 
-An easy way to pass the host names to MPI is using a host file. A host file
-looks like the following, where ``host1`` and ``host2`` should be the fully
-qualified domain names or IPs for these hosts.
+The following examples aim to clarify the backend initialization logic in MLX:
 
-.. code::
+.. code:: python
 
-    host1 slots=1
-    host2 slots=1
+    # Case 1: Initialize MPI regardless if it was possible to initialize the ring backend
+    world = mx.distributed.init(backend="mpi")
+    world2 = mx.distributed.init()  # subsequent calls return the MPI backend!
 
-When using MLX, it is very likely that you want to use 1 slot per host, ie one
-process per host.  The hostfile also needs to contain the current
-host if you want to run on the local host. Passing the host file to
-``mpirun`` is simply done using the ``--hostfile`` command line argument.
+    # Case 2: Initialize any backend
+    world = mx.distributed.init(backend="any")  # equivalent to no arguments
+    world2 = mx.distributed.init()  # same as above
+
+    # Case 3: Initialize both backends at the same time
+    world_mpi = mx.distributed.init(backend="mpi")
+    world_ring = mx.distributed.init(backend="ring")
+    world_any = mx.distributed.init()  # same as MPI because it was initialized first!
 
 Training Example
 ----------------
@@ -155,13 +166,179 @@ everything else remaining the same.
         optimizer.update(model, grads)
         return loss
 
-Tuning All Reduce
------------------
+Utilizing ``nn.average_gradients``
+^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
 
-We are working on improving the performance of all reduce on MLX but for now
-the two main things one can do to extract the most out of distributed training with MLX are:
+Although the code example above works correctly; it performs one communication
+per gradient. It is significantly more efficient to aggregate several gradients
+together and perform fewer communication steps.
 
-1. Perform a few large reductions instead of many small ones to improve
-   bandwidth and latency
-2. Pass ``--mca btl_tcp_links 4`` to ``mpirun`` to configure it to use 4 tcp
-   connections between each host to improve bandwidth
+This is the purpose of :func:`mlx.nn.average_gradients`. The final code looks
+almost identical to the example above:
+
+.. code:: python
+
+    model = ...
+    optimizer = ...
+    dataset = ...
+
+    def step(model, x, y):
+        loss, grads = loss_grad_fn(model, x, y)
+        grads = mlx.nn.average_gradients(grads) # <---- This line was added
+        optimizer.update(model, grads)
+        return loss
+
+    for x, y in dataset:
+        loss = step(model, x, y)
+        mx.eval(loss, model.parameters())
+
+
+Getting Started with MPI
+------------------------
+
+MLX already comes with the ability to "talk" to MPI if it is installed on the
+machine. Launching distributed MLX programs that use MPI can be done with
+``mpirun`` as expected. However, in the following examples we will be using
+``mlx.launch --backend mpi`` which takes care of some nuisances such as setting
+absolute paths for the ``mpirun`` executable and the ``libmpi.dyld`` shared
+library.
+
+The simplest possible usage is the following which, assuming the minimal
+example in the beginning of this page, should result in:
+
+.. code:: shell
+
+    $ mlx.launch --backend mpi -n 2 test.py
+    1 array([2, 2, 2, ..., 2, 2, 2], dtype=float32)
+    0 array([2, 2, 2, ..., 2, 2, 2], dtype=float32)
+
+The above launches two processes on the same (local) machine and we can see
+both standard output streams. The processes send the array of 1s to each other
+and compute the sum which is printed. Launching with ``mlx.launch -n 4 ...`` would
+print 4 etc.
+
+Installing MPI
+^^^^^^^^^^^^^^
+
+MPI can be installed with Homebrew, using the Anaconda package manager or
+compiled from source. Most of our testing is done using ``openmpi`` installed
+with the Anaconda package manager as follows:
+
+.. code:: shell
+
+    $ conda install conda-forge::openmpi
+
+Installing with Homebrew may require specifying the location of ``libmpi.dyld``
+so that MLX can find it and load it at runtime. This can simply be achieved by
+passing the ``DYLD_LIBRARY_PATH`` environment variable to ``mpirun`` and it is
+done automatically by ``mlx.launch``.
+
+.. code:: shell
+
+    $ mpirun -np 2 -x DYLD_LIBRARY_PATH=/opt/homebrew/lib/ python test.py
+    $ # or simply
+    $ mlx.launch -n 2 test.py
+
+Setting up Remote Hosts
+^^^^^^^^^^^^^^^^^^^^^^^
+
+MPI can automatically connect to remote hosts and set up the communication over
+the network if the remote hosts can be accessed via ssh. A good checklist to
+debug connectivity issues is the following:
+
+* ``ssh hostname`` works from all machines to all machines without asking for
+  password or host confirmation
+* ``mpirun`` is accessible on all machines.
+* Ensure that the ``hostname`` used by MPI is the one that you have configured
+  in the ``.ssh/config`` files on all machines.
+
+Tuning MPI All Reduce
+^^^^^^^^^^^^^^^^^^^^^
+
+.. note::
+
+    For faster all reduce consider using the ring backend either with Thunderbolt
+    connections or over Ethernet.
+
+Configure MPI to use N tcp connections between each host to improve bandwidth
+by passing ``--mca btl_tcp_links N``.
+
+Force MPI to use the most performant network interface by setting ``--mca
+btl_tcp_if_include <iface>`` where ``<iface>`` should be the interface you want
+to use.
+
+Getting Started with Ring
+-------------------------
+
+The ring backend does not depend on any third party library so it is always
+available. It uses TCP sockets so the nodes need to be reachable via a network.
+As the name suggests the nodes are connected in a ring which means that rank 1
+can only communicate with rank 0 and rank 2, rank 2 only with rank 1 and rank 3
+and so on and so forth. As a result :func:`send` and :func:`recv` with
+arbitrary sender and receiver is not supported in the ring backend.
+
+Defining a Ring
+^^^^^^^^^^^^^^^
+
+The easiest way to define and use a ring is via a JSON hostfile and the
+``mlx.launch`` :doc:`helper script <launching_distributed>`. For each node one
+defines a hostname to ssh into to run commands on this node and one or more IPs
+that this node will listen to for connections.
+
+For example the hostfile below defines a 4 node ring. ``hostname1`` will be
+rank 0, ``hostname2`` rank 1 etc.
+
+.. code:: json
+
+    [
+        {"ssh": "hostname1", "ips": ["123.123.123.1"]},
+        {"ssh": "hostname2", "ips": ["123.123.123.2"]},
+        {"ssh": "hostname3", "ips": ["123.123.123.3"]},
+        {"ssh": "hostname4", "ips": ["123.123.123.4"]}
+    ]
+
+Running ``mlx.launch --hostfile ring-4.json my_script.py`` will ssh into each
+node, run the script which will listen for connections in each of the provided
+IPs. Specifically, ``hostname1`` will connect to ``123.123.123.2`` and accept a
+connection from ``123.123.123.4`` and so on and so forth.
+
+Thunderbolt Ring
+^^^^^^^^^^^^^^^^
+
+Although the ring backend can have benefits over MPI even for Ethernet, its
+main purpose is to use Thunderbolt rings for higher bandwidth communication.
+Setting up such thunderbolt rings can be done manually, but is a relatively
+tedious process. To simplify this, we provide the utility ``mlx.distributed_config``.
+
+To use ``mlx.distributed_config`` your computers need to be accessible by ssh via
+Ethernet or Wi-Fi. Subsequently, connect them via thunderbolt cables and then call the
+utility as follows:
+
+.. code:: shell
+
+   mlx.distributed_config --verbose --hosts host1,host2,host3,host4
+
+By default the script will attempt to discover the thunderbolt ring and provide
+you with the commands to configure each node as well as the ``hostfile.json``
+to use with ``mlx.launch``. If password-less ``sudo`` is available on the nodes
+then ``--auto-setup`` can be used to configure them automatically.
+
+To validate your connection without configuring anything
+``mlx.distributed_config`` can also plot the ring using DOT format.
+
+.. code:: shell
+
+   mlx.distributed_config --verbose --hosts host1,host2,host3,host4 --dot >ring.dot
+   dot -Tpng ring.dot >ring.png
+   open ring.png
+
+If you want to go through the process manually, the steps are as follows:
+
+* Disable the thunderbolt bridge interface
+* For the cable connecting rank ``i`` to rank ``i + 1`` find the interfaces
+  corresponding to that cable in nodes ``i`` and ``i + 1``.
+* Set up a unique subnetwork connecting the two nodes for the corresponding
+  interfaces. For instance if the cable corresponds to ``en2`` on node ``i``
+  and ``en2`` also on node ``i + 1`` then we may assign IPs ``192.168.0.1`` and
+  ``192.168.0.2`` respectively to the two nodes. For more details you can see
+  the commands prepared by the utility script.

docs/src/usage/export.rst

@@ -7,17 +7,17 @@ Exporting Functions
 
 MLX has an API to export and import functions to and from a file. This lets you
 run computations written in one MLX front-end (e.g. Python) in another MLX
-front-end (e.g. C++). 
+front-end (e.g. C++).
 
 This guide walks through the basics of the MLX export API with some examples.
 To see the full list of functions check-out the :ref:`API documentation
 <export>`.
 
-Basics of Exporting 
+Basics of Exporting
 -------------------
 
 Let's start with a simple example:
- 
+
 .. code-block:: python
 
   def fun(x, y):
@@ -67,7 +67,7 @@ specified as variable positional arguments or as a tuple of arrays:
 
   x = mx.array(1.0)
   y = mx.array(1.0)
-   
+
   # Both arguments to fun are positional
   mx.export_function("add.mlxfn", fun, x, y)
 
@@ -133,7 +133,7 @@ parameters are also saved to the ``model.mlxfn`` file.
    For enclosed arrays inside an exported function, be extra careful to ensure
    they are evaluated. The computation graph that gets exported will include
    the computation that produces enclosed inputs.
-  
+
    If the above example was missing ``mx.eval(model.parameters()``, the
    exported function would include the random initialization of the
    :obj:`mlx.nn.Module` parameters.
@@ -150,8 +150,8 @@ parameters, pass them as inputs to the ``call`` wrapper:
      # Set the model's parameters to the input parameters
      model.update(tree_unflatten(list(params.items())))
      return model(x)
- 
-   params = dict(tree_flatten(model.parameters()))
+
+   params = tree_flatten(model.parameters(), destination={})
    mx.export_function("model.mlxfn", call, (mx.zeros(4),), params)
 
 
@@ -169,8 +169,8 @@ to export a function which can be used for inputs with variable shapes:
 
   # Ok
   out, = imported_abs(mx.array(-1.0))
-  
-  # Also ok 
+
+  # Also ok
   out, = imported_abs(mx.array([-1.0, -2.0]))
 
 With ``shapeless=False`` (which is the default), the second call to
@@ -197,7 +197,7 @@ a single file by creating an exporting context manager with :func:`exporter`:
   def fun(x, y=None):
       constant = mx.array(3.0)
       if y is not None:
-        x += y 
+        x += y
       return x + constant
 
   with mx.exporter("fun.mlxfn", fun) as exporter:
@@ -215,7 +215,7 @@ a single file by creating an exporting context manager with :func:`exporter`:
   print(out)
 
 In the above example the function constant data, (i.e. ``constant``), is only
-saved once. 
+saved once.
 
 Transformations with Imported Functions
 ---------------------------------------
@@ -238,7 +238,7 @@ on imported functions just like regular Python functions:
   # Prints: array(1, dtype=float32)
   print(dfdx(x))
 
-  # Compile the imported function 
+  # Compile the imported function
   mx.compile(imported_fun)
   # Prints: array(0, dtype=float32)
   print(compiled_fun(x)[0])
@@ -275,7 +275,7 @@ Import and run the function in C++ with only a few lines of code:
   // Prints: array(2, dtype=float32)
   std::cout << outputs[0] << std::endl;
 
-Imported functions can be transformed in C++ just like in Python. Use 
+Imported functions can be transformed in C++ just like in Python. Use
 ``std::vector<mx::array>`` for positional arguments and ``std::map<std::string,
 mx::array>`` for keyword arguments when calling imported functions in C++.
 

docs/src/usage/indexing.rst

@@ -107,6 +107,16 @@ same array:
   >>> a
   array([1, 2, 0], dtype=int32)
 
+
+Note, unlike NumPy, updates to the same location are nondeterministic:
+
+.. code-block:: shell
+
+  >>> a = mx.array([1, 2, 3])
+  >>> a[[0, 0]] = mx.array([4, 5])
+
+The first element of ``a`` could be ``4`` or ``5``.
+
 Transformations of functions which use in-place updates are allowed and work as
 expected. For example:
 

docs/src/usage/launching_distributed.rst

@@ -0,0 +1,105 @@
+:orphan:
+
+.. _usage_launch_distributed:
+
+Launching Distributed Programs
+==============================
+
+.. currentmodule:: mlx.core.distributed
+
+Installing the MLX python package provides a helper script ``mlx.launch`` that
+can be used to run python scripts distributed on several nodes. It allows
+launching using either the MPI backend or the ring backend. See the
+:doc:`distributed docs <distributed>` for the different backends.
+
+Usage
+-----
+
+The minimal usage example of ``mlx.launch`` is simply
+
+.. code:: shell
+
+    mlx.launch --hosts ip1,ip2 my_script.py
+
+or for testing on localhost
+
+.. code:: shell
+
+    mlx.launch -n 2 my_script.py
+
+The ``mlx.launch`` command connects to the provided host and launches the input
+script on each host. It monitors each of the launched processes and terminates
+the rest if one of them fails unexpectedly or if ``mlx.launch`` is terminated.
+It also takes care of forwarding the output of each remote process to stdout
+and stderr respectively.
+
+Providing Hosts
+^^^^^^^^^^^^^^^^
+
+Hosts can be provided as command line arguments, like above, but the way that
+allows to fully define a list of hosts is via a JSON hostfile. The hostfile has
+a very simple schema. It is simply a list of objects that define each host via
+a hostname to ssh to and a list of IPs to utilize for the communication.
+
+.. code:: json
+
+    [
+        {"ssh": "hostname1", "ips": ["123.123.1.1", "123.123.2.1"]},
+        {"ssh": "hostname2", "ips": ["123.123.1.2", "123.123.2.2"]}
+    ]
+
+You can use ``mlx.distributed_config --over ethernet`` to create a hostfile
+with IPs corresponding to the ``en0`` interface.
+
+Setting up Remote Hosts
+^^^^^^^^^^^^^^^^^^^^^^^^
+
+In order to be able to launch the script on each host we need to be able to
+connect via ssh. Moreover the input script and python binary need to be on each
+host and on the same path. A good checklist to debug errors is the following:
+
+* ``ssh hostname`` works without asking for password or host confirmation
+* the python binary is available on all hosts at the same path. You can use
+  ``mlx.launch --print-python`` to see what that path is.
+* the script you want to run is available on all hosts at the same path
+
+.. _mpi_specifics:
+
+MPI Specifics
+-------------
+
+One can use MPI by passing ``--backend mpi`` to ``mlx.launch``. In that case,
+``mlx.launch`` is a thin wrapper over ``mpirun``. Moreover,
+
+* The IPs in the hostfile are ignored
+* The ssh connectivity requirement is stronger as every node needs to be able
+  to connect to every other node
+* ``mpirun`` needs to be available on every node at the same path
+
+Finally, one can pass arguments to ``mpirun`` using ``--mpi-arg``. For instance
+to choose a specific interface for the byte-transfer-layer of MPI we can call
+``mlx.launch`` as follows:
+
+.. code:: shell
+
+    mlx.launch --backend mpi --mpi-arg '--mca btl_tcp_if_include en0' --hostfile hosts.json my_script.py
+
+
+.. _ring_specifics:
+
+Ring Specifics
+--------------
+
+The ring backend, which is also the default backend, can be explicitly selected
+with the argument ``--backend ring``. The ring backend has some specific
+requirements and arguments that are different to MPI:
+
+* The argument ``--hosts`` only accepts IPs and not hostnames. If we need to
+  ssh to a hostname that does not correspond to the IP we want to bind to we
+  have to provide a hostfile.
+* ``--starting-port`` defines the port to bind to on the remote hosts.
+  Specifically rank 0 for the first IP will use this port and each subsequent
+  IP or rank will add 1 to this port.
+* ``--connections-per-ip`` allows us to increase the number of connections
+  between neighboring nodes. This corresponds to ``--mca btl_tcp_links 2`` for
+  ``mpirun``.

examples/extensions/CMakeLists.txt

@@ -10,7 +10,6 @@ set(CMAKE_POSITION_INDEPENDENT_CODE ON)
 option(BUILD_SHARED_LIBS "Build extensions as a shared library" ON)
 
 # ----------------------------- Dependencies -----------------------------
-find_package(MLX CONFIG REQUIRED)
 find_package(
   Python 3.8
   COMPONENTS Interpreter Development.Module
@@ -21,6 +20,12 @@ execute_process(
   OUTPUT_VARIABLE nanobind_ROOT)
 find_package(nanobind CONFIG REQUIRED)
 
+execute_process(
+  COMMAND "${Python_EXECUTABLE}" -m mlx --cmake-dir
+  OUTPUT_STRIP_TRAILING_WHITESPACE
+  OUTPUT_VARIABLE MLX_ROOT)
+find_package(MLX CONFIG REQUIRED)
+
 # ----------------------------- Extensions -----------------------------
 
 # Add library

examples/extensions/axpby/axpby.cpp

@@ -1,20 +1,15 @@
-// Copyright © 2023-2024 Apple Inc.
+// Copyright © 2023-2025 Apple Inc.
 
-#include <cassert>
+#include <dlfcn.h>
 #include <iostream>
 #include <sstream>
 
-#include "mlx/backend/common/copy.h"
 #include "mlx/backend/common/utils.h"
-#include "mlx/backend/cpu/copy.h"
+#include "mlx/backend/cpu/encoder.h"
 #include "mlx/utils.h"
 
 #include "axpby/axpby.h"
 
-#ifdef ACCELERATE_NEW_LAPACK
-#include <vecLib/cblas_new.h>
-#endif
-
 #ifdef _METAL_
 #include "mlx/backend/metal/device.h"
 #include "mlx/backend/metal/utils.h"
@@ -22,6 +17,19 @@
 
 namespace my_ext {
 
+// A helper function to find the location of the current binary on disk.
+// The Metal library ("mlx_ext.mtllib"), should be in the same directory.
+std::string current_binary_dir() {
+  static std::string binary_dir = []() {
+    Dl_info info;
+    if (!dladdr(reinterpret_cast<void*>(&current_binary_dir), &info)) {
+      throw std::runtime_error("Unable to get current binary dir.");
+    }
+    return std::filesystem::path(info.dli_fname).parent_path().string();
+  }();
+  return binary_dir;
+}
+
 ///////////////////////////////////////////////////////////////////////////////
 // Operation Implementation
 ///////////////////////////////////////////////////////////////////////////////
@@ -76,136 +84,65 @@ void axpby_impl(
     const mx::array& y,
     mx::array& out,
     float alpha_,
-    float beta_) {
-  // We only allocate memory when we are ready to fill the output
-  // malloc_or_wait synchronously allocates available memory
-  // There may be a wait executed here if the allocation is requested
-  // under memory-pressured conditions
-  out.set_data(mx::allocator::malloc_or_wait(out.nbytes()));
+    float beta_,
+    mx::Stream stream) {
+  out.set_data(mx::allocator::malloc(out.nbytes()));
 
-  // Collect input and output data pointers
-  const T* x_ptr = x.data<T>();
-  const T* y_ptr = y.data<T>();
-  T* out_ptr = out.data<T>();
+  // Get the CPU command encoder and register input and output arrays
+  auto& encoder = mx::cpu::get_command_encoder(stream);
+  encoder.set_input_array(x);
+  encoder.set_input_array(y);
+  encoder.set_output_array(out);
 
-  // Cast alpha and beta to the relevant types
-  T alpha = static_cast<T>(alpha_);
-  T beta = static_cast<T>(beta_);
+  // Launch the CPU kernel
+  encoder.dispatch([x_ptr = x.data<T>(),
+                    y_ptr = y.data<T>(),
+                    out_ptr = out.data<T>(),
+                    size = out.size(),
+                    shape = out.shape(),
+                    x_strides = x.strides(),
+                    y_strides = y.strides(),
+                    alpha_,
+                    beta_]() {
+    // Cast alpha and beta to the relevant types
+    T alpha = static_cast<T>(alpha_);
+    T beta = static_cast<T>(beta_);
 
-  // Do the element-wise operation for each output
-  for (size_t out_idx = 0; out_idx < out.size(); out_idx++) {
-    // Map linear indices to offsets in x and y
-    auto x_offset = mx::elem_to_loc(out_idx, x.shape(), x.strides());
-    auto y_offset = mx::elem_to_loc(out_idx, y.shape(), y.strides());
+    // Do the element-wise operation for each output
+    for (size_t out_idx = 0; out_idx < size; out_idx++) {
+      // Map linear indices to offsets in x and y
+      auto x_offset = mx::elem_to_loc(out_idx, shape, x_strides);
+      auto y_offset = mx::elem_to_loc(out_idx, shape, y_strides);
 
-    // We allocate the output to be contiguous and regularly strided
-    // (defaults to row major) and hence it doesn't need additional mapping
-    out_ptr[out_idx] = alpha * x_ptr[x_offset] + beta * y_ptr[y_offset];
-  }
+      // We allocate the output to be contiguous and regularly strided
+      // (defaults to row major) and hence it doesn't need additional mapping
+      out_ptr[out_idx] = alpha * x_ptr[x_offset] + beta * y_ptr[y_offset];
+    }
+  });
 }
 
-/** Fall back implementation for evaluation on CPU */
-void Axpby::eval(
+void Axpby::eval_cpu(
     const std::vector<mx::array>& inputs,
     std::vector<mx::array>& outputs) {
-  // Check the inputs (registered in the op while constructing the out array)
-  assert(inputs.size() == 2);
   auto& x = inputs[0];
   auto& y = inputs[1];
   auto& out = outputs[0];
 
   // Dispatch to the correct dtype
   if (out.dtype() == mx::float32) {
-    return axpby_impl<float>(x, y, out, alpha_, beta_);
+    return axpby_impl<float>(x, y, out, alpha_, beta_, stream());
   } else if (out.dtype() == mx::float16) {
-    return axpby_impl<mx::float16_t>(x, y, out, alpha_, beta_);
+    return axpby_impl<mx::float16_t>(x, y, out, alpha_, beta_, stream());
   } else if (out.dtype() == mx::bfloat16) {
-    return axpby_impl<mx::bfloat16_t>(x, y, out, alpha_, beta_);
+    return axpby_impl<mx::bfloat16_t>(x, y, out, alpha_, beta_, stream());
   } else if (out.dtype() == mx::complex64) {
-    return axpby_impl<mx::complex64_t>(x, y, out, alpha_, beta_);
+    return axpby_impl<mx::complex64_t>(x, y, out, alpha_, beta_, stream());
   } else {
     throw std::runtime_error(
         "Axpby is only supported for floating point types.");
   }
 }
 
-///////////////////////////////////////////////////////////////////////////////
-// Primitive Accelerate Backend Implementation
-///////////////////////////////////////////////////////////////////////////////
-
-#ifdef ACCELERATE_NEW_LAPACK
-
-template <typename T>
-void axpby_impl_accelerate(
-    const mx::array& x,
-    const mx::array& y,
-    mx::array& out,
-    float alpha_,
-    float beta_) {
-  // Accelerate library provides catlas_saxpby which does
-  // Y = (alpha * X) + (beta * Y) in place
-  // To use it, we first copy the data in y over to the output array
-
-  // This specialization requires both x and y be contiguous in the same mode
-  // i.e: corresponding linear indices in both point to corresponding elements
-  // The data in the output array is allocated to match the strides in y
-  // such that x, y, and out are contiguous in the same mode and
-  // no transposition is needed
-  out.set_data(mx::allocator::malloc_or_wait(out.nbytes()));
-
-  // We then copy over the elements using the contiguous vector specialization
-  copy_inplace(y, out, mx::CopyType::Vector);
-
-  // Get x and y pointers for catlas_saxpby
-  const T* x_ptr = x.data<T>();
-  T* y_ptr = out.data<T>();
-
-  T alpha = static_cast<T>(alpha_);
-  T beta = static_cast<T>(beta_);
-
-  // Call the inplace accelerate operator
-  catlas_saxpby(
-      /* N = */ out.size(),
-      /* ALPHA = */ alpha,
-      /* X = */ x_ptr,
-      /* INCX = */ 1,
-      /* BETA = */ beta,
-      /* Y = */ y_ptr,
-      /* INCY = */ 1);
-}
-
-/** Evaluate primitive on CPU using accelerate specializations */
-void Axpby::eval_cpu(
-    const std::vector<mx::array>& inputs,
-    std::vector<mx::array>& outputs) {
-  assert(inputs.size() == 2);
-  auto& x = inputs[0];
-  auto& y = inputs[1];
-  auto& out = outputs[0];
-
-  // Accelerate specialization for contiguous single precision float arrays
-  if (out.dtype() == mx::float32 &&
-      ((x.flags().row_contiguous && y.flags().row_contiguous) ||
-       (x.flags().col_contiguous && y.flags().col_contiguous))) {
-    axpby_impl_accelerate<float>(x, y, out, alpha_, beta_);
-    return;
-  }
-
-  // Fall back to common backend if specializations are not available
-  eval(inputs, outputs);
-}
-
-#else // Accelerate not available
-
-/** Evaluate primitive on CPU falling back to common backend */
-void Axpby::eval_cpu(
-    const std::vector<mx::array>& inputs,
-    std::vector<mx::array>& outputs) {
-  eval(inputs, outputs);
-}
-
-#endif
-
 ///////////////////////////////////////////////////////////////////////////////
 // Primitive Metal Backend Implementation
 ///////////////////////////////////////////////////////////////////////////////
@@ -217,7 +154,6 @@ void Axpby::eval_gpu(
     const std::vector<mx::array>& inputs,
     std::vector<mx::array>& outputs) {
   // Prepare inputs
-  assert(inputs.size() == 2);
   auto& x = inputs[0];
   auto& y = inputs[1];
   auto& out = outputs[0];
@@ -236,25 +172,24 @@ void Axpby::eval_gpu(
   // Allocate output memory with strides based on specialization
   if (contiguous_kernel) {
     out.set_data(
-        mx::allocator::malloc_or_wait(x.data_size() * out.itemsize()),
+        mx::allocator::malloc(x.data_size() * out.itemsize()),
         x.data_size(),
         x.strides(),
         x.flags());
   } else {
-    out.set_data(mx::allocator::malloc_or_wait(out.nbytes()));
+    out.set_data(mx::allocator::malloc(out.nbytes()));
   }
 
   // Resolve name of kernel (corresponds to axpby.metal)
-  std::ostringstream kname;
-  kname << "axpby_";
-  kname << (contiguous_kernel ? "contiguous_" : "general_");
-  kname << type_to_name(out);
+  std::string kname = "axpby_";
+  kname += (contiguous_kernel ? "contiguous_" : "general_");
+  kname += type_to_name(out);
 
-  // Make sure the metal library is available
-  d.register_library("mlx_ext");
+  // Load the metal library
+  auto lib = d.get_library("mlx_ext", current_binary_dir());
 
   // Make a kernel from this metal library
-  auto kernel = d.get_kernel(kname.str(), "mlx_ext");
+  auto kernel = d.get_kernel(kname, lib);
 
   // Prepare to encode kernel
   auto& compute_encoder = d.get_command_encoder(s.index);

examples/extensions/axpby/axpby.h

@@ -1,4 +1,4 @@
-// Copyright © 2023 Apple Inc.
+// Copyright © 2023-2025 Apple Inc.
 
 #pragma once
 
@@ -74,9 +74,9 @@ class Axpby : public mx::Primitive {
       const std::vector<mx::array>& inputs,
       const std::vector<int>& axes) override;
 
-  /** Print the primitive. */
-  void print(std::ostream& os) override {
-    os << "Axpby";
+  /** The name of primitive. */
+  const char* name() const override {
+    return "Axpby";
   }
 
   /** Equivalence check **/
@@ -85,11 +85,6 @@ class Axpby : public mx::Primitive {
  private:
   float alpha_;
   float beta_;
-
-  /** Fall back implementation for evaluation on CPU */
-  void eval(
-      const std::vector<mx::array>& inputs,
-      std::vector<mx::array>& outputs);
 };
 
 } // namespace my_ext

examples/extensions/axpby/axpby.metal

@@ -1,4 +1,4 @@
-// Copyright © 2023 Apple Inc.
+// Copyright © 2023-2025 Apple Inc.
 
 #include <metal_stdlib>
 

examples/extensions/requirements.txt

@@ -1,4 +1,4 @@
 setuptools>=42
 cmake>=3.25
 mlx>=0.21.0
-nanobind==2.2.0
+nanobind==2.4.0

examples/extensions/test.py

@@ -3,8 +3,10 @@ from mlx_sample_extensions import axpby
 
 a = mx.ones((3, 4))
 b = mx.ones((3, 4))
-c = axpby(a, b, 4.0, 2.0, stream=mx.cpu)
+c_cpu = axpby(a, b, 4.0, 2.0, stream=mx.cpu)
+c_gpu = axpby(a, b, 4.0, 2.0, stream=mx.gpu)
 
-print(f"c shape: {c.shape}")
-print(f"c dtype: {c.dtype}")
-print(f"c correct: {mx.all(c == 6.0).item()}")
+print(f"c shape: {c_cpu.shape}")
+print(f"c dtype: {c_cpu.dtype}")
+print(f"c_cpu correct: {mx.all(c_cpu == 6.0).item()}")
+print(f"c_gpu correct: {mx.all(c_gpu == 6.0).item()}")

mlx/CMakeLists.txt

@@ -5,6 +5,7 @@ target_sources(
           ${CMAKE_CURRENT_SOURCE_DIR}/compile.cpp
           ${CMAKE_CURRENT_SOURCE_DIR}/device.cpp
           ${CMAKE_CURRENT_SOURCE_DIR}/dtype.cpp
+          ${CMAKE_CURRENT_SOURCE_DIR}/dtype_utils.cpp
           ${CMAKE_CURRENT_SOURCE_DIR}/export.cpp
           ${CMAKE_CURRENT_SOURCE_DIR}/einsum.cpp
           ${CMAKE_CURRENT_SOURCE_DIR}/fast.cpp
@@ -19,6 +20,11 @@ target_sources(
           ${CMAKE_CURRENT_SOURCE_DIR}/linalg.cpp
           ${CMAKE_CURRENT_SOURCE_DIR}/backend/metal/metal.h)
 
+# Define MLX_VERSION only in the version.cpp file.
+add_library(mlx_version OBJECT ${CMAKE_CURRENT_SOURCE_DIR}/version.cpp)
+target_compile_definitions(mlx_version PRIVATE MLX_VERSION="${MLX_VERSION}")
+target_link_libraries(mlx PRIVATE $<BUILD_INTERFACE:mlx_version>)
+
 if(MSVC)
   # Disable some MSVC warnings to speed up compilation.
   target_compile_options(mlx PUBLIC /wd4068 /wd4244 /wd4267 /wd4804)
@@ -43,5 +49,19 @@ add_subdirectory(${CMAKE_CURRENT_SOURCE_DIR}/io)
 if(MLX_BUILD_METAL)
   add_subdirectory(${CMAKE_CURRENT_SOURCE_DIR}/backend/metal)
 else()
-  add_subdirectory(${CMAKE_CURRENT_SOURCE_DIR}/backend/no_metal)
+  target_sources(mlx
+                 PRIVATE ${CMAKE_CURRENT_SOURCE_DIR}/backend/metal/no_metal.cpp)
+endif()
+
+if(MLX_BUILD_CUDA)
+  add_subdirectory(${CMAKE_CURRENT_SOURCE_DIR}/backend/cuda)
+else()
+  target_sources(mlx
+                 PRIVATE ${CMAKE_CURRENT_SOURCE_DIR}/backend/cuda/no_cuda.cpp)
+endif()
+
+if(MLX_BUILD_METAL OR MLX_BUILD_CUDA)
+  add_subdirectory(${CMAKE_CURRENT_SOURCE_DIR}/backend/gpu)
+else()
+  add_subdirectory(${CMAKE_CURRENT_SOURCE_DIR}/backend/no_gpu)
 endif()

mlx/allocator.cpp

@@ -4,12 +4,11 @@
 #include <sstream>
 
 #include "mlx/allocator.h"
-#include "mlx/scheduler.h"
 
 namespace mlx::core::allocator {
 
 Buffer malloc(size_t size) {
-  auto buffer = allocator().malloc(size, /* allow_swap */ true);
+  auto buffer = allocator().malloc(size);
   if (size && !buffer.ptr()) {
     std::ostringstream msg;
     msg << "[malloc] Unable to allocate " << size << " bytes.";
@@ -22,45 +21,4 @@ void free(Buffer buffer) {
   allocator().free(buffer);
 }
 
-Buffer CommonAllocator::malloc(size_t size, bool) {
-  void* ptr = std::malloc(size + sizeof(size_t));
-  if (ptr != nullptr) {
-    *static_cast<size_t*>(ptr) = size;
-  }
-  return Buffer{ptr};
-}
-
-void CommonAllocator::free(Buffer buffer) {
-  std::free(buffer.ptr());
-}
-
-size_t CommonAllocator::size(Buffer buffer) const {
-  if (buffer.ptr() == nullptr) {
-    return 0;
-  }
-  return *static_cast<size_t*>(buffer.ptr());
-}
-
-Buffer malloc_or_wait(size_t size) {
-  auto buffer = allocator().malloc(size);
-
-  while (size && !buffer.ptr() && scheduler::n_active_tasks() > 0) {
-    scheduler::wait_for_one();
-    buffer = allocator().malloc(size);
-  }
-
-  // Try swapping if needed
-  if (size && !buffer.ptr()) {
-    buffer = allocator().malloc(size, /* allow_swap = */ true);
-  }
-
-  if (size && !buffer.ptr()) {
-    std::ostringstream msg;
-    msg << "[malloc_or_wait] Unable to allocate " << size << " bytes.";
-    throw std::runtime_error(msg.str());
-  }
-
-  return buffer;
-}
-
 } // namespace mlx::core::allocator

mlx/allocator.h

@@ -32,14 +32,10 @@ Buffer malloc(size_t size);
 
 void free(Buffer buffer);
 
-// Wait for running tasks to finish and free up memory
-// if allocation fails
-Buffer malloc_or_wait(size_t size);
-
 class Allocator {
   /** Abstract base class for a memory allocator. */
  public:
-  virtual Buffer malloc(size_t size, bool allow_swap = false) = 0;
+  virtual Buffer malloc(size_t size) = 0;
   virtual void free(Buffer buffer) = 0;
   virtual size_t size(Buffer buffer) const = 0;
 
@@ -53,16 +49,4 @@ class Allocator {
 
 Allocator& allocator();
 
-class CommonAllocator : public Allocator {
-  /** A general CPU allocator. */
- public:
-  virtual Buffer malloc(size_t size, bool allow_swap = false) override;
-  virtual void free(Buffer buffer) override;
-  virtual size_t size(Buffer buffer) const override;
-
- private:
-  CommonAllocator() = default;
-  friend Allocator& allocator();
-};
-
 } // namespace mlx::core::allocator

mlx/array.cpp

@@ -56,6 +56,18 @@ std::vector<array> array::make_arrays(
   return outputs;
 }
 
+array array::unsafe_weak_copy(const array& other) {
+  auto cpy = array(other.shape(), other.dtype(), nullptr, {});
+  cpy.set_data(
+      other.buffer(),
+      other.data_size(),
+      other.strides(),
+      other.flags(),
+      [](auto) {});
+  cpy.array_desc_->data_ptr = other.array_desc_->data_ptr;
+  return cpy;
+}
+
 array::array(std::initializer_list<float> data)
     : array_desc_(std::make_shared<ArrayDesc>(
           Shape{static_cast<ShapeElem>(data.size())},
@@ -76,35 +88,27 @@ array::array(allocator::Buffer data, Shape shape, Dtype dtype, Deleter deleter)
   set_data(data, deleter);
 }
 
-array::array(
-    allocator::Buffer data,
-    Shape shape,
-    Dtype dtype,
-    Strides strides,
-    size_t data_size,
-    Flags flags,
-    Deleter deleter)
-    : array_desc_(std::make_shared<ArrayDesc>(std::move(shape), dtype)) {
-  set_data(data, data_size, std::move(strides), flags, deleter);
-}
-
 void array::detach() {
+  array_desc_->primitive = nullptr;
+  for (auto& s : array_desc_->siblings) {
+    s.array_desc_->primitive = nullptr;
+  }
   for (auto& s : array_desc_->siblings) {
     s.array_desc_->inputs.clear();
     s.array_desc_->siblings.clear();
     s.array_desc_->position = 0;
-    s.array_desc_->primitive = nullptr;
   }
   array_desc_->inputs.clear();
   array_desc_->siblings.clear();
   array_desc_->position = 0;
-  array_desc_->primitive = nullptr;
 }
 
 bool array::is_available() const {
   if (status() == Status::available) {
     return true;
-  } else if (status() == Status::evaluated && event().is_signaled()) {
+  } else if (
+      status() == Status::evaluated &&
+      (!event().valid() || event().is_signaled())) {
     set_status(Status::available);
     return true;
   }
@@ -113,7 +117,10 @@ bool array::is_available() const {
 
 void array::wait() {
   if (!is_available()) {
-    event().wait();
+    if (event().valid()) {
+      event().wait();
+      detach_event();
+    }
     set_status(Status::available);
   }
 }
@@ -174,34 +181,13 @@ void array::copy_shared_buffer(const array& other) {
   copy_shared_buffer(other, other.strides(), other.flags(), other.data_size());
 }
 
-void array::move_shared_buffer(
-    array other,
-    const Strides& strides,
-    Flags flags,
-    size_t data_size,
-    size_t offset /* = 0 */) {
-  array_desc_->data = std::move(other.array_desc_->data);
-  array_desc_->strides = strides;
-  array_desc_->flags = flags;
-  array_desc_->data_size = data_size;
-  auto char_offset = sizeof(char) * itemsize() * offset;
-  auto data_ptr = other.array_desc_->data_ptr;
-  other.array_desc_->data_ptr = nullptr;
-  array_desc_->data_ptr =
-      static_cast<void*>(static_cast<char*>(data_ptr) + char_offset);
-}
-
-void array::move_shared_buffer(array other) {
-  move_shared_buffer(other, other.strides(), other.flags(), other.data_size());
-}
-
 array::~array() {
   if (array_desc_ == nullptr) {
     return;
   }
 
-  // Ignore arrays that might be detached during eval
-  if (status() == array::Status::scheduled) {
+  // Detached/detaching
+  if (array_desc_->primitive == nullptr) {
     return;
   }
 

mlx/array.h

@@ -10,6 +10,7 @@
 #include "mlx/allocator.h"
 #include "mlx/dtype.h"
 #include "mlx/event.h"
+#include "mlx/small_vector.h"
 
 namespace mlx::core {
 
@@ -18,8 +19,8 @@ class Primitive;
 
 using Deleter = std::function<void(allocator::Buffer)>;
 using ShapeElem = int32_t;
-using Shape = std::vector<ShapeElem>;
-using Strides = std::vector<int64_t>;
+using Shape = SmallVector<ShapeElem>;
+using Strides = SmallVector<int64_t>;
 
 class array {
   /* An array is really a node in a graph. It contains a shared ArrayDesc
@@ -199,6 +200,13 @@ class array {
       const std::shared_ptr<Primitive>& primitive,
       const std::vector<array>& inputs);
 
+  /**
+   * Get a new array that refers to the same data as the input but with a
+   * non-owning pointer to it. Note the array is detached from the graph and has
+   * no inputs, siblings or primitive.
+   */
+  static array unsafe_weak_copy(const array& other);
+
   /** A unique identifier for an array. */
   std::uintptr_t id() const {
     return reinterpret_cast<std::uintptr_t>(array_desc_.get());
@@ -217,6 +225,10 @@ class array {
     // Not copyable
     Data(const Data& d) = delete;
     Data& operator=(const Data& d) = delete;
+    Data(Data&& o) : buffer(o.buffer), d(o.d) {
+      o.buffer = allocator::Buffer(nullptr);
+      o.d = [](allocator::Buffer) {};
+    }
     ~Data() {
       d(buffer);
     }
@@ -243,18 +255,6 @@ class array {
     bool col_contiguous : 1;
   };
 
-  /** Build an array from all the info held by the array description. Including
-   * the buffer, strides, flags.
-   */
-  explicit array(
-      allocator::Buffer data,
-      Shape shape,
-      Dtype dtype,
-      Strides strides,
-      size_t data_size,
-      Flags flags,
-      Deleter deleter = allocator::free);
-
   /** The array's primitive. */
   Primitive& primitive() const {
     return *(array_desc_->primitive);
@@ -344,11 +344,11 @@ class array {
     return allocator::allocator().size(buffer());
   }
 
-  // Return a copy of the shared pointer
-  // to the array::Data struct
-  std::shared_ptr<Data> data_shared_ptr() const {
+  // Return the shared pointer to the array::Data struct
+  const std::shared_ptr<Data>& data_shared_ptr() const {
     return array_desc_->data;
   }
+
   // Return a raw pointer to the arrays data
   template <typename T>
   T* data() {
@@ -361,15 +361,10 @@ class array {
   }
 
   enum Status {
-    // The ouptut of a computation which has not been scheduled.
+    // The output of a computation which has not been scheduled.
     // For example, the status of `x` in `auto x = a + b`.
     unscheduled,
 
-    // The ouptut of a computation which has been scheduled but `eval_*` has
-    // not yet been called on the array's primitive. A possible
-    // status of `x` in `auto x = a + b; eval(x);`
-    scheduled,
-
     // The array's `eval_*` function has been run, but the computation is not
     // necessarily complete. The array will have memory allocated and if it is
     // not a tracer then it will be detached from the graph.
@@ -406,6 +401,10 @@ class array {
     array_desc_->event = std::move(e);
   }
 
+  void detach_event() const {
+    array_desc_->event = Event{};
+  }
+
   // Mark the array as a tracer array (true) or not.
   void set_tracer(bool is_tracer) {
     array_desc_->is_tracer = is_tracer;
@@ -431,15 +430,6 @@ class array {
 
   void copy_shared_buffer(const array& other);
 
-  void move_shared_buffer(
-      array other,
-      const Strides& strides,
-      Flags flags,
-      size_t data_size,
-      size_t offset = 0);
-
-  void move_shared_buffer(array other);
-
   void overwrite_descriptor(const array& other) {
     array_desc_ = other.array_desc_;
   }

mlx/backend/common/CMakeLists.txt

@@ -1,6 +1,7 @@
 target_sources(
   mlx
-  PRIVATE ${CMAKE_CURRENT_SOURCE_DIR}/compiled.cpp
+  PRIVATE ${CMAKE_CURRENT_SOURCE_DIR}/broadcasting.cpp
+          ${CMAKE_CURRENT_SOURCE_DIR}/compiled.cpp
           ${CMAKE_CURRENT_SOURCE_DIR}/common.cpp
           ${CMAKE_CURRENT_SOURCE_DIR}/load.cpp
           ${CMAKE_CURRENT_SOURCE_DIR}/reduce.cpp

mlx/backend/common/binary.h

@@ -38,25 +38,20 @@ inline void set_binary_op_output_data(
     const array& a,
     const array& b,
     array& out,
-    BinaryOpType bopt,
-    bool donate_with_move = false) {
+    BinaryOpType bopt) {
   bool b_donatable = is_donatable(b, out);
   bool a_donatable = is_donatable(a, out);
   switch (bopt) {
     case BinaryOpType::ScalarScalar:
       out.set_data(
-          allocator::malloc_or_wait(out.itemsize()), 1, a.strides(), a.flags());
+          allocator::malloc(out.itemsize()), 1, a.strides(), a.flags());
       break;
     case BinaryOpType::ScalarVector:
       if (b_donatable) {
-        if (donate_with_move) {
-          out.move_shared_buffer(b);
-        } else {
-          out.copy_shared_buffer(b);
-        }
+        out.copy_shared_buffer(b);
       } else {
         out.set_data(
-            allocator::malloc_or_wait(b.data_size() * out.itemsize()),
+            allocator::malloc(b.data_size() * out.itemsize()),
             b.data_size(),
             b.strides(),
             b.flags());
@@ -64,14 +59,10 @@ inline void set_binary_op_output_data(
       break;
     case BinaryOpType::VectorScalar:
       if (a_donatable) {
-        if (donate_with_move) {
-          out.move_shared_buffer(a);
-        } else {
-          out.copy_shared_buffer(a);
-        }
+        out.copy_shared_buffer(a);
       } else {
         out.set_data(
-            allocator::malloc_or_wait(a.data_size() * out.itemsize()),
+            allocator::malloc(a.data_size() * out.itemsize()),
             a.data_size(),
             a.strides(),
             a.flags());
@@ -79,20 +70,12 @@ inline void set_binary_op_output_data(
       break;
     case BinaryOpType::VectorVector:
       if (a_donatable) {
-        if (donate_with_move) {
-          out.move_shared_buffer(a);
-        } else {
-          out.copy_shared_buffer(a);
-        }
+        out.copy_shared_buffer(a);
       } else if (b_donatable) {
-        if (donate_with_move) {
-          out.move_shared_buffer(b);
-        } else {
-          out.copy_shared_buffer(b);
-        }
+        out.copy_shared_buffer(b);
       } else {
         out.set_data(
-            allocator::malloc_or_wait(a.data_size() * out.itemsize()),
+            allocator::malloc(a.data_size() * out.itemsize()),
             a.data_size(),
             a.strides(),
             a.flags());
@@ -100,20 +83,12 @@ inline void set_binary_op_output_data(
       break;
     case BinaryOpType::General:
       if (a_donatable && a.flags().row_contiguous && a.size() == out.size()) {
-        if (donate_with_move) {
-          out.move_shared_buffer(a);
-        } else {
-          out.copy_shared_buffer(a);
-        }
+        out.copy_shared_buffer(a);
       } else if (
           b_donatable && b.flags().row_contiguous && b.size() == out.size()) {
-        if (donate_with_move) {
-          out.move_shared_buffer(b);
-        } else {
-          out.copy_shared_buffer(b);
-        }
+        out.copy_shared_buffer(b);
       } else {
-        out.set_data(allocator::malloc_or_wait(out.nbytes()));
+        out.set_data(allocator::malloc(out.nbytes()));
       }
       break;
   }

mlx/backend/common/broadcasting.cpp

@@ -0,0 +1,24 @@
+// Copyright © 2024 Apple Inc.
+
+#include "mlx/backend/common/utils.h"
+
+namespace mlx::core {
+
+void broadcast(const array& in, array& out) {
+  if (out.size() == 0) {
+    out.set_data(nullptr);
+    return;
+  }
+  Strides strides(out.ndim(), 0);
+  int diff = out.ndim() - in.ndim();
+  for (int i = in.ndim() - 1; i >= 0; --i) {
+    strides[i + diff] = (in.shape()[i] == 1) ? 0 : in.strides()[i];
+  }
+  auto flags = in.flags();
+  if (out.size() > in.size()) {
+    flags.row_contiguous = flags.col_contiguous = false;
+  }
+  out.copy_shared_buffer(in, strides, flags, in.data_size());
+}
+
+} // namespace mlx::core

mlx/backend/common/broadcasting.h

@@ -1,10 +1,11 @@
 // Copyright © 2024 Apple Inc.
+
 #pragma once
 
+#include "mlx/array.h"
+
 namespace mlx::core {
 
-void encode_wait(Event e);
-
-void encode_signal(Event e);
+void broadcast(const array& in, array& out);
 
 } // namespace mlx::core

mlx/backend/common/buffer_cache.h

@@ -0,0 +1,157 @@
+// Copyright © 2025 Apple Inc.
+
+#pragma once
+
+#include <cassert>
+#include <functional>
+#include <map>
+
+namespace mlx::core {
+
+template <typename T>
+class BufferCache {
+ public:
+  BufferCache(
+      size_t page_size,
+      std::function<size_t(T*)> get_size,
+      std::function<void(T*)> free)
+      : page_size_(page_size),
+        get_size_(std::move(get_size)),
+        free_(std::move(free)) {}
+
+  ~BufferCache() {
+    clear();
+  }
+
+  BufferCache(const BufferCache&) = delete;
+  BufferCache& operator=(const BufferCache&) = delete;
+
+  T* reuse_from_cache(size_t size) {
+    // Find the closest buffer in pool.
+    auto it = buffer_pool_.lower_bound(size);
+    if (it == buffer_pool_.end() ||
+        it->first >= std::min(2 * size, size + 2 * page_size_)) {
+      return nullptr;
+    }
+
+    // Collect from the cache.
+    T* buf = it->second->buf;
+    pool_size_ -= it->first;
+
+    // Remove from record.
+    remove_from_list(it->second);
+    buffer_pool_.erase(it);
+    return buf;
+  }
+
+  void recycle_to_cache(T* buf) {
+    assert(buf);
+    // Add to cache.
+    BufferHolder* bh = new BufferHolder(buf);
+    add_at_head(bh);
+    size_t size = get_size_(buf);
+    pool_size_ += size;
+    buffer_pool_.emplace(size, bh);
+  }
+
+  int release_cached_buffers(size_t min_bytes_to_free) {
+    if (min_bytes_to_free >= 0.9 * pool_size_) {
+      return clear();
+    } else {
+      int n_release = 0;
+      size_t total_bytes_freed = 0;
+
+      while (tail_ && (total_bytes_freed < min_bytes_to_free)) {
+        // Release buffer.
+        size_t size = get_size_(tail_->buf);
+        total_bytes_freed += size;
+        free_(tail_->buf);
+        n_release++;
+
+        // Remove from record.
+        auto its = buffer_pool_.equal_range(size);
+        auto it = std::find_if(its.first, its.second, [this](const auto& el) {
+          return el.second == tail_;
+        });
+        assert(it != buffer_pool_.end());
+        buffer_pool_.erase(it);
+        remove_from_list(tail_);
+      }
+
+      pool_size_ -= total_bytes_freed;
+      return n_release;
+    }
+  }
+
+  int clear() {
+    int n_release = 0;
+    for (auto& [size, holder] : buffer_pool_) {
+      free_(holder->buf);
+      n_release++;
+      delete holder;
+    }
+    buffer_pool_.clear();
+    pool_size_ = 0;
+    head_ = nullptr;
+    tail_ = nullptr;
+    return n_release;
+  }
+
+  size_t cache_size() const {
+    return pool_size_;
+  }
+
+  size_t page_size() const {
+    return page_size_;
+  }
+
+ private:
+  struct BufferHolder {
+   public:
+    explicit BufferHolder(T* buf_) : buf(buf_) {}
+
+    BufferHolder* prev{nullptr};
+    BufferHolder* next{nullptr};
+    T* buf;
+  };
+
+  void add_at_head(BufferHolder* to_add) {
+    if (!head_) {
+      head_ = to_add;
+      tail_ = to_add;
+    } else {
+      head_->prev = to_add;
+      to_add->next = head_;
+      head_ = to_add;
+    }
+  }
+
+  void remove_from_list(BufferHolder* to_remove) {
+    if (to_remove->prev && to_remove->next) { // if middle
+      to_remove->prev->next = to_remove->next;
+      to_remove->next->prev = to_remove->prev;
+    } else if (to_remove->prev && to_remove == tail_) { // if tail
+      tail_ = to_remove->prev;
+      tail_->next = nullptr;
+    } else if (to_remove == head_ && to_remove->next) { // if head
+      head_ = to_remove->next;
+      head_->prev = nullptr;
+    } else if (to_remove == head_ && to_remove == tail_) { // if only element
+      head_ = nullptr;
+      tail_ = nullptr;
+    }
+
+    delete to_remove;
+  }
+
+  std::multimap<size_t, BufferHolder*> buffer_pool_;
+  BufferHolder* head_{nullptr};
+  BufferHolder* tail_{nullptr};
+  size_t pool_size_{0};
+
+  const size_t page_size_;
+  std::function<size_t(T*)> get_size_;
+  std::function<void(T*)> free_;
+};
+
+} // namespace mlx::core

mlx/backend/common/common.cpp

@@ -1,6 +1,7 @@
 // Copyright © 2024 Apple Inc.
 #include <cassert>
 
+#include "mlx/backend/common/broadcasting.h"
 #include "mlx/backend/common/utils.h"
 #include "mlx/primitives.h"
 
@@ -39,24 +40,7 @@ void AsStrided::eval(const std::vector<array>& inputs, array& out) {
   // rely on data_size anyway.
   size_t data_size = out.size();
 
-  return move_or_copy(in, out, strides_, flags, data_size, offset_);
-}
-
-void broadcast(const array& in, array& out) {
-  if (out.size() == 0) {
-    out.set_data(nullptr);
-    return;
-  }
-  Strides strides(out.ndim(), 0);
-  int diff = out.ndim() - in.ndim();
-  for (int i = in.ndim() - 1; i >= 0; --i) {
-    strides[i + diff] = (in.shape()[i] == 1) ? 0 : in.strides()[i];
-  }
-  auto flags = in.flags();
-  if (out.size() > in.size()) {
-    flags.row_contiguous = flags.col_contiguous = false;
-  }
-  move_or_copy(in, out, strides, flags, in.data_size());
+  return out.copy_shared_buffer(in, strides_, flags, data_size, offset_);
 }
 
 void Broadcast::eval(const std::vector<array>& inputs, array& out) {
@@ -69,7 +53,7 @@ void BroadcastAxes::eval(const std::vector<array>& inputs, array& out) {
 
 void Copy::eval(const std::vector<array>& inputs, array& out) {
   assert(inputs.size() == 1);
-  move_or_copy(inputs[0], out);
+  out.copy_shared_buffer(inputs[0]);
 }
 
 void CustomTransforms::eval(
@@ -78,7 +62,7 @@ void CustomTransforms::eval(
   assert(inputs.size() > outputs.size());
   for (int i = 0, j = inputs.size() - outputs.size(); i < outputs.size();
        i++, j++) {
-    move_or_copy(inputs[j], outputs[i]);
+    outputs[i].copy_shared_buffer(inputs[j]);
   }
 }
 
@@ -87,7 +71,7 @@ void Depends::eval(
     std::vector<array>& outputs) {
   assert(inputs.size() > outputs.size());
   for (int i = 0; i < outputs.size(); i++) {
-    move_or_copy(inputs[i], outputs[i]);
+    outputs[i].copy_shared_buffer(inputs[i]);
   }
 }
 
@@ -98,12 +82,12 @@ void ExpandDims::eval(const std::vector<array>& inputs, array& out) {
   for (auto ax : axes_) {
     strides.insert(strides.begin() + ax, 1);
   }
-  move_or_copy(in, out, strides, in.flags(), in.data_size());
+  out.copy_shared_buffer(in, strides, in.flags(), in.data_size());
 }
 
 void NumberOfElements::eval(const std::vector<array>& inputs, array& out) {
   assert(inputs.size() == 1);
-  out.set_data(allocator::malloc_or_wait(out.nbytes()));
+  out.set_data(allocator::malloc(out.nbytes()));
 
   double numel = 1;
   for (auto ax : axes_) {
@@ -210,7 +194,7 @@ void shared_buffer_reshape(
     auto max_dim = std::max_element(out.shape().begin(), out.shape().end());
     flags.col_contiguous = out.size() <= 1 || out.size() == *max_dim;
   }
-  move_or_copy(in, out, out_strides, flags, in.data_size());
+  out.copy_shared_buffer(in, out_strides, flags, in.data_size());
 }
 
 void Split::eval(
@@ -276,12 +260,12 @@ void Squeeze::eval(const std::vector<array>& inputs, array& out) {
       strides.push_back(in.strides(i));
     }
   }
-  move_or_copy(in, out, strides, in.flags(), in.data_size());
+  out.copy_shared_buffer(in, strides, in.flags(), in.data_size());
 }
 
 void StopGradient::eval(const std::vector<array>& inputs, array& out) {
   assert(inputs.size() == 1);
-  move_or_copy(inputs[0], out);
+  out.copy_shared_buffer(inputs[0]);
 }
 
 void Transpose::eval(const std::vector<array>& inputs, array& out) {
@@ -315,7 +299,7 @@ void Transpose::eval(const std::vector<array>& inputs, array& out) {
       b_stride *= out.shape(ri);
     }
   }
-  move_or_copy(in, out, out_strides, flags, in.data_size());
+  out.copy_shared_buffer(in, out_strides, flags, in.data_size());
 }
 
 } // namespace mlx::core

mlx/backend/common/compiled.cpp

@@ -1,8 +1,7 @@
 // Copyright © 2023-2024 Apple Inc.
 
 #include "mlx/backend/common/compiled.h"
-#include "mlx/graph_utils.h"
-#include "mlx/primitives.h"
+#include "mlx/backend/common/utils.h"
 #include "mlx/utils.h"
 
 namespace mlx::core {
@@ -15,6 +14,8 @@ void print_constant(std::ostream& os, const array& x) {
       return print_float_constant<float16_t>(os, x);
     case bfloat16:
       return print_float_constant<bfloat16_t>(os, x);
+    case float64:
+      return print_float_constant<double>(os, x);
     case complex64:
       return print_complex_constant<complex64_t>(os, x);
     case int8:
@@ -51,6 +52,8 @@ std::string get_type_string(Dtype d) {
       return "float16_t";
     case bfloat16:
       return "bfloat16_t";
+    case float64:
+      return "double";
     case complex64:
       return "complex64_t";
     case bool_:
@@ -79,55 +82,6 @@ std::string get_type_string(Dtype d) {
   }
 }
 
-std::string build_lib_name(
-    const std::vector<array>& inputs,
-    const std::vector<array>& outputs,
-    const std::vector<array>& tape,
-    const std::unordered_set<uintptr_t>& constant_ids) {
-  NodeNamer namer;
-  std::ostringstream os;
-  std::ostringstream constant_hasher;
-
-  // Fill the input names. This is not really necessary, I just like having A,
-  // B, C, ... as the inputs.
-  for (auto& x : inputs) {
-    namer.get_name(x);
-  }
-
-  // The primitives describing the tape. For unary and binary primitives this
-  // must be enough to describe the full computation.
-  for (auto& a : tape) {
-    // name and type of output
-    os << namer.get_name(a) << kindof(a.dtype()) << a.itemsize();
-    // computation performed
-    a.primitive().print(os);
-    // name of inputs to the function
-    for (auto& inp : a.inputs()) {
-      os << namer.get_name(inp);
-    }
-  }
-  os << "_";
-
-  for (auto& x : inputs) {
-    if (constant_ids.find(x.id()) != constant_ids.end()) {
-      os << "C";
-      print_constant(constant_hasher, x);
-    } else {
-      os << (is_scalar(x) ? "S" : "V");
-    }
-  }
-  os << "_";
-  for (auto& x : inputs) {
-    if (constant_ids.find(x.id()) != constant_ids.end()) {
-      continue;
-    }
-    os << kindof(x.dtype()) << x.itemsize();
-  }
-  os << "_" << std::hash<std::string>{}(constant_hasher.str());
-
-  return os.str();
-}
-
 bool compiled_check_contiguity(
     const std::vector<array>& inputs,
     const Shape& shape) {
@@ -159,10 +113,8 @@ bool compiled_check_contiguity(
 void compiled_allocate_outputs(
     const std::vector<array>& inputs,
     std::vector<array>& outputs,
-    const std::vector<array>& inputs_,
-    const std::unordered_set<uintptr_t>& constant_ids_,
-    bool contiguous,
-    bool move_buffers /* = false */) {
+    const std::function<bool(size_t)>& is_constant,
+    bool contiguous) {
   if (contiguous) {
     int o = 0;
     Strides strides;
@@ -176,13 +128,8 @@ void compiled_allocate_outputs(
       // - Donatable
       // - Not a constant
       if (in.itemsize() == outputs[o].itemsize() && !is_scalar(in) &&
-          in.is_donatable() &&
-          constant_ids_.find(inputs_[i].id()) == constant_ids_.end()) {
-        if (move_buffers) {
-          outputs[o++].move_shared_buffer(in);
-        } else {
-          outputs[o++].copy_shared_buffer(in);
-        }
+          in.is_donatable() && is_constant(i)) {
+        outputs[o++].copy_shared_buffer(in);
       }
       // Get representative input flags to properly set non-donated outputs
       if (strides.empty() && in.size() == outputs[0].size()) {
@@ -193,7 +140,7 @@ void compiled_allocate_outputs(
     }
     for (; o < outputs.size(); ++o) {
       outputs[o].set_data(
-          allocator::malloc_or_wait(data_size * outputs[o].itemsize()),
+          allocator::malloc(data_size * outputs[o].itemsize()),
           data_size,
           strides,
           flags);
@@ -209,21 +156,86 @@ void compiled_allocate_outputs(
       // - Not a constant
       if (in.flags().row_contiguous && in.size() == outputs[o].size() &&
           in.itemsize() == outputs[o].itemsize() && in.is_donatable() &&
-          constant_ids_.find(inputs_[i].id()) == constant_ids_.end()) {
-        if (move_buffers) {
-          outputs[o].move_shared_buffer(
-              in, outputs[o].strides(), in.flags(), in.data_size());
-        } else {
-          outputs[o].copy_shared_buffer(
-              in, outputs[o].strides(), in.flags(), in.data_size());
-        }
+          is_constant(i)) {
+        outputs[o].copy_shared_buffer(
+            in, outputs[o].strides(), in.flags(), in.data_size());
         o++;
       }
     }
     for (; o < outputs.size(); ++o) {
-      outputs[o].set_data(allocator::malloc_or_wait(outputs[o].nbytes()));
+      outputs[o].set_data(allocator::malloc(outputs[o].nbytes()));
     }
   }
 }
 
+std::tuple<bool, Shape, std::vector<Strides>> compiled_collapse_contiguous_dims(
+    const std::vector<array>& inputs,
+    const array& out,
+    const std::function<bool(size_t)>& is_constant) {
+  const Shape& shape = out.shape();
+  bool contiguous = compiled_check_contiguity(inputs, shape);
+  if (contiguous) {
+    return {true, shape, {}};
+  }
+
+  std::vector<Strides> strides_vec{out.strides()};
+  for (size_t i = 0; i < inputs.size(); ++i) {
+    // Skip constants.
+    if (is_constant(i)) {
+      continue;
+    }
+
+    // Skip scalar inputs.
+    const auto& x = inputs[i];
+    if (is_scalar(x)) {
+      continue;
+    }
+
+    // Broadcast the inputs to the output shape.
+    Strides xstrides;
+    size_t j = 0;
+    for (; j < shape.size() - x.ndim(); ++j) {
+      if (shape[j] == 1) {
+        xstrides.push_back(out.strides()[j]);
+      } else {
+        xstrides.push_back(0);
+      }
+    }
+    for (size_t i = 0; i < x.ndim(); ++i, ++j) {
+      if (x.shape(i) == 1) {
+        if (shape[j] == 1) {
+          xstrides.push_back(out.strides()[j]);
+        } else {
+          xstrides.push_back(0);
+        }
+      } else {
+        xstrides.push_back(x.strides()[i]);
+      }
+    }
+    strides_vec.push_back(std::move(xstrides));
+  }
+
+  auto tup = collapse_contiguous_dims(shape, strides_vec, INT32_MAX);
+  return {false, std::move(std::get<0>(tup)), std::move(std::get<1>(tup))};
+}
+
+bool compiled_use_large_index(
+    const std::vector<array>& inputs,
+    const std::vector<array>& outputs,
+    bool contiguous) {
+  if (contiguous) {
+    size_t max_size = 0;
+    for (const auto& in : inputs) {
+      max_size = std::max(max_size, in.data_size());
+    }
+    return max_size > UINT32_MAX;
+  } else {
+    size_t max_size = 0;
+    for (const auto& o : outputs) {
+      max_size = std::max(max_size, o.size());
+    }
+    return max_size > UINT32_MAX;
+  }
+}
+
 } // namespace mlx::core

mlx/backend/common/compiled.h

@@ -1,9 +1,8 @@
 // Copyright © 2023-2024 Apple Inc.
 #pragma once
 
+#include <functional>
 #include <iomanip>
-#include <sstream>
-#include <unordered_set>
 
 #include "mlx/array.h"
 #include "mlx/primitives.h"
@@ -14,19 +13,17 @@ inline bool is_static_cast(const Primitive& p) {
   return (typeid(p) == typeid(Broadcast) || typeid(p) == typeid(AsType));
 }
 
-std::string build_lib_name(
-    const std::vector<array>& inputs,
-    const std::vector<array>& outputs,
-    const std::vector<array>& tape,
-    const std::unordered_set<uintptr_t>& constant_ids);
-
 std::string get_type_string(Dtype d);
 
 template <typename T>
 void print_float_constant(std::ostream& os, const array& x) {
   auto old_precision = os.precision();
-  os << std::setprecision(std::numeric_limits<float>::digits10 + 1)
-     << x.item<T>() << std::setprecision(old_precision);
+  if constexpr (std::is_same_v<T, double>) {
+    os << std::setprecision(std::numeric_limits<double>::digits10 + 1);
+  } else {
+    os << std::setprecision(std::numeric_limits<float>::digits10 + 1);
+  }
+  os << x.item<T>() << std::setprecision(old_precision);
 }
 
 template <typename T>
@@ -60,9 +57,19 @@ bool compiled_check_contiguity(
 void compiled_allocate_outputs(
     const std::vector<array>& inputs,
     std::vector<array>& outputs,
-    const std::vector<array>& inputs_,
-    const std::unordered_set<uintptr_t>& constant_ids_,
-    bool contiguous,
-    bool move_buffers = false);
+    const std::function<bool(size_t)>& is_constant,
+    bool contiguous);
+
+// Collapse contiguous dims ignoring scalars and constants.
+std::tuple<bool, Shape, std::vector<Strides>> compiled_collapse_contiguous_dims(
+    const std::vector<array>& inputs,
+    const array& out,
+    const std::function<bool(size_t)>& is_constant);
+
+// Return whether the kernel should use large index.
+bool compiled_use_large_index(
+    const std::vector<array>& inputs,
+    const std::vector<array>& outputs,
+    bool contiguous);
 
 } // namespace mlx::core

mlx/backend/common/copy.h

@@ -2,7 +2,7 @@
 
 #pragma once
 
-#include "mlx/array.h"
+#include "mlx/backend/common/utils.h"
 
 namespace mlx::core {
 
@@ -22,4 +22,25 @@ enum class CopyType {
   GeneralGeneral
 };
 
+inline bool set_copy_output_data(const array& in, array& out, CopyType ctype) {
+  if (ctype == CopyType::Vector) {
+    // If the input is donateable, we are doing a vector copy and the types
+    // have the same size, then the input buffer can hold the output.
+    if (is_donatable(in, out)) {
+      out.copy_shared_buffer(in);
+      return true;
+    } else {
+      out.set_data(
+          allocator::malloc(in.data_size() * out.itemsize()),
+          in.data_size(),
+          in.strides(),
+          in.flags());
+      return false;
+    }
+  } else {
+    out.set_data(allocator::malloc(out.nbytes()));
+    return false;
+  }
+}
+
 } // namespace mlx::core

mlx/backend/common/hadamard.h

@@ -99,7 +99,11 @@ inline std::pair<int, int> decompose_hadamard(int n) {
           "[hadamard] Only supports n = m*2^k where m in (1, 12, 20, 28).");
     }
   }
+  if (n > (1 << 26)) {
+    throw std::invalid_argument(
+        "[hadamard] Only supports n = m*2^k where k <= 26");
+  }
   return {n, m};
 }
 
-} // namespace mlx::core
+} // namespace mlx::core

mlx/backend/common/load.cpp

@@ -3,7 +3,8 @@
 #include <algorithm>
 #include <utility>
 
-#include "mlx/backend/common/load.h"
+#include "mlx/primitives.h"
+#include "mlx/scheduler.h"
 
 namespace {
 
@@ -26,26 +27,31 @@ void swap_endianness(uint8_t* data_bytes, size_t N) {
 
 namespace mlx::core {
 
-void load(
-    array& out,
-    size_t offset,
-    const std::shared_ptr<io::Reader>& reader,
-    bool swap_endianness_) {
-  reader->read(out.data<char>(), out.nbytes(), offset);
-
-  if (swap_endianness_) {
-    switch (out.itemsize()) {
-      case 2:
-        swap_endianness<2>(out.data<uint8_t>(), out.data_size());
-        break;
-      case 4:
-        swap_endianness<4>(out.data<uint8_t>(), out.data_size());
-        break;
-      case 8:
-        swap_endianness<8>(out.data<uint8_t>(), out.data_size());
-        break;
+void Load::eval_cpu(const std::vector<array>& inputs, array& out) {
+  out.set_data(allocator::malloc(out.nbytes()));
+  auto read_task = [out_ptr = out.data<char>(),
+                    size = out.size(),
+                    itemsize = out.itemsize(),
+                    offset = offset_,
+                    reader = reader_,
+                    swap_endianness_ = swap_endianness_]() mutable {
+    reader->read(out_ptr, size * itemsize, offset);
+    if (swap_endianness_) {
+      switch (itemsize) {
+        case 2:
+          swap_endianness<2>(reinterpret_cast<uint8_t*>(out_ptr), size);
+          break;
+        case 4:
+          swap_endianness<4>(reinterpret_cast<uint8_t*>(out_ptr), size);
+          break;
+        case 8:
+          swap_endianness<8>(reinterpret_cast<uint8_t*>(out_ptr), size);
+          break;
+      }
     }
-  }
+  };
+  auto fut = io::thread_pool().enqueue(std::move(read_task)).share();
+  scheduler::enqueue(stream(), [fut = std::move(fut)]() { fut.wait(); });
 }
 
 } // namespace mlx::core

mlx/backend/common/load.h

@@ -1,14 +0,0 @@
-// Copyright © 2024 Apple Inc.
-
-#include "mlx/array.h"
-#include "mlx/io/load.h"
-
-namespace mlx::core {
-
-void load(
-    array& out,
-    size_t offset,
-    const std::shared_ptr<io::Reader>& reader,
-    bool swap_endianess);
-
-} // namespace mlx::core

mlx/backend/common/matmul.h

@@ -0,0 +1,67 @@
+// Copyright © 2025 Apple Inc.
+
+#pragma once
+
+#include "mlx/backend/common/utils.h"
+#include "mlx/utils.h"
+
+#include <sstream>
+
+namespace mlx::core {
+
+inline std::tuple<Shape, Strides, Strides> collapse_batches(
+    const array& a,
+    const array& b) {
+  if (a.ndim() == 2) {
+    return {{1}, {0}, {0}};
+  }
+
+  Shape A_bshape{a.shape().begin(), a.shape().end() - 2};
+  Strides A_bstride{a.strides().begin(), a.strides().end() - 2};
+  Strides B_bstride{b.strides().begin(), b.strides().end() - 2};
+
+  auto [batch_shape, batch_strides] =
+      collapse_contiguous_dims(A_bshape, std::vector{A_bstride, B_bstride});
+
+  auto a_batch_strides = batch_strides[0];
+  auto b_batch_strides = batch_strides[1];
+
+  if (batch_shape.empty()) {
+    batch_shape.push_back(1);
+    a_batch_strides.push_back(0);
+    b_batch_strides.push_back(0);
+  }
+
+  return std::make_tuple(batch_shape, a_batch_strides, b_batch_strides);
+}
+
+inline std::tuple<Shape, Strides, Strides, Strides>
+collapse_batches(const array& a, const array& b, const array& c) {
+  if (a.ndim() == 2) {
+    return {{1}, {0}, {0}, {0}};
+  }
+
+  Shape A_bshape{a.shape().begin(), a.shape().end() - 2};
+  Strides A_bstride{a.strides().begin(), a.strides().end() - 2};
+  Strides B_bstride{b.strides().begin(), b.strides().end() - 2};
+  Strides C_bstride{c.strides().begin(), c.strides().end() - 2};
+
+  auto [batch_shape, batch_strides] = collapse_contiguous_dims(
+      A_bshape, std::vector{A_bstride, B_bstride, C_bstride});
+
+  auto A_batch_stride = batch_strides[0];
+  auto B_batch_stride = batch_strides[1];
+  auto C_batch_stride = batch_strides[2];
+
+  if (batch_shape.empty()) {
+    batch_shape.push_back(1);
+    A_batch_stride.push_back(0);
+    B_batch_stride.push_back(0);
+    C_batch_stride.push_back(0);
+  }
+
+  return std::make_tuple(
+      batch_shape, A_batch_stride, B_batch_stride, C_batch_stride);
+}
+
+} // namespace mlx::core

mlx/backend/common/reduce.cpp

@@ -5,11 +5,9 @@
 namespace mlx::core {
 
 std::pair<Shape, Strides> shapes_without_reduction_axes(
-    const array& x,
+    Shape shape,
+    Strides strides,
     const std::vector<int>& axes) {
-  auto shape = x.shape();
-  auto strides = x.strides();
-
   for (int i = axes.size() - 1; i >= 0; i--) {
     int a = axes[i];
     shape.erase(shape.begin() + a);
@@ -19,6 +17,15 @@ std::pair<Shape, Strides> shapes_without_reduction_axes(
   return std::make_pair(shape, strides);
 }
 
+std::pair<Shape, Strides> shapes_without_reduction_axes(
+    const array& x,
+    const std::vector<int>& axes) {
+  auto shape = x.shape();
+  auto strides = x.strides();
+  return shapes_without_reduction_axes(
+      std::move(shape), std::move(strides), axes);
+}
+
 ReductionPlan get_reduction_plan(const array& x, const std::vector<int>& axes) {
   // The data is all there and we are reducing over everything
   if (x.size() == x.data_size() && axes.size() == x.ndim() &&

mlx/backend/common/reduce.h

@@ -51,5 +51,9 @@ ReductionPlan get_reduction_plan(const array& x, const std::vector<int>& axes);
 std::pair<Shape, Strides> shapes_without_reduction_axes(
     const array& x,
     const std::vector<int>& axes);
+std::pair<Shape, Strides> shapes_without_reduction_axes(
+    Shape shape,
+    Strides strides,
+    const std::vector<int>& axes);
 
 } // namespace mlx::core

mlx/backend/common/slicing.cpp

@@ -14,6 +14,10 @@ std::tuple<int64_t, Strides> prepare_slice(
     data_offset += start_indices[i] * in.strides()[i];
     inp_strides[i] = in.strides()[i] * strides[i];
   }
+  // Normalize the offset
+  if (data_offset < 0) {
+    data_offset += in.data_size();
+  }
   return std::make_tuple(data_offset, inp_strides);
 }
 
@@ -32,7 +36,7 @@ void shared_buffer_slice(
   flags.col_contiguous = is_col_contiguous;
   flags.contiguous = (no_bsx_size == data_size);
 
-  move_or_copy(in, out, out_strides, flags, data_size, data_offset);
+  out.copy_shared_buffer(in, out_strides, flags, data_size, data_offset);
 }
 
 void slice(
@@ -54,9 +58,10 @@ void slice(
       data_end += end_idx * in.strides()[i];
     }
   }
-  // data_end can be -1
-  size_t data_size =
-      data_end < 0 ? (data_offset - data_end) : (data_end - data_offset);
+  if (data_end < 0) {
+    data_end += in.data_size();
+  }
+  size_t data_size = (data_end - data_offset);
   shared_buffer_slice(in, inp_strides, data_offset, data_size, out);
 }
 

mlx/backend/common/ternary.h

@@ -36,15 +36,10 @@ inline void set_ternary_op_output_data(
     const array& b,
     const array& c,
     array& out,
-    TernaryOpType topt,
-    bool donate_with_move = false) {
-  auto maybe_donate = [&out, donate_with_move](const array& x) {
+    TernaryOpType topt) {
+  auto maybe_donate = [&out](const array& x) {
     if (is_donatable(x, out)) {
-      if (donate_with_move) {
-        out.move_shared_buffer(x);
-      } else {
-        out.copy_shared_buffer(x);
-      }
+      out.copy_shared_buffer(x);
       return true;
     }
     return false;
@@ -53,12 +48,12 @@ inline void set_ternary_op_output_data(
   switch (topt) {
     case TernaryOpType::ScalarScalarScalar:
       out.set_data(
-          allocator::malloc_or_wait(out.itemsize()), 1, b.strides(), b.flags());
+          allocator::malloc(out.itemsize()), 1, b.strides(), b.flags());
       break;
     case TernaryOpType::VectorVectorVector:
       if (!(maybe_donate(a) || maybe_donate(b) || maybe_donate(c))) {
         out.set_data(
-            allocator::malloc_or_wait(out.itemsize() * b.data_size()),
+            allocator::malloc(out.itemsize() * b.data_size()),
             b.data_size(),
             b.strides(),
             b.flags());
@@ -69,7 +64,7 @@ inline void set_ternary_op_output_data(
       if (!((a.flags().row_contiguous && maybe_donate(a)) ||
             (b.flags().row_contiguous && maybe_donate(b)) ||
             (c.flags().row_contiguous && maybe_donate(c)))) {
-        out.set_data(allocator::malloc_or_wait(out.nbytes()));
+        out.set_data(allocator::malloc(out.nbytes()));
       }
       break;
   }

mlx/backend/common/unary.h

@@ -0,0 +1,26 @@
+// Copyright © 2025 Apple Inc.
+
+#pragma once
+
+#include "mlx/allocator.h"
+#include "mlx/backend/common/utils.h"
+
+namespace mlx::core {
+
+inline void set_unary_output_data(const array& in, array& out) {
+  if (in.flags().contiguous) {
+    if (is_donatable(in, out)) {
+      out.copy_shared_buffer(in);
+    } else {
+      out.set_data(
+          allocator::malloc(in.data_size() * out.itemsize()),
+          in.data_size(),
+          in.strides(),
+          in.flags());
+    }
+  } else {
+    out.set_data(allocator::malloc(out.nbytes()));
+  }
+}
+
+} // namespace mlx::core

mlx/backend/common/utils.cpp

@@ -1,29 +1,20 @@
 // Copyright © 2023-2024 Apple Inc.
 
+#include <dlfcn.h>
+
 #include "mlx/backend/common/utils.h"
 
 namespace mlx::core {
 
-void move_or_copy(const array& in, array& out) {
-  if (in.is_donatable()) {
-    out.move_shared_buffer(in);
-  } else {
-    out.copy_shared_buffer(in);
-  }
-}
-
-void move_or_copy(
-    const array& in,
-    array& out,
-    const Strides& strides,
-    array::Flags flags,
-    size_t data_size,
-    size_t offset /* = 0 */) {
-  if (in.is_donatable()) {
-    out.move_shared_buffer(in, strides, flags, data_size, offset);
-  } else {
-    out.copy_shared_buffer(in, strides, flags, data_size, offset);
-  }
+std::filesystem::path current_binary_dir() {
+  static std::filesystem::path binary_dir = []() {
+    Dl_info info;
+    if (!dladdr(reinterpret_cast<void*>(&current_binary_dir), &info)) {
+      throw std::runtime_error("Unable to get current binary dir.");
+    }
+    return std::filesystem::path(info.dli_fname).parent_path();
+  }();
+  return binary_dir;
 }
 
 std::tuple<Shape, std::vector<Strides>> collapse_contiguous_dims(
@@ -123,4 +114,118 @@ std::pair<Shape, Strides> collapse_contiguous_dims(
   return collapse_contiguous_dims(a.shape(), a.strides(), size_cap);
 }
 
+Dims get_block_dims_common(int dim0, int dim1, int dim2, int pow2 /* = 10 */) {
+  int pows[3] = {0, 0, 0};
+  int sum = 0;
+  while (true) {
+    int presum = sum;
+    // Check all the pows
+    if (dim0 >= (1 << (pows[0] + 1))) {
+      pows[0]++;
+      sum++;
+    }
+    if (sum == 10) {
+      break;
+    }
+    if (dim1 >= (1 << (pows[1] + 1))) {
+      pows[1]++;
+      sum++;
+    }
+    if (sum == 10) {
+      break;
+    }
+    if (dim2 >= (1 << (pows[2] + 1))) {
+      pows[2]++;
+      sum++;
+    }
+    if (sum == presum || sum == pow2) {
+      break;
+    }
+  }
+  return std::make_tuple(1ul << pows[0], 1ul << pows[1], 1ul << pows[2]);
+}
+
+Dims get_2d_grid_dims_common(const Shape& shape, const Strides& strides) {
+  // Dims with strides of 0 are ignored as they
+  // correspond to broadcasted dimensions
+  size_t grid_x = 1;
+  size_t grid_y = 1;
+  for (int i = 0; i < shape.size(); ++i) {
+    if (strides[i] == 0) {
+      continue;
+    }
+    if (grid_x * shape[i] < UINT32_MAX) {
+      grid_x *= shape[i];
+    } else {
+      grid_y *= shape[i];
+    }
+  }
+  if (grid_y > UINT32_MAX || grid_x > UINT32_MAX) {
+    throw std::runtime_error("Unable to safely factor shape.");
+  }
+  if (grid_y > grid_x) {
+    std::swap(grid_x, grid_y);
+  }
+  return std::make_tuple(
+      static_cast<uint32_t>(grid_x), static_cast<uint32_t>(grid_y), 1);
+}
+
+Dims get_2d_grid_dims_common(
+    const Shape& shape,
+    const Strides& strides,
+    size_t divisor) {
+  // Compute the 2d grid dimensions such that the total size of the grid is
+  // divided by divisor.
+  size_t grid_x = 1;
+  size_t grid_y = 1;
+  for (int i = 0; i < shape.size(); ++i) {
+    if (strides[i] == 0) {
+      continue;
+    }
+
+    // No need to add this shape we can just remove it from the divisor.
+    if (divisor % shape[i] == 0) {
+      divisor /= shape[i];
+      continue;
+    }
+
+    if (grid_x * shape[i] < UINT32_MAX) {
+      grid_x *= shape[i];
+    } else {
+      grid_y *= shape[i];
+    }
+
+    if (divisor > 1) {
+      if (grid_x % divisor == 0) {
+        grid_x /= divisor;
+        divisor = 1;
+      } else if (grid_y % divisor == 0) {
+        grid_y /= divisor;
+        divisor = 1;
+      }
+    }
+  }
+  if (grid_y > UINT32_MAX || grid_x > UINT32_MAX) {
+    throw std::runtime_error("Unable to safely factor shape.");
+  }
+  if (grid_y > grid_x) {
+    std::swap(grid_x, grid_y);
+  }
+  if (divisor > 1) {
+    grid_x = ((grid_x + divisor - 1) / divisor) * divisor;
+  }
+  return std::make_tuple(
+      static_cast<uint32_t>(grid_x), static_cast<uint32_t>(grid_y), 1);
+}
+
+std::pair<Dims, Dims> get_grid_and_block_common(int dim0, int dim1, int dim2) {
+  auto [bx, by, bz] = get_block_dims_common(dim0, dim1, dim2);
+  auto gx = (dim0 + bx - 1) / bx;
+  auto gy = (dim1 + by - 1) / by;
+  auto gz = (dim2 + bz - 1) / bz;
+
+  return std::make_pair(
+      std::make_tuple(gx, gy, gz), std::make_tuple(bx, by, bz));
+}
+
 } // namespace mlx::core

mlx/backend/common/utils.h

@@ -2,12 +2,17 @@
 
 #pragma once
 
+#include <filesystem>
+#include <tuple>
 #include <vector>
 
 #include "mlx/array.h"
 
 namespace mlx::core {
 
+// Return the directory that contains current shared library.
+std::filesystem::path current_binary_dir();
+
 inline int64_t
 elem_to_loc(int elem, const Shape& shape, const Strides& strides) {
   int64_t loc = 0;
@@ -70,6 +75,31 @@ std::pair<Shape, Strides> collapse_contiguous_dims(
     const array& a,
     int64_t size_cap = std::numeric_limits<int32_t>::max());
 
+// Compute the thread block dimensions which fit the given
+// input dimensions.
+// - The thread block dimensions will be powers of two
+// - The thread block size will be less than 2^pow2
+using Dims = std::tuple<uint32_t, uint32_t, uint32_t>;
+Dims get_block_dims_common(int dim0, int dim1, int dim2, int pow2 = 10);
+
+// Computes a 2D grid where each element is < UINT_MAX
+// Assumes:
+// - overall size (product of non-broadcasted dimensions) is < UINT_MAX^2
+// - shape and strides correspond to a contiguous (no holes) but
+//   possibly broadcasted array
+Dims get_2d_grid_dims_common(const Shape& shape, const Strides& strides);
+
+// Same as above but we do an implicit division with divisor.
+// Basically, equivalent to factorizing
+//    Prod(s \forall s in shape if strides[s] > 0) / divisor.
+Dims get_2d_grid_dims_common(
+    const Shape& shape,
+    const Strides& strides,
+    size_t divisor);
+
+// Get both the block and a grid of blocks that covers dim0, dim1 and dim2.
+std::pair<Dims, Dims> get_grid_and_block_common(int dim0, int dim1, int dim2);
+
 struct ContiguousIterator {
   inline void step() {
     int dims = shape_.size();
@@ -159,19 +189,17 @@ inline bool is_donatable(const array& in, const array& out) {
       in.buffer_size() <= out.nbytes() + donation_extra;
 }
 
-void move_or_copy(const array& in, array& out);
-void move_or_copy(
-    const array& in,
-    array& out,
-    const Strides& strides,
-    array::Flags flags,
-    size_t data_size,
-    size_t offset = 0);
-
 std::pair<bool, Strides> prepare_reshape(const array& in, const array& out);
 
 void shared_buffer_reshape(
     const array& in,
     const Strides& out_strides,
     array& out);
+
+template <typename T>
+inline SmallVector<T> remove_index(SmallVector<T> vec, size_t index) {
+  vec.erase(std::next(vec.begin(), index));
+  return vec;
+}
+
 } // namespace mlx::core

mlx/backend/cpu/CMakeLists.txt

@@ -40,11 +40,15 @@ add_dependencies(mlx cpu_compiled_preamble)
 
 target_sources(
   mlx
-  PRIVATE ${CMAKE_CURRENT_SOURCE_DIR}/arg_reduce.cpp
+  PRIVATE ${CMAKE_CURRENT_SOURCE_DIR}/available.cpp
+          ${CMAKE_CURRENT_SOURCE_DIR}/arg_reduce.cpp
           ${CMAKE_CURRENT_SOURCE_DIR}/binary.cpp
           ${CMAKE_CURRENT_SOURCE_DIR}/conv.cpp
           ${CMAKE_CURRENT_SOURCE_DIR}/copy.cpp
+          ${CMAKE_CURRENT_SOURCE_DIR}/distributed.cpp
+          ${CMAKE_CURRENT_SOURCE_DIR}/eig.cpp
           ${CMAKE_CURRENT_SOURCE_DIR}/eigh.cpp
+          ${CMAKE_CURRENT_SOURCE_DIR}/encoder.cpp
           ${CMAKE_CURRENT_SOURCE_DIR}/fft.cpp
           ${CMAKE_CURRENT_SOURCE_DIR}/hadamard.cpp
           ${CMAKE_CURRENT_SOURCE_DIR}/matmul.cpp
@@ -56,6 +60,7 @@ target_sources(
           ${CMAKE_CURRENT_SOURCE_DIR}/scan.cpp
           ${CMAKE_CURRENT_SOURCE_DIR}/select.cpp
           ${CMAKE_CURRENT_SOURCE_DIR}/softmax.cpp
+          ${CMAKE_CURRENT_SOURCE_DIR}/logsumexp.cpp
           ${CMAKE_CURRENT_SOURCE_DIR}/sort.cpp
           ${CMAKE_CURRENT_SOURCE_DIR}/threefry.cpp
           ${CMAKE_CURRENT_SOURCE_DIR}/indexing.cpp
@@ -65,13 +70,14 @@ target_sources(
           ${CMAKE_CURRENT_SOURCE_DIR}/inverse.cpp
           ${CMAKE_CURRENT_SOURCE_DIR}/cholesky.cpp
           ${CMAKE_CURRENT_SOURCE_DIR}/unary.cpp
+          ${CMAKE_CURRENT_SOURCE_DIR}/eval.cpp
           ${CMAKE_CURRENT_BINARY_DIR}/compiled_preamble.cpp)
 
 if(MLX_BUILD_ACCELERATE)
   target_sources(mlx PRIVATE ${CMAKE_CURRENT_SOURCE_DIR}/gemms/bnns.cpp)
 else()
-  target_sources(mlx PRIVATE ${CMAKE_CURRENT_SOURCE_DIR}/gemms/no_fp16.cpp
-                             ${CMAKE_CURRENT_SOURCE_DIR}/gemms/no_bf16.cpp)
+  target_sources(mlx PRIVATE ${CMAKE_CURRENT_SOURCE_DIR}/gemms/simd_fp16.cpp
+                             ${CMAKE_CURRENT_SOURCE_DIR}/gemms/simd_bf16.cpp)
 endif()
 
 if(IOS)

mlx/backend/cpu/arange.h

@@ -2,76 +2,27 @@
 
 #pragma once
 
-#include "mlx/allocator.h"
 #include "mlx/array.h"
+#include "mlx/backend/cpu/encoder.h"
 
 namespace mlx::core {
 
 namespace {
 
 template <typename T>
-void arange(T start, T next, array& out, size_t size) {
+void arange(T start, T next, array& out, size_t size, Stream stream) {
   auto ptr = out.data<T>();
   auto step_size = next - start;
-  for (int i = 0; i < size; ++i) {
-    ptr[i] = start;
-    start += step_size;
-  }
+  auto& encoder = cpu::get_command_encoder(stream);
+  encoder.set_output_array(out);
+  encoder.dispatch([ptr, start, step_size, size]() mutable {
+    for (int i = 0; i < size; ++i) {
+      ptr[i] = start;
+      start += step_size;
+    }
+  });
 }
 
 } // namespace
 
-void arange(
-    const std::vector<array>& inputs,
-    array& out,
-    double start,
-    double step) {
-  assert(inputs.size() == 0);
-  out.set_data(allocator::malloc_or_wait(out.nbytes()));
-  switch (out.dtype()) {
-    case bool_:
-      throw std::runtime_error("Bool type unsupported for arange.");
-      break;
-    case uint8:
-      arange<uint8_t>(start, start + step, out, out.size());
-      break;
-    case uint16:
-      arange<uint16_t>(start, start + step, out, out.size());
-      break;
-    case uint32:
-      arange<uint32_t>(start, start + step, out, out.size());
-      break;
-    case uint64:
-      arange<uint64_t>(start, start + step, out, out.size());
-      break;
-    case int8:
-      arange<int8_t>(start, start + step, out, out.size());
-      break;
-    case int16:
-      arange<int16_t>(start, start + step, out, out.size());
-      break;
-    case int32:
-      arange<int32_t>(start, start + step, out, out.size());
-      break;
-    case int64:
-      arange<int64_t>(start, start + step, out, out.size());
-      break;
-    case float16:
-      arange<float16_t>(start, start + step, out, out.size());
-      break;
-    case float32:
-      arange<float>(start, start + step, out, out.size());
-      break;
-    case float64:
-      arange<double>(start, start + step, out, out.size());
-      break;
-    case bfloat16:
-      arange<bfloat16_t>(start, start + step, out, out.size());
-      break;
-    case complex64:
-      arange<complex64_t>(start, start + step, out, out.size());
-      break;
-  }
-}
-
 } // namespace mlx::core

mlx/backend/cpu/arg_reduce.cpp

@@ -3,6 +3,7 @@
 #include <cassert>
 
 #include "mlx/backend/common/utils.h"
+#include "mlx/backend/cpu/encoder.h"
 #include "mlx/primitives.h"
 
 namespace mlx::core {
@@ -13,19 +14,20 @@ template <typename InT, typename OpT>
 void arg_reduce(const array& in, array& out, const OpT& op, int axis) {
   auto axis_size = in.shape()[axis];
   auto axis_stride = in.strides()[axis];
-  Strides strides = in.strides();
-  Shape shape = in.shape();
-  strides.erase(strides.begin() + axis);
-  shape.erase(shape.begin() + axis);
+  Strides strides = remove_index(in.strides(), axis);
+  Shape shape = remove_index(in.shape(), axis);
+  auto in_ptr = in.data<InT>();
+  auto out_ptr = out.data<uint32_t>();
+
   for (uint32_t i = 0; i < out.size(); ++i) {
     auto loc = elem_to_loc(i, shape, strides);
-    auto in_ptr = in.data<InT>() + loc;
+    auto local_in_ptr = in_ptr + loc;
     uint32_t ind_v = 0;
-    InT v = (*in_ptr);
-    for (uint32_t j = 0; j < axis_size; ++j, in_ptr += axis_stride) {
-      op(j, (*in_ptr), &ind_v, &v);
+    InT v = (*local_in_ptr);
+    for (uint32_t j = 0; j < axis_size; ++j, local_in_ptr += axis_stride) {
+      op(j, (*local_in_ptr), &ind_v, &v);
     }
-    out.data<uint32_t>()[i] = ind_v;
+    out_ptr[i] = ind_v;
   }
 }
 
@@ -64,52 +66,59 @@ void arg_reduce_dispatch(
 void ArgReduce::eval_cpu(const std::vector<array>& inputs, array& out) {
   assert(inputs.size() == 1);
   auto& in = inputs[0];
-  out.set_data(allocator::malloc_or_wait(out.nbytes()));
-
-  switch (in.dtype()) {
-    case bool_:
-      arg_reduce_dispatch<bool>(in, out, reduce_type_, axis_);
-      break;
-    case uint8:
-      arg_reduce_dispatch<uint8_t>(in, out, reduce_type_, axis_);
-      break;
-    case uint16:
-      arg_reduce_dispatch<uint16_t>(in, out, reduce_type_, axis_);
-      break;
-    case uint32:
-      arg_reduce_dispatch<uint32_t>(in, out, reduce_type_, axis_);
-      break;
-    case uint64:
-      arg_reduce_dispatch<uint64_t>(in, out, reduce_type_, axis_);
-      break;
-    case int8:
-      arg_reduce_dispatch<int8_t>(in, out, reduce_type_, axis_);
-      break;
-    case int16:
-      arg_reduce_dispatch<int16_t>(in, out, reduce_type_, axis_);
-      break;
-    case int32:
-      arg_reduce_dispatch<int32_t>(in, out, reduce_type_, axis_);
-      break;
-    case int64:
-      arg_reduce_dispatch<int64_t>(in, out, reduce_type_, axis_);
-      break;
-    case float16:
-      arg_reduce_dispatch<float16_t>(in, out, reduce_type_, axis_);
-      break;
-    case float32:
-      arg_reduce_dispatch<float>(in, out, reduce_type_, axis_);
-      break;
-    case bfloat16:
-      arg_reduce_dispatch<bfloat16_t>(in, out, reduce_type_, axis_);
-      break;
-    case float64:
-      arg_reduce_dispatch<double>(in, out, reduce_type_, axis_);
-      break;
-    case complex64:
-      arg_reduce_dispatch<complex64_t>(in, out, reduce_type_, axis_);
-      break;
-  }
+  out.set_data(allocator::malloc(out.nbytes()));
+  auto& encoder = cpu::get_command_encoder(stream());
+  encoder.set_input_array(in);
+  encoder.set_output_array(out);
+  encoder.dispatch([in = array::unsafe_weak_copy(in),
+                    out = array::unsafe_weak_copy(out),
+                    reduce_type_ = reduce_type_,
+                    axis_ = axis_]() mutable {
+    switch (in.dtype()) {
+      case bool_:
+        arg_reduce_dispatch<bool>(in, out, reduce_type_, axis_);
+        break;
+      case uint8:
+        arg_reduce_dispatch<uint8_t>(in, out, reduce_type_, axis_);
+        break;
+      case uint16:
+        arg_reduce_dispatch<uint16_t>(in, out, reduce_type_, axis_);
+        break;
+      case uint32:
+        arg_reduce_dispatch<uint32_t>(in, out, reduce_type_, axis_);
+        break;
+      case uint64:
+        arg_reduce_dispatch<uint64_t>(in, out, reduce_type_, axis_);
+        break;
+      case int8:
+        arg_reduce_dispatch<int8_t>(in, out, reduce_type_, axis_);
+        break;
+      case int16:
+        arg_reduce_dispatch<int16_t>(in, out, reduce_type_, axis_);
+        break;
+      case int32:
+        arg_reduce_dispatch<int32_t>(in, out, reduce_type_, axis_);
+        break;
+      case int64:
+        arg_reduce_dispatch<int64_t>(in, out, reduce_type_, axis_);
+        break;
+      case float16:
+        arg_reduce_dispatch<float16_t>(in, out, reduce_type_, axis_);
+        break;
+      case float32:
+        arg_reduce_dispatch<float>(in, out, reduce_type_, axis_);
+        break;
+      case bfloat16:
+        arg_reduce_dispatch<bfloat16_t>(in, out, reduce_type_, axis_);
+        break;
+      case float64:
+        arg_reduce_dispatch<double>(in, out, reduce_type_, axis_);
+        break;
+      case complex64:
+        arg_reduce_dispatch<complex64_t>(in, out, reduce_type_, axis_);
+        break;
+    }
+  });
 }
 
 } // namespace mlx::core

mlx/backend/cpu/available.cpp

@@ -0,0 +1,11 @@
+// Copyright © 2025 Apple Inc.
+
+#include "mlx/backend/cpu/available.h"
+
+namespace mlx::core::cpu {
+
+bool is_available() {
+  return true;
+}
+
+} // namespace mlx::core::cpu

mlx/backend/cpu/available.h

@@ -0,0 +1,9 @@
+// Copyright © 2025 Apple Inc.
+
+#pragma once
+
+namespace mlx::core::cpu {
+
+bool is_available();
+
+} // namespace mlx::core::cpu

mlx/backend/cpu/binary.cpp

@@ -8,6 +8,7 @@
 #include "mlx/backend/cpu/binary.h"
 #include "mlx/backend/cpu/binary_ops.h"
 #include "mlx/backend/cpu/binary_two.h"
+#include "mlx/backend/cpu/encoder.h"
 #include "mlx/primitives.h"
 #include "mlx/utils.h"
 
@@ -16,51 +17,221 @@ namespace mlx::core {
 namespace {
 
 template <typename Op>
-void comparison_op(const array& a, const array& b, array& out, Op op) {
-  switch (a.dtype()) {
-    case bool_:
-      binary_op<bool, bool>(a, b, out, op);
-      break;
-    case uint8:
-      binary_op<uint8_t, bool>(a, b, out, op);
-      break;
-    case uint16:
-      binary_op<uint16_t, bool>(a, b, out, op);
-      break;
-    case uint32:
-      binary_op<uint32_t, bool>(a, b, out, op);
-      break;
-    case uint64:
-      binary_op<uint64_t, bool>(a, b, out, op);
-      break;
-    case int8:
-      binary_op<int8_t, bool>(a, b, out, op);
-      break;
-    case int16:
-      binary_op<int16_t, bool>(a, b, out, op);
-      break;
-    case int32:
-      binary_op<int32_t, bool>(a, b, out, op);
-      break;
-    case int64:
-      binary_op<int64_t, bool>(a, b, out, op);
-      break;
-    case float16:
-      binary_op<float16_t, bool>(a, b, out, op);
-      break;
-    case float32:
-      binary_op<float, bool>(a, b, out, op);
-      break;
-    case float64:
-      binary_op<double, bool>(a, b, out, op);
-      break;
-    case bfloat16:
-      binary_op<bfloat16_t, bool>(a, b, out, op);
-      break;
-    case complex64:
-      binary_op<complex64_t, bool>(a, b, out, op);
-      break;
-  }
+void binary(const array& a, const array& b, array& out, Op op, Stream stream) {
+  auto bopt = get_binary_op_type(a, b);
+  set_binary_op_output_data(a, b, out, bopt);
+
+  auto& encoder = cpu::get_command_encoder(stream);
+  encoder.set_input_array(a);
+  encoder.set_input_array(b);
+  encoder.set_output_array(out);
+  encoder.dispatch([a = array::unsafe_weak_copy(a),
+                    b = array::unsafe_weak_copy(b),
+                    out = array::unsafe_weak_copy(out),
+                    bopt]() mutable {
+    switch (out.dtype()) {
+      case bool_:
+        binary_op<bool, Op>(a, b, out, bopt);
+        break;
+      case uint8:
+        binary_op<uint8_t, Op>(a, b, out, bopt);
+        break;
+      case uint16:
+        binary_op<uint16_t, Op>(a, b, out, bopt);
+        break;
+      case uint32:
+        binary_op<uint32_t, Op>(a, b, out, bopt);
+        break;
+      case uint64:
+        binary_op<uint64_t, Op>(a, b, out, bopt);
+        break;
+      case int8:
+        binary_op<int8_t, Op>(a, b, out, bopt);
+        break;
+      case int16:
+        binary_op<int16_t, Op>(a, b, out, bopt);
+        break;
+      case int32:
+        binary_op<int32_t, Op>(a, b, out, bopt);
+        break;
+      case int64:
+        binary_op<int64_t, Op>(a, b, out, bopt);
+        break;
+      case float16:
+        binary_op<float16_t, Op>(a, b, out, bopt);
+        break;
+      case float32:
+        binary_op<float, Op>(a, b, out, bopt);
+        break;
+      case float64:
+        binary_op<double, Op>(a, b, out, bopt);
+        break;
+      case bfloat16:
+        binary_op<bfloat16_t, Op>(a, b, out, bopt);
+        break;
+      case complex64:
+        binary_op<complex64_t, Op>(a, b, out, bopt);
+        break;
+    }
+  });
+}
+
+template <typename Op>
+void comparison_op(
+    const array& a,
+    const array& b,
+    array& out,
+    Op op,
+    Stream stream) {
+  auto bopt = get_binary_op_type(a, b);
+  set_binary_op_output_data(a, b, out, bopt);
+
+  auto& encoder = cpu::get_command_encoder(stream);
+  encoder.set_input_array(a);
+  encoder.set_input_array(b);
+  encoder.set_output_array(out);
+  encoder.dispatch([a = array::unsafe_weak_copy(a),
+                    b = array::unsafe_weak_copy(b),
+                    out = array::unsafe_weak_copy(out),
+                    bopt]() mutable {
+    switch (a.dtype()) {
+      case bool_:
+        binary_op<bool, bool, Op>(a, b, out, bopt);
+        break;
+      case uint8:
+        binary_op<uint8_t, bool, Op>(a, b, out, bopt);
+        break;
+      case uint16:
+        binary_op<uint16_t, bool, Op>(a, b, out, bopt);
+        break;
+      case uint32:
+        binary_op<uint32_t, bool, Op>(a, b, out, bopt);
+        break;
+      case uint64:
+        binary_op<uint64_t, bool, Op>(a, b, out, bopt);
+        break;
+      case int8:
+        binary_op<int8_t, bool, Op>(a, b, out, bopt);
+        break;
+      case int16:
+        binary_op<int16_t, bool, Op>(a, b, out, bopt);
+        break;
+      case int32:
+        binary_op<int32_t, bool, Op>(a, b, out, bopt);
+        break;
+      case int64:
+        binary_op<int64_t, bool, Op>(a, b, out, bopt);
+        break;
+      case float16:
+        binary_op<float16_t, bool, Op>(a, b, out, bopt);
+        break;
+      case float32:
+        binary_op<float, bool, Op>(a, b, out, bopt);
+        break;
+      case float64:
+        binary_op<double, bool, Op>(a, b, out, bopt);
+        break;
+      case bfloat16:
+        binary_op<bfloat16_t, bool, Op>(a, b, out, bopt);
+        break;
+      case complex64:
+        binary_op<complex64_t, bool, Op>(a, b, out, bopt);
+        break;
+    }
+  });
+}
+
+template <typename Op>
+void binary_float(
+    const array& a,
+    const array& b,
+    array& out,
+    Op op,
+    Stream stream) {
+  auto bopt = get_binary_op_type(a, b);
+  set_binary_op_output_data(a, b, out, bopt);
+
+  auto& encoder = cpu::get_command_encoder(stream);
+  encoder.set_input_array(a);
+  encoder.set_input_array(b);
+  encoder.set_output_array(out);
+  encoder.dispatch([a = array::unsafe_weak_copy(a),
+                    b = array::unsafe_weak_copy(b),
+                    out = array::unsafe_weak_copy(out),
+                    bopt]() mutable {
+    switch (out.dtype()) {
+      case float16:
+        binary_op<float16_t, Op>(a, b, out, bopt);
+        break;
+      case float32:
+        binary_op<float, Op>(a, b, out, bopt);
+        break;
+      case float64:
+        binary_op<double, Op>(a, b, out, bopt);
+        break;
+      case bfloat16:
+        binary_op<bfloat16_t, Op>(a, b, out, bopt);
+        break;
+      case complex64:
+        binary_op<complex64_t, Op>(a, b, out, bopt);
+        break;
+      default:
+        throw std::runtime_error(
+            "[binary_float] Only supports floating point types.");
+    }
+  });
+}
+
+template <typename Op>
+void binary_int(
+    const array& a,
+    const array& b,
+    array& out,
+    Op op,
+    Stream stream) {
+  auto bopt = get_binary_op_type(a, b);
+  set_binary_op_output_data(a, b, out, bopt);
+
+  auto& encoder = cpu::get_command_encoder(stream);
+  encoder.set_input_array(a);
+  encoder.set_input_array(b);
+  encoder.set_output_array(out);
+  encoder.dispatch([a = array::unsafe_weak_copy(a),
+                    b = array::unsafe_weak_copy(b),
+                    out = array::unsafe_weak_copy(out),
+                    bopt]() mutable {
+    switch (out.dtype()) {
+      case bool_:
+        binary_op<bool, Op>(a, b, out, bopt);
+      case uint8:
+        binary_op<uint8_t, Op>(a, b, out, bopt);
+        break;
+      case uint16:
+        binary_op<uint16_t, Op>(a, b, out, bopt);
+        break;
+      case uint32:
+        binary_op<uint32_t, Op>(a, b, out, bopt);
+        break;
+      case uint64:
+        binary_op<uint64_t, Op>(a, b, out, bopt);
+        break;
+      case int8:
+        binary_op<int8_t, Op>(a, b, out, bopt);
+        break;
+      case int16:
+        binary_op<int16_t, Op>(a, b, out, bopt);
+        break;
+      case int32:
+        binary_op<int32_t, Op>(a, b, out, bopt);
+        break;
+      case int64:
+        binary_op<int64_t, Op>(a, b, out, bopt);
+        break;
+      default:
+        throw std::runtime_error("[binary_int] Type not supported");
+        break;
+    }
+  });
 }
 
 } // namespace
@@ -69,7 +240,7 @@ void Add::eval_cpu(const std::vector<array>& inputs, array& out) {
   assert(inputs.size() == 2);
   auto& a = inputs[0];
   auto& b = inputs[1];
-  binary(a, b, out, detail::Add());
+  binary(a, b, out, detail::Add(), stream());
 }
 
 void DivMod::eval_cpu(
@@ -78,70 +249,89 @@ void DivMod::eval_cpu(
   assert(inputs.size() == 2);
   auto& a = inputs[0];
   auto& b = inputs[1];
-  auto integral_op = [](auto x, auto y) {
-    return std::make_pair(x / y, x % y);
-  };
-  auto float_op = [](auto x, auto y) {
-    return std::make_pair(std::trunc(x / y), std::fmod(x, y));
-  };
-  switch (outputs[0].dtype()) {
-    case bool_:
-      binary_op<bool>(a, b, outputs, integral_op);
-    case uint8:
-      binary_op<uint8_t>(a, b, outputs, integral_op);
-      break;
-    case uint16:
-      binary_op<uint16_t>(a, b, outputs, integral_op);
-      break;
-    case uint32:
-      binary_op<uint32_t>(a, b, outputs, integral_op);
-      break;
-    case uint64:
-      binary_op<uint64_t>(a, b, outputs, integral_op);
-      break;
-    case int8:
-      binary_op<int8_t>(a, b, outputs, integral_op);
-      break;
-    case int16:
-      binary_op<int16_t>(a, b, outputs, integral_op);
-      break;
-    case int32:
-      binary_op<int32_t>(a, b, outputs, integral_op);
-      break;
-    case int64:
-      binary_op<int64_t>(a, b, outputs, integral_op);
-      break;
-    case float16:
-      binary_op<float16_t>(a, b, outputs, float_op);
-      break;
-    case float32:
-      binary_op<float>(a, b, outputs, float_op);
-      break;
-    case float64:
-      binary_op<double>(a, b, outputs, float_op);
-      break;
-    case bfloat16:
-      binary_op<bfloat16_t>(a, b, outputs, float_op);
-      break;
-    case complex64:
-      // Should never get here
-      throw std::runtime_error("[DivMod] Complex type not supported");
-      break;
-  }
+  auto bopt = get_binary_op_type(a, b);
+  auto& out_a = outputs[0];
+  auto& out_b = outputs[1];
+  set_binary_op_output_data(a, b, out_a, bopt);
+  set_binary_op_output_data(a, b, out_b, bopt);
+
+  auto& encoder = cpu::get_command_encoder(stream());
+  encoder.set_input_array(a);
+  encoder.set_input_array(b);
+  encoder.set_output_array(out_a);
+  encoder.set_output_array(out_b);
+
+  encoder.dispatch([a = array::unsafe_weak_copy(a),
+                    b = array::unsafe_weak_copy(b),
+                    out_a = array::unsafe_weak_copy(out_a),
+                    out_b = array::unsafe_weak_copy(out_b),
+                    bopt]() mutable {
+    auto integral_op = [](auto x, auto y) {
+      return std::make_pair(x / y, x % y);
+    };
+    auto float_op = [](auto x, auto y) {
+      return std::make_pair(std::trunc(x / y), std::fmod(x, y));
+    };
+
+    switch (out_a.dtype()) {
+      case bool_:
+        binary_op<bool>(a, b, out_a, out_b, integral_op, bopt);
+      case uint8:
+        binary_op<uint8_t>(a, b, out_a, out_b, integral_op, bopt);
+        break;
+      case uint16:
+        binary_op<uint16_t>(a, b, out_a, out_b, integral_op, bopt);
+        break;
+      case uint32:
+        binary_op<uint32_t>(a, b, out_a, out_b, integral_op, bopt);
+        break;
+      case uint64:
+        binary_op<uint64_t>(a, b, out_a, out_b, integral_op, bopt);
+        break;
+      case int8:
+        binary_op<int8_t>(a, b, out_a, out_b, integral_op, bopt);
+        break;
+      case int16:
+        binary_op<int16_t>(a, b, out_a, out_b, integral_op, bopt);
+        break;
+      case int32:
+        binary_op<int32_t>(a, b, out_a, out_b, integral_op, bopt);
+        break;
+      case int64:
+        binary_op<int64_t>(a, b, out_a, out_b, integral_op, bopt);
+        break;
+      case float16:
+        binary_op<float16_t>(a, b, out_a, out_b, float_op, bopt);
+        break;
+      case float32:
+        binary_op<float>(a, b, out_a, out_b, float_op, bopt);
+        break;
+      case float64:
+        binary_op<double>(a, b, out_a, out_b, float_op, bopt);
+        break;
+      case bfloat16:
+        binary_op<bfloat16_t>(a, b, out_a, out_b, float_op, bopt);
+        break;
+      case complex64:
+        // Should never get here
+        throw std::runtime_error("[DivMod] Complex type not supported");
+        break;
+    }
+  });
 }
 
 void Divide::eval_cpu(const std::vector<array>& inputs, array& out) {
   assert(inputs.size() == 2);
   auto& a = inputs[0];
   auto& b = inputs[1];
-  binary(a, b, out, detail::Divide());
+  binary(a, b, out, detail::Divide(), stream());
 }
 
 void Remainder::eval_cpu(const std::vector<array>& inputs, array& out) {
   assert(inputs.size() == 2);
   auto& a = inputs[0];
   auto& b = inputs[1];
-  binary(a, b, out, detail::Remainder());
+  binary(a, b, out, detail::Remainder(), stream());
 }
 
 void Equal::eval_cpu(const std::vector<array>& inputs, array& out) {
@@ -149,181 +339,143 @@ void Equal::eval_cpu(const std::vector<array>& inputs, array& out) {
   auto& a = inputs[0];
   auto& b = inputs[1];
   if (equal_nan_) {
-    switch (a.dtype()) {
-      case float16:
-        binary_op<float16_t, bool>(a, b, out, detail::NaNEqual());
-        break;
-      case float32:
-        binary_op<float, bool>(a, b, out, detail::NaNEqual());
-        break;
-      case float64:
-        binary_op<double, bool>(a, b, out, detail::NaNEqual());
-        break;
-      case bfloat16:
-        binary_op<bfloat16_t, bool>(a, b, out, detail::NaNEqual());
-        break;
-      case complex64:
-        binary_op<complex64_t, bool>(a, b, out, detail::NaNEqual());
-        break;
-      default:
-        throw std::runtime_error(
-            "[NanEqual::eval_cpu] Only for floating point types.");
-    }
+    auto bopt = get_binary_op_type(a, b);
+    set_binary_op_output_data(a, b, out, bopt);
+
+    auto& encoder = cpu::get_command_encoder(stream());
+    encoder.set_input_array(a);
+    encoder.set_input_array(b);
+    encoder.set_output_array(out);
+    encoder.dispatch([a = array::unsafe_weak_copy(a),
+                      b = array::unsafe_weak_copy(b),
+                      out = array::unsafe_weak_copy(out),
+                      bopt]() mutable {
+      switch (a.dtype()) {
+        case float16:
+          binary_op<float16_t, bool, detail::NaNEqual>(a, b, out, bopt);
+          break;
+        case float32:
+          binary_op<float, bool, detail::NaNEqual>(a, b, out, bopt);
+          break;
+        case float64:
+          binary_op<double, bool, detail::NaNEqual>(a, b, out, bopt);
+          break;
+        case bfloat16:
+          binary_op<bfloat16_t, bool, detail::NaNEqual>(a, b, out, bopt);
+          break;
+        case complex64:
+          binary_op<complex64_t, bool, detail::NaNEqual>(a, b, out, bopt);
+          break;
+        default:
+          throw std::runtime_error(
+              "[NanEqual::eval_cpu] Only for floating point types.");
+      }
+    });
   } else {
-    comparison_op(a, b, out, detail::Equal());
+    comparison_op(a, b, out, detail::Equal(), stream());
   }
 }
 
 void Greater::eval_cpu(const std::vector<array>& inputs, array& out) {
   assert(inputs.size() == 2);
-  comparison_op(inputs[0], inputs[1], out, detail::Greater());
+  comparison_op(inputs[0], inputs[1], out, detail::Greater(), stream());
 }
 
 void GreaterEqual::eval_cpu(const std::vector<array>& inputs, array& out) {
   assert(inputs.size() == 2);
-  comparison_op(inputs[0], inputs[1], out, detail::GreaterEqual());
+  comparison_op(inputs[0], inputs[1], out, detail::GreaterEqual(), stream());
 }
 
 void Less::eval_cpu(const std::vector<array>& inputs, array& out) {
   assert(inputs.size() == 2);
-  comparison_op(inputs[0], inputs[1], out, detail::Less());
+  comparison_op(inputs[0], inputs[1], out, detail::Less(), stream());
 }
 
 void LessEqual::eval_cpu(const std::vector<array>& inputs, array& out) {
   assert(inputs.size() == 2);
-  comparison_op(inputs[0], inputs[1], out, detail::LessEqual());
+  comparison_op(inputs[0], inputs[1], out, detail::LessEqual(), stream());
 }
 
 void LogAddExp::eval_cpu(const std::vector<array>& inputs, array& out) {
   assert(inputs.size() == 2);
   auto& a = inputs[0];
   auto& b = inputs[1];
-  switch (out.dtype()) {
-    case float16:
-      binary_op<float16_t>(a, b, out, detail::LogAddExp());
-      break;
-    case float32:
-      binary_op<float>(a, b, out, detail::LogAddExp());
-      break;
-    case float64:
-      binary_op<double>(a, b, out, detail::LogAddExp());
-      break;
-    case bfloat16:
-      binary_op<bfloat16_t>(a, b, out, detail::LogAddExp());
-      break;
-    default:
-      throw std::runtime_error(
-          "[LogAddExp::eval_cpu] Only supports non-complex floating point types.");
-  }
+  binary_float(a, b, out, detail::LogAddExp(), stream());
 }
 
 void LogicalAnd::eval_cpu(const std::vector<array>& inputs, array& out) {
   assert(inputs.size() == 2); // LogicalAnd requires two input arrays
   auto& in1 = inputs[0];
   auto& in2 = inputs[1];
-  binary(in1, in2, out, detail::LogicalAnd());
+  binary(in1, in2, out, detail::LogicalAnd(), stream());
 }
 
 void LogicalOr::eval_cpu(const std::vector<array>& inputs, array& out) {
   assert(inputs.size() == 2); // LogicalOr requires two input arrays
   auto& in1 = inputs[0];
   auto& in2 = inputs[1];
-  binary(in1, in2, out, detail::LogicalOr());
+  binary(in1, in2, out, detail::LogicalOr(), stream());
 }
 
 void Maximum::eval_cpu(const std::vector<array>& inputs, array& out) {
   assert(inputs.size() == 2);
   auto& a = inputs[0];
   auto& b = inputs[1];
-  binary(a, b, out, detail::Maximum());
+  binary(a, b, out, detail::Maximum(), stream());
 }
 
 void Minimum::eval_cpu(const std::vector<array>& inputs, array& out) {
   assert(inputs.size() == 2);
   auto& a = inputs[0];
   auto& b = inputs[1];
-  binary(a, b, out, detail::Minimum());
+  binary(a, b, out, detail::Minimum(), stream());
 }
 
 void Multiply::eval_cpu(const std::vector<array>& inputs, array& out) {
   assert(inputs.size() == 2);
   auto& a = inputs[0];
   auto& b = inputs[1];
-  binary(a, b, out, detail::Multiply());
+  binary(a, b, out, detail::Multiply(), stream());
 }
 
 void NotEqual::eval_cpu(const std::vector<array>& inputs, array& out) {
   assert(inputs.size() == 2);
-  comparison_op(inputs[0], inputs[1], out, detail::NotEqual());
+  comparison_op(inputs[0], inputs[1], out, detail::NotEqual(), stream());
 }
 
 void Power::eval_cpu(const std::vector<array>& inputs, array& out) {
   assert(inputs.size() == 2);
   auto& a = inputs[0];
   auto& b = inputs[1];
-  binary(a, b, out, detail::Power());
+  binary(a, b, out, detail::Power(), stream());
 }
 
 void Subtract::eval_cpu(const std::vector<array>& inputs, array& out) {
   assert(inputs.size() == 2);
   auto& a = inputs[0];
   auto& b = inputs[1];
-  binary(a, b, out, detail::Subtract());
+  binary(a, b, out, detail::Subtract(), stream());
 }
 
 void BitwiseBinary::eval_cpu(const std::vector<array>& inputs, array& out) {
   assert(inputs.size() == 2);
   auto& a = inputs[0];
   auto& b = inputs[1];
-  auto dispatch_type = [&a, &b, &out](auto op) {
-    switch (out.dtype()) {
-      case bool_:
-        binary_op<bool>(a, b, out, op);
-      case uint8:
-        binary_op<uint8_t>(a, b, out, op);
-        break;
-      case uint16:
-        binary_op<uint16_t>(a, b, out, op);
-        break;
-      case uint32:
-        binary_op<uint32_t>(a, b, out, op);
-        break;
-      case uint64:
-        binary_op<uint64_t>(a, b, out, op);
-        break;
-      case int8:
-        binary_op<int8_t>(a, b, out, op);
-        break;
-      case int16:
-        binary_op<int16_t>(a, b, out, op);
-        break;
-      case int32:
-        binary_op<int32_t>(a, b, out, op);
-        break;
-      case int64:
-        binary_op<int64_t>(a, b, out, op);
-        break;
-      default:
-        throw std::runtime_error(
-            "[BitwiseBinary::eval_cpu] Type not supported");
-        break;
-    }
-  };
   switch (op_) {
     case BitwiseBinary::And:
-      dispatch_type(detail::BitwiseAnd());
+      binary_int(a, b, out, detail::BitwiseAnd(), stream());
       break;
     case BitwiseBinary::Or:
-      dispatch_type(detail::BitwiseOr());
+      binary_int(a, b, out, detail::BitwiseOr(), stream());
       break;
     case BitwiseBinary::Xor:
-      dispatch_type(detail::BitwiseXor());
+      binary_int(a, b, out, detail::BitwiseXor(), stream());
       break;
     case BitwiseBinary::LeftShift:
-      dispatch_type(detail::LeftShift());
+      binary_int(a, b, out, detail::LeftShift(), stream());
       break;
     case BitwiseBinary::RightShift:
-      dispatch_type(detail::RightShift());
+      binary_int(a, b, out, detail::RightShift(), stream());
       break;
   }
 }
@@ -332,23 +484,7 @@ void ArcTan2::eval_cpu(const std::vector<array>& inputs, array& out) {
   assert(inputs.size() == 2);
   const auto& a = inputs[0];
   const auto& b = inputs[1];
-  switch (out.dtype()) {
-    case float16:
-      binary_op<float16_t>(a, b, out, detail::ArcTan2());
-      break;
-    case float32:
-      binary_op<float>(a, b, out, detail::ArcTan2());
-      break;
-    case float64:
-      binary_op<double>(a, b, out, detail::ArcTan2());
-      break;
-    case bfloat16:
-      binary_op<bfloat16_t>(a, b, out, detail::ArcTan2());
-      break;
-    default:
-      throw std::runtime_error(
-          "[ArcTan2::eval_cpu] Only supports non-complex floating point types.");
-  }
+  binary_float(a, b, out, detail::ArcTan2(), stream());
 }
 
 } // namespace mlx::core

mlx/backend/cpu/binary.h

@@ -3,7 +3,6 @@
 #pragma once
 #include <cassert>
 
-#include "mlx/allocator.h"
 #include "mlx/array.h"
 #include "mlx/backend/common/binary.h"
 #include "mlx/backend/common/utils.h"
@@ -14,22 +13,18 @@ namespace mlx::core {
 
 template <typename Op>
 struct VectorScalar {
-  Op op;
-
-  VectorScalar(Op op_) : op(op_) {}
-
   template <typename T, typename U>
   void operator()(const T* a, const T* b, U* dst, int size) {
     T scalar = *b;
     constexpr int N = simd::max_size<T>;
     while (size >= N) {
-      simd::store(dst, op(simd::load<T, N>(a), simd::Simd<T, N>(scalar)));
+      simd::store(dst, Op{}(simd::load<T, N>(a), simd::Simd<T, N>(scalar)));
       dst += N;
       a += N;
       size -= N;
     }
     while (size-- > 0) {
-      *dst = op(*a, scalar);
+      *dst = Op{}(*a, scalar);
       dst++;
       a++;
     }
@@ -38,22 +33,18 @@ struct VectorScalar {
 
 template <typename Op>
 struct ScalarVector {
-  Op op;
-
-  ScalarVector(Op op_) : op(op_) {}
-
   template <typename T, typename U>
   void operator()(const T* a, const T* b, U* dst, int size) {
     T scalar = *a;
     constexpr int N = simd::max_size<T>;
     while (size >= N) {
-      simd::store(dst, op(simd::Simd<T, N>(scalar), simd::load<T, N>(b)));
+      simd::store(dst, Op{}(simd::Simd<T, N>(scalar), simd::load<T, N>(b)));
       dst += N;
       b += N;
       size -= N;
     }
     while (size-- > 0) {
-      *dst = op(scalar, *b);
+      *dst = Op{}(scalar, *b);
       dst++;
       b++;
     }
@@ -62,22 +53,18 @@ struct ScalarVector {
 
 template <typename Op>
 struct VectorVector {
-  Op op;
-
-  VectorVector(Op op_) : op(op_) {}
-
   template <typename T, typename U>
   void operator()(const T* a, const T* b, U* dst, int size) {
     constexpr int N = simd::max_size<T>;
     while (size >= N) {
-      simd::store(dst, op(simd::load<T, N>(a), simd::load<T, N>(b)));
+      simd::store(dst, Op{}(simd::load<T, N>(a), simd::load<T, N>(b)));
       dst += N;
       a += N;
       b += N;
       size -= N;
     }
     while (size-- > 0) {
-      *dst = op(*a, *b);
+      *dst = Op{}(*a, *b);
       dst++;
       a++;
       b++;
@@ -90,7 +77,6 @@ void binary_op_dims(
     const T* a,
     const T* b,
     U* out,
-    Op op,
     const Shape& shape,
     const Strides& a_strides,
     const Strides& b_strides,
@@ -104,12 +90,12 @@ void binary_op_dims(
   for (int i = 0; i < N; i++) {
     if constexpr (D > 1) {
       binary_op_dims<T, U, Op, D - 1, Strided>(
-          a, b, out, op, shape, a_strides, b_strides, out_strides, axis + 1);
+          a, b, out, shape, a_strides, b_strides, out_strides, axis + 1);
     } else {
       if constexpr (Strided) {
-        op(a, b, out, stride_out);
+        Op{}(a, b, out, stride_out);
       } else {
-        *out = op(*a, *b);
+        *out = Op{}(*a, *b);
       }
     }
     out += stride_out;
@@ -120,66 +106,38 @@ void binary_op_dims(
 
 template <typename T, typename U, bool Strided, typename Op>
 void binary_op_dispatch_dims(
-    const array& a,
-    const array& b,
-    array& out,
-    Op op,
+    const T* a,
+    const T* b,
+    U* out,
     int dim,
+    int size,
     const Shape& shape,
     const Strides& a_strides,
     const Strides& b_strides,
     const Strides& out_strides) {
-  const T* a_ptr = a.data<T>();
-  const T* b_ptr = b.data<T>();
-  U* out_ptr = out.data<U>();
   switch (dim) {
     case 1:
       binary_op_dims<T, U, Op, 1, Strided>(
-          a_ptr,
-          b_ptr,
-          out_ptr,
-          op,
-          shape,
-          a_strides,
-          b_strides,
-          out_strides,
-          0);
+          a, b, out, shape, a_strides, b_strides, out_strides, 0);
       return;
     case 2:
       binary_op_dims<T, U, Op, 2, Strided>(
-          a_ptr,
-          b_ptr,
-          out_ptr,
-          op,
-          shape,
-          a_strides,
-          b_strides,
-          out_strides,
-          0);
+          a, b, out, shape, a_strides, b_strides, out_strides, 0);
       return;
     case 3:
       binary_op_dims<T, U, Op, 3, Strided>(
-          a_ptr,
-          b_ptr,
-          out_ptr,
-          op,
-          shape,
-          a_strides,
-          b_strides,
-          out_strides,
-          0);
+          a, b, out, shape, a_strides, b_strides, out_strides, 0);
       return;
   }
 
   ContiguousIterator a_it(shape, a_strides, dim - 3);
   ContiguousIterator b_it(shape, b_strides, dim - 3);
   auto stride = out_strides[dim - 4];
-  for (int64_t elem = 0; elem < a.size(); elem += stride) {
+  for (int64_t elem = 0; elem < size; elem += stride) {
     binary_op_dims<T, U, Op, 3, Strided>(
-        a_ptr + a_it.loc,
-        b_ptr + b_it.loc,
-        out_ptr + elem,
-        op,
+        a + a_it.loc,
+        b + b_it.loc,
+        out + elem,
         shape,
         a_strides,
         b_strides,
@@ -191,40 +149,41 @@ void binary_op_dispatch_dims(
 }
 
 template <typename T, typename U, typename Op>
-void binary_op(const array& a, const array& b, array& out, Op op) {
-  auto bopt = get_binary_op_type(a, b);
-  set_binary_op_output_data(a, b, out, bopt);
-
+void binary_op(const array& a, const array& b, array& out, BinaryOpType bopt) {
   // The full computation is scalar scalar so call the base op once
+  auto a_ptr = a.data<T>();
+  auto b_ptr = b.data<T>();
+
+  auto out_ptr = out.data<U>();
   if (bopt == BinaryOpType::ScalarScalar) {
-    *(out.data<U>()) = op(*a.data<T>(), *b.data<T>());
+    *out_ptr = Op{}(*a_ptr, *b_ptr);
     return;
   }
 
   // The full computation is scalar vector so delegate to the op
   if (bopt == BinaryOpType::ScalarVector) {
-    ScalarVector{op}(a.data<T>(), b.data<T>(), out.data<U>(), b.data_size());
+    ScalarVector<Op>{}(a_ptr, b_ptr, out_ptr, b.data_size());
     return;
   }
 
   // The full computation is vector scalar so delegate to the op
   if (bopt == BinaryOpType::VectorScalar) {
-    VectorScalar{op}(a.data<T>(), b.data<T>(), out.data<U>(), a.data_size());
+    VectorScalar<Op>{}(a_ptr, b_ptr, out_ptr, a.data_size());
     return;
   }
 
   // The full computation is vector vector so delegate to the op
   if (bopt == BinaryOpType::VectorVector) {
-    VectorVector{op}(a.data<T>(), b.data<T>(), out.data<U>(), out.size());
+    VectorVector<Op>{}(a_ptr, b_ptr, out_ptr, a.size());
     return;
   }
 
   // General computation so let's try to optimize
   auto [new_shape, new_strides] = collapse_contiguous_dims(
       a.shape(), {a.strides(), b.strides(), out.strides()});
-  const auto& a_strides = new_strides[0];
-  const auto& b_strides = new_strides[1];
-  const auto& strides = new_strides[2];
+  auto& a_strides = new_strides[0];
+  auto& b_strides = new_strides[1];
+  auto& strides = new_strides[2];
 
   // Get the left-most dim such that the array is row contiguous after
   auto leftmost_rc_dim = [&strides](const auto& arr_strides) {
@@ -248,7 +207,8 @@ void binary_op(const array& a, const array& b, array& out, Op op) {
 
   auto ndim = new_shape.size();
 
-  // Case 1: LxM and FxM where L and F are broadcastable and M is row contiguous
+  // Case 1: LxM and FxM where L and F are broadcastable and M is row
+  // contiguous
   int dim = ndim;
   if (int d = std::max(a_rc_dim, b_rc_dim); d < ndim) {
     bopt = BinaryOpType::VectorVector;
@@ -275,99 +235,59 @@ void binary_op(const array& a, const array& b, array& out, Op op) {
 
   switch (bopt) {
     case BinaryOpType::VectorVector:
-      binary_op_dispatch_dims<T, U, true>(
-          a,
-          b,
-          out,
-          VectorVector{op},
+      binary_op_dispatch_dims<T, U, true, VectorVector<Op>>(
+          a_ptr,
+          b_ptr,
+          out_ptr,
           dim,
+          a.size(),
           new_shape,
           a_strides,
           b_strides,
           strides);
       break;
     case BinaryOpType::VectorScalar:
-      binary_op_dispatch_dims<T, U, true>(
-          a,
-          b,
-          out,
-          VectorScalar{op},
+      binary_op_dispatch_dims<T, U, true, VectorScalar<Op>>(
+          a_ptr,
+          b_ptr,
+          out_ptr,
           dim,
+          a.size(),
           new_shape,
           a_strides,
           b_strides,
           strides);
       break;
     case BinaryOpType::ScalarVector:
-      binary_op_dispatch_dims<T, U, true>(
-          a,
-          b,
-          out,
-          ScalarVector{op},
+      binary_op_dispatch_dims<T, U, true, ScalarVector<Op>>(
+          a_ptr,
+          b_ptr,
+          out_ptr,
           dim,
+          a.size(),
           new_shape,
           a_strides,
           b_strides,
           strides);
       break;
     default:
-      binary_op_dispatch_dims<T, U, false>(
-          a, b, out, op, dim, new_shape, a_strides, b_strides, strides);
+      binary_op_dispatch_dims<T, U, false, Op>(
+          a_ptr,
+          b_ptr,
+          out_ptr,
+          dim,
+          a.size(),
+          new_shape,
+          a_strides,
+          b_strides,
+          strides);
       break;
   }
 }
 
 template <typename T, typename Op>
-void binary_op(const array& a, const array& b, array& out, Op op) {
-  binary_op<T, T>(a, b, out, op);
-}
-
-template <typename Op>
-void binary(const array& a, const array& b, array& out, Op op) {
-  switch (out.dtype()) {
-    case bool_:
-      binary_op<bool>(a, b, out, op);
-      break;
-    case uint8:
-      binary_op<uint8_t>(a, b, out, op);
-      break;
-    case uint16:
-      binary_op<uint16_t>(a, b, out, op);
-      break;
-    case uint32:
-      binary_op<uint32_t>(a, b, out, op);
-      break;
-    case uint64:
-      binary_op<uint64_t>(a, b, out, op);
-      break;
-    case int8:
-      binary_op<int8_t>(a, b, out, op);
-      break;
-    case int16:
-      binary_op<int16_t>(a, b, out, op);
-      break;
-    case int32:
-      binary_op<int32_t>(a, b, out, op);
-      break;
-    case int64:
-      binary_op<int64_t>(a, b, out, op);
-      break;
-    case float16:
-      binary_op<float16_t>(a, b, out, op);
-      break;
-    case float32:
-      binary_op<float>(a, b, out, op);
-      break;
-    case float64:
-      binary_op<double>(a, b, out, op);
-      break;
-    case bfloat16:
-      binary_op<bfloat16_t>(a, b, out, op);
-      break;
-    case complex64:
-      binary_op<complex64_t>(a, b, out, op);
-      break;
-  }
+void binary_op(const array& a, const array& b, array& out, BinaryOpType bopt) {
+  binary_op<T, T, Op>(a, b, out, bopt);
 }
 
 } // namespace mlx::core

mlx/backend/cpu/binary_two.h

@@ -58,14 +58,14 @@ void binary_op_dispatch_dims(
     Op op) {
   auto [shape, strides] = collapse_contiguous_dims(
       a.shape(), {a.strides(), b.strides(), out_a.strides()});
-  const auto& a_strides = strides[0];
-  const auto& b_strides = strides[1];
-  const auto& out_strides = strides[2];
   const T* a_ptr = a.data<T>();
   const T* b_ptr = b.data<T>();
   U* out_a_ptr = out_a.data<U>();
   U* out_b_ptr = out_b.data<U>();
 
+  const auto& a_strides = strides[0];
+  const auto& b_strides = strides[1];
+  const auto& out_strides = strides[2];
   int ndim = shape.size();
   switch (ndim) {
     case 1:
@@ -120,14 +120,10 @@ template <typename T, typename U = T, typename Op>
 void binary_op(
     const array& a,
     const array& b,
-    std::vector<array>& outputs,
-    Op op) {
-  auto bopt = get_binary_op_type(a, b);
-  auto& out_a = outputs[0];
-  auto& out_b = outputs[1];
-  set_binary_op_output_data(a, b, out_a, bopt);
-  set_binary_op_output_data(a, b, out_b, bopt);
-
+    array& out_a,
+    array& out_b,
+    Op op,
+    BinaryOpType bopt) {
   // The full computation is scalar scalar so call the base op once
   if (bopt == BinaryOpType::General) {
     binary_op_dispatch_dims<T, U, Op>(a, b, out_a, out_b, op);
@@ -141,14 +137,14 @@ void binary_op(
   if (bopt == BinaryOpType::ScalarScalar) {
     std::tie(*out_a_ptr, *out_b_ptr) = op(*a_ptr, *b_ptr);
   } else if (bopt == BinaryOpType::ScalarVector) {
-    for (size_t i = 0; i < b.size(); ++i) {
+    for (size_t i = 0; i < b.data_size(); ++i) {
       std::tie(*out_a_ptr, *out_b_ptr) = op(*a_ptr, *b_ptr);
       out_a_ptr++;
       out_b_ptr++;
       b_ptr++;
     }
   } else if (bopt == BinaryOpType::VectorScalar) {
-    for (size_t i = 0; i < a.size(); ++i) {
+    for (size_t i = 0; i < a.data_size(); ++i) {
       std::tie(*out_a_ptr, *out_b_ptr) = op(*a_ptr, *b_ptr);
       out_a_ptr++;
       out_b_ptr++;
@@ -165,58 +161,6 @@ void binary_op(
   }
 }
 
-template <typename Op>
-void binary(
-    const array& a,
-    const array& b,
-    std::vector<array>& outputs,
-    Op op) {
-  switch (outputs[0].dtype()) {
-    case bool_:
-      binary_op<bool>(a, b, outputs, op);
-      break;
-    case uint8:
-      binary_op<uint8_t>(a, b, outputs, op);
-      break;
-    case uint16:
-      binary_op<uint16_t>(a, b, outputs, op);
-      break;
-    case uint32:
-      binary_op<uint32_t>(a, b, outputs, op);
-      break;
-    case uint64:
-      binary_op<uint64_t>(a, b, outputs, op);
-      break;
-    case int8:
-      binary_op<int8_t>(a, b, outputs, op);
-      break;
-    case int16:
-      binary_op<int16_t>(a, b, outputs, op);
-      break;
-    case int32:
-      binary_op<int32_t>(a, b, outputs, op);
-      break;
-    case int64:
-      binary_op<int64_t>(a, b, outputs, op);
-      break;
-    case float16:
-      binary_op<float16_t>(a, b, outputs, op);
-      break;
-    case float32:
-      binary_op<float>(a, b, outputs, op);
-      break;
-    case float64:
-      binary_op<double>(a, b, outputs, op);
-      break;
-    case bfloat16:
-      binary_op<bfloat16_t>(a, b, outputs, op);
-      break;
-    case complex64:
-      binary_op<complex64_t>(a, b, outputs, op);
-      break;
-  }
-}
-
 } // namespace
 
 } // namespace mlx::core

mlx/backend/cpu/cholesky.cpp

@@ -2,13 +2,15 @@
 
 #include "mlx/allocator.h"
 #include "mlx/backend/cpu/copy.h"
+#include "mlx/backend/cpu/encoder.h"
 #include "mlx/backend/cpu/lapack.h"
 #include "mlx/linalg.h"
 #include "mlx/primitives.h"
 
 namespace mlx::core {
 
-void cholesky_impl(const array& a, array& factor, bool upper) {
+template <typename T>
+void cholesky_impl(const array& a, array& factor, bool upper, Stream stream) {
   // Lapack uses the column-major convention. We take advantage of the fact that
   // the matrix should be symmetric:
   //   (A)ᵀ = A
@@ -16,59 +18,68 @@ void cholesky_impl(const array& a, array& factor, bool upper) {
   // triangular matrix, so uplo is the opposite of what we would expect from
   // upper
 
-  char uplo = (upper) ? 'L' : 'U';
-
   // The decomposition is computed in place, so just copy the input to the
   // output.
-  copy(
+  copy_cpu(
       a,
       factor,
-      a.flags().row_contiguous ? CopyType::Vector : CopyType::General);
+      a.flags().row_contiguous ? CopyType::Vector : CopyType::General,
+      stream);
 
-  const int N = a.shape(-1);
-  const size_t num_matrices = a.size() / (N * N);
+  auto& encoder = cpu::get_command_encoder(stream);
+  encoder.set_output_array(factor);
+  encoder.dispatch([matrix = factor.data<T>(),
+                    upper,
+                    N = a.shape(-1),
+                    size = a.size()]() mutable {
+    char uplo = (upper) ? 'L' : 'U';
+    size_t num_matrices = size / (N * N);
+    for (int i = 0; i < num_matrices; i++) {
+      // Compute Cholesky factorization.
+      int info;
+      potrf<T>(
+          /* uplo = */ &uplo,
+          /* n = */ &N,
+          /* a = */ matrix,
+          /* lda = */ &N,
+          /* info = */ &info);
 
-  float* matrix = factor.data<float>();
-
-  for (int i = 0; i < num_matrices; i++) {
-    // Compute Cholesky factorization.
-    int info;
-    MLX_LAPACK_FUNC(spotrf)
-    (
-        /* uplo = */ &uplo,
-        /* n = */ &N,
-        /* a = */ matrix,
-        /* lda = */ &N,
-        /* info = */ &info);
-
-    // TODO: We do nothing when the matrix is not positive semi-definite
-    // because throwing an error would result in a crash. If we figure out how
-    // to catch errors from the implementation we should throw.
-    if (info < 0) {
-      std::stringstream msg;
-      msg << "[cholesky] Cholesky decomposition failed with error code "
-          << info;
-      throw std::runtime_error(msg.str());
-    }
-
-    // Zero out the upper/lower triangle while advancing the pointer to the
-    // next matrix at the same time.
-    for (int row = 0; row < N; row++) {
-      if (upper) {
-        std::fill(matrix, matrix + row, 0);
-      } else {
-        std::fill(matrix + row + 1, matrix + N, 0);
+      // TODO: We do nothing when the matrix is not positive semi-definite
+      // because throwing an error would result in a crash. If we figure out how
+      // to catch errors from the implementation we should throw.
+      if (info < 0) {
+        std::stringstream msg;
+        msg << "[Cholesky::eval_cpu] Cholesky decomposition failed with error code "
+            << info;
+        throw std::runtime_error(msg.str());
+      }
+
+      // Zero out the upper/lower triangle while advancing the pointer to the
+      // next matrix at the same time.
+      for (int row = 0; row < N; row++) {
+        if (upper) {
+          std::fill(matrix, matrix + row, 0);
+        } else {
+          std::fill(matrix + row + 1, matrix + N, 0);
+        }
+        matrix += N;
       }
-      matrix += N;
     }
-  }
+  });
 }
 
 void Cholesky::eval_cpu(const std::vector<array>& inputs, array& output) {
-  if (inputs[0].dtype() != float32) {
-    throw std::runtime_error("[Cholesky::eval] only supports float32.");
+  switch (inputs[0].dtype()) {
+    case float32:
+      cholesky_impl<float>(inputs[0], output, upper_, stream());
+      break;
+    case float64:
+      cholesky_impl<double>(inputs[0], output, upper_, stream());
+      break;
+    default:
+      throw std::runtime_error(
+          "[Cholesky::eval_cpu] only supports float32 or float64.");
   }
-  cholesky_impl(inputs[0], output, upper_);
 }
 
 } // namespace mlx::core

mlx/backend/cpu/compiled.cpp

@@ -11,9 +11,11 @@
 
 #include "mlx/backend/common/compiled.h"
 #include "mlx/backend/cpu/compiled_preamble.h"
+#include "mlx/backend/cpu/encoder.h"
 #include "mlx/backend/cpu/jit_compiler.h"
 #include "mlx/device.h"
 #include "mlx/graph_utils.h"
+#include "mlx/version.h"
 
 namespace mlx::core {
 
@@ -39,7 +41,10 @@ struct CompilerCache {
   std::shared_mutex mtx;
 };
 
-static CompilerCache cache{};
+static CompilerCache& cache() {
+  static CompilerCache cache_;
+  return cache_;
+};
 
 // GPU compile is always available if the GPU is available and since we are in
 // this file CPU compile is also available.
@@ -55,14 +60,16 @@ void* compile(
     const std::string& kernel_name,
     const std::function<std::string(void)>& source_builder) {
   {
-    std::shared_lock lock(cache.mtx);
-    if (auto it = cache.kernels.find(kernel_name); it != cache.kernels.end()) {
+    std::shared_lock lock(cache().mtx);
+    if (auto it = cache().kernels.find(kernel_name);
+        it != cache().kernels.end()) {
       return it->second;
     }
   }
 
-  std::unique_lock lock(cache.mtx);
-  if (auto it = cache.kernels.find(kernel_name); it != cache.kernels.end()) {
+  std::unique_lock lock(cache().mtx);
+  if (auto it = cache().kernels.find(kernel_name);
+      it != cache().kernels.end()) {
     return it->second;
   }
   std::string source_code = source_builder();
@@ -88,7 +95,11 @@ void* compile(
     kernel_file_name = kernel_name;
   }
 
-  auto output_dir = std::filesystem::temp_directory_path();
+  auto output_dir =
+      std::filesystem::temp_directory_path() / "mlx" / version() / "cpu";
+  if (!std::filesystem::exists(output_dir)) {
+    std::filesystem::create_directories(output_dir);
+  }
 
   std::string shared_lib_name = "lib" + kernel_file_name + ".so";
   auto shared_lib_path = (output_dir / shared_lib_name).string();
@@ -119,10 +130,10 @@ void* compile(
   }
 
   // load library
-  cache.libs.emplace_back(shared_lib_path);
+  cache().libs.emplace_back(shared_lib_path);
 
   // Load function
-  void* fun = dlsym(cache.libs.back().lib, kernel_name.c_str());
+  void* fun = dlsym(cache().libs.back().lib, kernel_name.c_str());
   if (!fun) {
     std::ostringstream msg;
     msg << "[Compile::eval_cpu] Failed to load compiled function "
@@ -130,7 +141,7 @@ void* compile(
         << dlerror();
     throw std::runtime_error(msg.str());
   }
-  cache.kernels.insert({kernel_name, fun});
+  cache().kernels.insert({kernel_name, fun});
   return fun;
 }
 
@@ -140,18 +151,9 @@ inline void build_kernel(
     const std::vector<array>& inputs,
     const std::vector<array>& outputs,
     const std::vector<array>& tape,
-    const std::unordered_set<uintptr_t>& constant_ids,
+    const std::function<bool(size_t)>& is_constant,
     bool contiguous,
     int ndim) {
-  // All outputs should have the exact same shape and will be row contiguous
-  auto output_shape = outputs[0].shape();
-  auto output_strides = outputs[0].strides();
-
-  // Constants are scalars that are captured by value and cannot change
-  auto is_constant = [&constant_ids](const array& x) {
-    return constant_ids.find(x.id()) != constant_ids.end();
-  };
-
   NodeNamer namer;
 
 #ifdef _MSC_VER
@@ -160,25 +162,28 @@ inline void build_kernel(
 #endif
 
   // Start the kernel
-  os << "void " << kernel_name << "(void** args) {" << std::endl;
+  os << "void " << kernel_name
+     << "(int* shape, int64_t** strides, void** args) {" << std::endl;
 
   // Add the input arguments
   int cnt = 0;
-  for (auto& x : inputs) {
-    auto& xname = namer.get_name(x);
-
+  int strides_index = 1;
+  for (size_t i = 0; i < inputs.size(); ++i) {
     // Skip constants from the input list
-    if (is_constant(x)) {
+    if (is_constant(i)) {
       continue;
     }
 
+    const auto& x = inputs[i];
+    auto& xname = namer.get_name(x);
+
     auto tstr = get_type_string(x.dtype());
     os << "  " << tstr << "* " << xname << " = (" << tstr << "*)args[" << cnt++
        << "];" << std::endl;
     // Scalars and contiguous need no strides
     if (!is_scalar(x) && !contiguous) {
-      os << "  const size_t* " << xname << "_strides = (size_t*)args[" << cnt++
-         << "];" << std::endl;
+      os << "  const int64_t* " << xname << "_strides = strides["
+         << strides_index++ << "];" << std::endl;
     }
   }
 
@@ -188,10 +193,8 @@ inline void build_kernel(
     os << "  " << tstr << "* " << namer.get_name(x) << " = (" << tstr
        << "*)args[" << cnt++ << "];" << std::endl;
   }
-  // Add output strides and shape to extract the indices.
-  if (!contiguous) {
-    os << "  const int* shape = (int*)args[" << cnt++ << "];" << std::endl;
-  } else {
+  // Add output size
+  if (contiguous) {
     os << "  const size_t size = (size_t)args[" << cnt++ << "];" << std::endl;
   }
 
@@ -205,10 +208,11 @@ inline void build_kernel(
   }
 
   // Read the inputs in tmps
-  for (auto& x : inputs) {
+  for (size_t i = 0; i < inputs.size(); ++i) {
+    const auto& x = inputs[i];
     auto& xname = namer.get_name(x);
 
-    if (is_constant(x)) {
+    if (is_constant(i)) {
       os << "  " << get_type_string(x.dtype()) << " tmp_" << xname << " = ";
       print_constant(os, x);
       os << ";" << std::endl;
@@ -232,7 +236,7 @@ inline void build_kernel(
       os << "static_cast<" << get_type_string(x.dtype()) << ">(tmp_"
          << namer.get_name(x.inputs()[0]) << ");" << std::endl;
     } else {
-      x.primitive().print(os);
+      os << x.primitive().name();
       os << "()(";
       for (int i = 0; i < x.inputs().size() - 1; i++) {
         os << "tmp_" << namer.get_name(x.inputs()[i]) << ", ";
@@ -258,8 +262,9 @@ inline void build_kernel(
   } else {
     for (int d = ndim - 1; d >= 0; --d) {
       // Update pointers
-      for (auto& x : inputs) {
-        if (is_constant(x) || is_scalar(x)) {
+      for (size_t i = 0; i < inputs.size(); ++i) {
+        const auto& x = inputs[i];
+        if (is_constant(i) || is_scalar(x)) {
           continue;
         }
         auto& xname = namer.get_name(x);
@@ -281,63 +286,33 @@ inline void build_kernel(
 void Compiled::eval_cpu(
     const std::vector<array>& inputs,
     std::vector<array>& outputs) {
-  if (kernel_lib_.empty()) {
-    kernel_lib_ = build_lib_name(inputs_, outputs_, tape_, constant_ids_);
-  }
+  auto& encoder = cpu::get_command_encoder(stream());
 
-  // Figure out which kernel we are using
-  auto& shape = outputs[0].shape();
-  auto contiguous = compiled_check_contiguity(inputs, shape);
+  // Collapse contiguous dims to route to a faster kernel if possible. Also
+  // handle all broadcasting.
+  auto [contiguous, shape, strides] =
+      compiled_collapse_contiguous_dims(inputs, outputs[0], is_constant_);
 
-  // Handle all broadcasting and collect function input arguments
+  // Collect function input arguments.
   std::vector<void*> args;
-  std::vector<std::vector<size_t>> strides;
-  for (int i = 0; i < inputs.size(); i++) {
-    // Skip constants.
-    if (constant_ids_.find(inputs_[i].id()) != constant_ids_.end()) {
+  for (size_t i = 0; i < inputs.size(); ++i) {
+    if (is_constant_(i)) {
       continue;
     }
-    auto& x = inputs[i];
+    const auto& x = inputs[i];
+    encoder.set_input_array(x);
     args.push_back((void*)x.data<void>());
-
-    if (contiguous || is_scalar(x)) {
-      continue;
-    }
-
-    // Broadcast the input to the output shape.
-    std::vector<size_t> xstrides;
-    int j = 0;
-    for (; j < shape.size() - x.ndim(); j++) {
-      if (shape[j] == 1) {
-        xstrides.push_back(outputs[0].strides()[j]);
-      } else {
-        xstrides.push_back(0);
-      }
-    }
-    for (int i = 0; i < x.ndim(); i++, j++) {
-      if (x.shape(i) == 1) {
-        if (shape[j] == 1) {
-          xstrides.push_back(outputs[0].strides()[j]);
-        } else {
-          xstrides.push_back(0);
-        }
-      } else {
-        xstrides.push_back(x.strides()[i]);
-      }
-    }
-    strides.push_back(std::move(xstrides));
-    args.push_back(strides.back().data());
   }
 
   // Get the kernel name from the lib
   int ndim = shape.size();
   auto kernel_name = kernel_lib_ + (contiguous ? "_contiguous" : "_strided_");
   if (!contiguous) {
-    kernel_name += std::to_string(shape.size());
+    kernel_name += std::to_string(ndim);
   }
 
   // Get the function
-  auto fn_ptr = compile(kernel_name, [&]() {
+  auto fn_ptr = compile(kernel_name, [&, contiguous = contiguous]() {
     std::ostringstream kernel;
     kernel << get_kernel_preamble() << std::endl;
     kernel << "extern \"C\"  {" << std::endl;
@@ -347,7 +322,7 @@ void Compiled::eval_cpu(
         inputs_,
         outputs_,
         tape_,
-        constant_ids_,
+        is_constant_,
         contiguous,
         ndim);
     // Close extern "C"
@@ -355,19 +330,26 @@ void Compiled::eval_cpu(
     return kernel.str();
   });
 
-  compiled_allocate_outputs(
-      inputs, outputs, inputs_, constant_ids_, contiguous, false);
+  compiled_allocate_outputs(inputs, outputs, is_constant_, contiguous);
 
   for (auto& x : outputs) {
     args.push_back(x.data<void>());
+    encoder.set_output_array(x);
   }
-  if (!contiguous) {
-    args.push_back((void*)outputs[0].shape().data());
-  } else {
+  if (contiguous) {
     args.push_back((void*)outputs[0].data_size());
   }
-  auto fun = (void (*)(void**))fn_ptr;
-  fun(args.data());
+  auto fun = reinterpret_cast<void (*)(int*, int64_t**, void**)>(fn_ptr);
+  encoder.dispatch([fun,
+                    args = std::move(args),
+                    strides = std::move(strides),
+                    shape = std::move(shape)]() mutable {
+    SmallVector<int64_t*> strides_ptrs;
+    for (auto& s : strides) {
+      strides_ptrs.push_back(s.data());
+    }
+    fun(shape.data(), strides_ptrs.data(), args.data());
+  });
 }
 
 } // namespace mlx::core

mlx/backend/cpu/copy.cpp

@@ -5,6 +5,7 @@
 #include "mlx/allocator.h"
 #include "mlx/backend/common/utils.h"
 #include "mlx/backend/cpu/copy.h"
+#include "mlx/backend/cpu/encoder.h"
 #include "mlx/backend/cpu/simd/simd.h"
 
 namespace mlx::core {
@@ -13,19 +14,19 @@ namespace {
 
 template <typename SrcT, typename DstT>
 void copy_single(const array& src, array& dst) {
-  auto val = static_cast<DstT>(src.data<SrcT>()[0]);
+  auto src_ptr = src.data<SrcT>();
   auto dst_ptr = dst.data<DstT>();
-  for (int i = 0; i < dst.size(); ++i) {
-    dst_ptr[i] = val;
-  }
+  auto size = dst.size();
+  auto val = static_cast<DstT>(src_ptr[0]);
+  std::fill_n(dst_ptr, size, val);
 }
 
 template <typename SrcT, typename DstT>
 void copy_vector(const array& src, array& dst) {
   auto src_ptr = src.data<SrcT>();
   auto dst_ptr = dst.data<DstT>();
-  size_t size = src.data_size();
-  std::copy(src_ptr, src_ptr + src.data_size(), dst_ptr);
+  auto size = src.data_size();
+  std::copy(src_ptr, src_ptr + size, dst_ptr);
 }
 
 template <typename SrcT, typename DstT, int D>
@@ -60,36 +61,57 @@ void copy_general_general(
     const Strides& i_strides,
     const Strides& o_strides,
     int64_t i_offset,
-    int64_t o_offset) {
+    int64_t o_offset,
+    const std::optional<array>& dynamic_i_offset,
+    const std::optional<array>& dynamic_o_offset) {
+  auto src_ptr = src.data<SrcT>() + i_offset;
+  auto dst_ptr = dst.data<DstT>() + o_offset;
+  auto i_offset_ptr =
+      dynamic_i_offset ? dynamic_i_offset->data<int64_t>() : nullptr;
+  auto o_offset_ptr =
+      dynamic_o_offset ? dynamic_o_offset->data<int64_t>() : nullptr;
+  auto size = src.size();
   if (data_shape.empty()) {
-    auto val = static_cast<DstT>(*(src.data<SrcT>() + i_offset));
-    auto dst_ptr = dst.data<DstT>() + o_offset;
+    auto val = static_cast<DstT>(*src_ptr);
     *dst_ptr = val;
     return;
   }
   auto [shape, strides] =
       collapse_contiguous_dims(data_shape, {i_strides, o_strides});
-  auto src_ptr = src.data<SrcT>() + i_offset;
-  auto dst_ptr = dst.data<DstT>() + o_offset;
+
   int ndim = shape.size();
-  if (ndim == 1) {
-    copy_dims<SrcT, DstT, 1>(
-        src_ptr, dst_ptr, shape, strides[0], strides[1], 0);
-    return;
-  } else if (ndim == 2) {
-    copy_dims<SrcT, DstT, 2>(
-        src_ptr, dst_ptr, shape, strides[0], strides[1], 0);
-    return;
-  } else if (ndim == 3) {
-    copy_dims<SrcT, DstT, 3>(
-        src_ptr, dst_ptr, shape, strides[0], strides[1], 0);
+  if (ndim < 3) {
+    if (i_offset_ptr) {
+      src_ptr += i_offset_ptr[0];
+    }
+    if (o_offset_ptr) {
+      dst_ptr += o_offset_ptr[0];
+    }
+
+    if (ndim == 1) {
+      copy_dims<SrcT, DstT, 1>(
+          src_ptr, dst_ptr, shape, strides[0], strides[1], 0);
+    } else if (ndim == 2) {
+      copy_dims<SrcT, DstT, 2>(
+          src_ptr, dst_ptr, shape, strides[0], strides[1], 0);
+    } else if (ndim == 3) {
+      copy_dims<SrcT, DstT, 3>(
+          src_ptr, dst_ptr, shape, strides[0], strides[1], 0);
+    }
     return;
   }
+  if (i_offset_ptr) {
+    src_ptr += i_offset_ptr[0];
+  }
+  if (o_offset_ptr) {
+    dst_ptr += o_offset_ptr[0];
+  }
+
   ContiguousIterator in(shape, strides[0], ndim - 3);
   ContiguousIterator out(shape, strides[1], ndim - 3);
   auto stride = std::accumulate(
       shape.end() - 3, shape.end(), 1, std::multiplies<int64_t>());
-  for (int64_t elem = 0; elem < src.size(); elem += stride) {
+  for (int64_t elem = 0; elem < size; elem += stride) {
     copy_dims<SrcT, DstT, 3>(
         src_ptr + in.loc,
         dst_ptr + out.loc,
@@ -105,7 +127,15 @@ void copy_general_general(
 template <typename SrcT, typename DstT>
 inline void copy_general_general(const array& src, array& dst) {
   copy_general_general<SrcT, DstT>(
-      src, dst, src.shape(), src.strides(), dst.strides(), 0, 0);
+      src,
+      dst,
+      src.shape(),
+      src.strides(),
+      dst.strides(),
+      0,
+      0,
+      std::nullopt,
+      std::nullopt);
 }
 
 template <typename SrcT, typename DstT>
@@ -116,7 +146,9 @@ void copy_general(
     const Strides& i_strides,
     const Strides&,
     int64_t i_offset,
-    int64_t o_offset) {
+    int64_t o_offset,
+    const std::optional<array>& dynamic_i_offset,
+    const std::optional<array>& dynamic_o_offset) {
   copy_general_general<SrcT, DstT>(
       src,
       dst,
@@ -124,7 +156,9 @@ void copy_general(
       i_strides,
       make_contiguous_strides(data_shape),
       i_offset,
-      o_offset);
+      o_offset,
+      dynamic_i_offset,
+      dynamic_o_offset);
 }
 
 template <typename SrcT, typename DstT>
@@ -136,7 +170,9 @@ inline void copy_general(const array& src, array& dst) {
       src.strides(),
       make_contiguous_strides(src.shape()),
       0,
-      0);
+      0,
+      std::nullopt,
+      std::nullopt);
 }
 
 template <typename SrcT, typename DstT, typename... Args>
@@ -259,38 +295,34 @@ inline void copy_inplace_dispatch(
 
 } // namespace
 
-void copy_inplace(const array& src, array& dst, CopyType ctype) {
-  copy_inplace_dispatch(src, dst, ctype);
+void copy_cpu_inplace(
+    const array& src,
+    array& dst,
+    CopyType ctype,
+    Stream stream) {
+  auto& encoder = cpu::get_command_encoder(stream);
+  encoder.set_input_array(src);
+  encoder.set_output_array(dst);
+  encoder.dispatch(
+      [src = array::unsafe_weak_copy(src),
+       dst = array::unsafe_weak_copy(dst),
+       ctype]() mutable { copy_inplace_dispatch(src, dst, ctype); });
 }
 
-void copy(const array& src, array& dst, CopyType ctype) {
-  // Allocate the output
-  switch (ctype) {
-    case CopyType::Vector:
-      if (src.is_donatable() && src.itemsize() == dst.itemsize()) {
-        dst.copy_shared_buffer(src);
-      } else {
-        auto size = src.data_size();
-        dst.set_data(
-            allocator::malloc_or_wait(size * dst.itemsize()),
-            size,
-            src.strides(),
-            src.flags());
-      }
-      break;
-    case CopyType::Scalar:
-    case CopyType::General:
-    case CopyType::GeneralGeneral:
-      dst.set_data(allocator::malloc_or_wait(dst.nbytes()));
-      break;
+void copy_cpu(const array& src, array& dst, CopyType ctype, Stream stream) {
+  bool donated = set_copy_output_data(src, dst, ctype);
+  if (donated && src.dtype() == dst.dtype()) {
+    // If the output has the same type as the input then there is nothing to
+    // copy, just use the buffer.
+    return;
   }
   if (ctype == CopyType::GeneralGeneral) {
     ctype = CopyType::General;
   }
-  copy_inplace(src, dst, ctype);
+  copy_cpu_inplace(src, dst, ctype, stream);
 }
 
-void copy_inplace(
+void copy_cpu_inplace(
     const array& src,
     array& dst,
     const Shape& data_shape,
@@ -298,24 +330,57 @@ void copy_inplace(
     const Strides& o_strides,
     int64_t i_offset,
     int64_t o_offset,
-    CopyType ctype) {
-  switch (ctype) {
-    case CopyType::General:
-    case CopyType::GeneralGeneral:
-      copy_inplace_dispatch(
-          src,
-          dst,
-          ctype,
-          data_shape,
-          i_strides,
-          o_strides,
-          i_offset,
-          o_offset);
-      break;
-    case CopyType::Scalar:
-    case CopyType::Vector:
-      copy_inplace_dispatch(src, dst, ctype);
-  }
+    CopyType ctype,
+    Stream stream,
+    const std::optional<array>& dynamic_i_offset, /* = std::nullopt */
+    const std::optional<array>& dynamic_o_offset /* = std::nullopt */) {
+  auto& encoder = cpu::get_command_encoder(stream);
+  encoder.set_input_array(src);
+  encoder.set_output_array(dst);
+  auto weak_copy_if_set = [](auto x) -> std::optional<array> {
+    if (x) {
+      return array::unsafe_weak_copy(*x);
+    } else {
+      return std::nullopt;
+    }
+  };
+  encoder.dispatch(
+      [src = array::unsafe_weak_copy(src),
+       dst = array::unsafe_weak_copy(dst),
+       data_shape,
+       i_strides,
+       o_strides,
+       i_offset,
+       o_offset,
+       ctype,
+       dynamic_i_offset = weak_copy_if_set(dynamic_i_offset),
+       dynamic_o_offset = weak_copy_if_set(dynamic_o_offset)]() mutable {
+        switch (ctype) {
+          case CopyType::General:
+          case CopyType::GeneralGeneral:
+            copy_inplace_dispatch(
+                src,
+                dst,
+                ctype,
+                data_shape,
+                i_strides,
+                o_strides,
+                i_offset,
+                o_offset,
+                dynamic_i_offset,
+                dynamic_o_offset);
+            break;
+          case CopyType::Scalar:
+          case CopyType::Vector:
+            copy_inplace_dispatch(src, dst, ctype);
+        }
+      });
+}
+
+array contiguous_copy_cpu(const array& arr, Stream stream) {
+  array arr_copy(arr.shape(), arr.dtype(), nullptr, {});
+  copy_cpu(arr, arr_copy, CopyType::General, stream);
+  return arr_copy;
 }
 
 } // namespace mlx::core

mlx/backend/cpu/copy.h

@@ -2,16 +2,22 @@
 
 #pragma once
 
+#include <optional>
+
 #include "mlx/array.h"
 #include "mlx/backend/common/copy.h"
 #include "mlx/backend/common/utils.h"
 
 namespace mlx::core {
 
-void copy(const array& src, array& dst, CopyType ctype);
-void copy_inplace(const array& src, array& dst, CopyType ctype);
+void copy_cpu(const array& src, array& dst, CopyType ctype, Stream stream);
+void copy_cpu_inplace(
+    const array& src,
+    array& dst,
+    CopyType ctype,
+    Stream stream);
 
-void copy_inplace(
+void copy_cpu_inplace(
     const array& src,
     array& dst,
     const Shape& data_shape,
@@ -19,6 +25,12 @@ void copy_inplace(
     const Strides& o_strides,
     int64_t i_offset,
     int64_t o_offset,
-    CopyType ctype);
+    CopyType ctype,
+    Stream stream,
+    const std::optional<array>& dynamic_i_offset = std::nullopt,
+    const std::optional<array>& dynamic_o_offset = std::nullopt);
+
+// Return a contiguous array with same shape that copies the data of |arr|.
+array contiguous_copy_cpu(const array& arr, Stream stream);
 
 } // namespace mlx::core

mlx/backend/cpu/distributed.cpp

@@ -0,0 +1,98 @@
+// Copyright © 2024 Apple Inc.
+
+#include <cassert>
+
+#include "mlx/allocator.h"
+#include "mlx/backend/cpu/copy.h"
+#include "mlx/backend/cpu/encoder.h"
+#include "mlx/distributed/primitives.h"
+
+namespace mlx::core::distributed {
+
+std::pair<array, bool> ensure_row_contiguous(const array& arr, Stream stream) {
+  if (arr.flags().row_contiguous) {
+    return {arr, false};
+  } else {
+    return {contiguous_copy_cpu(arr, stream), true};
+  }
+};
+
+void AllReduce::eval_cpu(
+    const std::vector<array>& inputs,
+    std::vector<array>& outputs) {
+  assert(inputs.size() == 1);
+  assert(outputs.size() == 1);
+
+  auto donate_or_copy = [s = stream()](const array& in, array& out) {
+    if (in.flags().row_contiguous) {
+      if (in.is_donatable()) {
+        out.copy_shared_buffer(in);
+      } else {
+        out.set_data(allocator::malloc(out.nbytes()));
+      }
+      return in;
+    } else {
+      array arr_copy = contiguous_copy_cpu(in, s);
+      out.copy_shared_buffer(arr_copy);
+      return arr_copy;
+    }
+  };
+
+  auto in = donate_or_copy(inputs[0], outputs[0]);
+  switch (reduce_type_) {
+    case Sum:
+      distributed::detail::all_sum(group(), in, outputs[0], stream());
+      break;
+    case Max:
+      distributed::detail::all_max(group(), in, outputs[0], stream());
+      break;
+    case Min:
+      distributed::detail::all_min(group(), in, outputs[0], stream());
+      break;
+    default:
+      throw std::runtime_error(
+          "Only all reduce sum, min and max are supported for now");
+  }
+}
+
+void AllGather::eval_cpu(
+    const std::vector<array>& inputs,
+    std::vector<array>& outputs) {
+  assert(inputs.size() == 1);
+  assert(outputs.size() == 1);
+
+  auto [in, copied] = ensure_row_contiguous(inputs[0], stream());
+  outputs[0].set_data(allocator::malloc(outputs[0].nbytes()));
+  distributed::detail::all_gather(group(), in, outputs[0], stream());
+  if (copied) {
+    auto& enc = cpu::get_command_encoder(stream());
+    enc.add_temporary(in);
+  }
+}
+
+void Send::eval_cpu(
+    const std::vector<array>& inputs,
+    std::vector<array>& outputs) {
+  assert(inputs.size() == 1);
+  assert(outputs.size() == 1);
+
+  auto [in, copied] = ensure_row_contiguous(inputs[0], stream());
+  distributed::detail::send(group(), in, dst_, stream());
+  outputs[0].copy_shared_buffer(inputs[0]);
+  if (copied) {
+    auto& enc = cpu::get_command_encoder(stream());
+    enc.add_temporary(in);
+  }
+}
+
+void Recv::eval_cpu(
+    const std::vector<array>& inputs,
+    std::vector<array>& outputs) {
+  assert(inputs.size() == 0);
+  assert(outputs.size() == 1);
+
+  outputs[0].set_data(allocator::malloc(outputs[0].nbytes()));
+  distributed::detail::recv(group(), outputs[0], src_, stream());
+}
+
+} // namespace mlx::core::distributed

mlx/backend/cpu/eig.cpp

@@ -0,0 +1,174 @@
+// Copyright © 2025 Apple Inc.
+
+#include "mlx/allocator.h"
+#include "mlx/array.h"
+#include "mlx/backend/cpu/copy.h"
+#include "mlx/backend/cpu/encoder.h"
+#include "mlx/backend/cpu/lapack.h"
+#include "mlx/linalg.h"
+#include "mlx/primitives.h"
+
+namespace mlx::core {
+
+namespace {
+
+template <typename T>
+void eig_impl(
+    array& a,
+    array& vectors,
+    array& values,
+    bool compute_eigenvectors,
+    Stream stream) {
+  using OT = std::complex<T>;
+  auto a_ptr = a.data<T>();
+  auto eig_ptr = values.data<OT>();
+
+  auto& encoder = cpu::get_command_encoder(stream);
+  encoder.set_input_array(a);
+  encoder.set_output_array(values);
+  OT* vec_ptr = nullptr;
+  if (compute_eigenvectors) {
+    encoder.set_output_array(vectors);
+    vec_ptr = vectors.data<OT>();
+  }
+  encoder.dispatch([a_ptr,
+                    vec_ptr,
+                    eig_ptr,
+                    compute_eigenvectors,
+                    N = vectors.shape(-1),
+                    size = vectors.size()]() mutable {
+    // Work query
+    char jobr = 'N';
+    char jobl = compute_eigenvectors ? 'V' : 'N';
+    int n_vecs_r = 1;
+    int n_vecs_l = compute_eigenvectors ? N : 1;
+    int lwork = -1;
+    int info;
+    {
+      T work;
+      int iwork;
+      geev<T>(
+          &jobl,
+          &jobr,
+          &N,
+          nullptr,
+          &N,
+          nullptr,
+          nullptr,
+          nullptr,
+          &n_vecs_l,
+          nullptr,
+          &n_vecs_r,
+          &work,
+          &lwork,
+          &info);
+      lwork = static_cast<int>(work);
+    }
+
+    auto eig_tmp_data = array::Data{allocator::malloc(sizeof(T) * N * 2)};
+    auto vec_tmp_data =
+        array::Data{allocator::malloc(vec_ptr ? sizeof(T) * N * N * 2 : 0)};
+    auto eig_tmp = static_cast<T*>(eig_tmp_data.buffer.raw_ptr());
+    auto vec_tmp = static_cast<T*>(vec_tmp_data.buffer.raw_ptr());
+    auto work_buf = array::Data{allocator::malloc(sizeof(T) * lwork)};
+    for (size_t i = 0; i < size / (N * N); ++i) {
+      geev<T>(
+          &jobl,
+          &jobr,
+          &N,
+          a_ptr,
+          &N,
+          eig_tmp,
+          eig_tmp + N,
+          vec_tmp,
+          &n_vecs_l,
+          nullptr,
+          &n_vecs_r,
+          static_cast<T*>(work_buf.buffer.raw_ptr()),
+          &lwork,
+          &info);
+      for (int i = 0; i < N; ++i) {
+        eig_ptr[i] = {eig_tmp[i], eig_tmp[N + i]};
+      }
+      if (vec_ptr) {
+        for (int i = 0; i < N; ++i) {
+          if (eig_ptr[i].imag() != 0) {
+            // This vector and the next are a pair
+            for (int j = 0; j < N; ++j) {
+              vec_ptr[i * N + j] = {
+                  vec_tmp[i * N + j], -vec_tmp[(i + 1) * N + j]};
+              vec_ptr[(i + 1) * N + j] = {
+                  vec_tmp[i * N + j], vec_tmp[(i + 1) * N + j]};
+            }
+            i += 1;
+          } else {
+            for (int j = 0; j < N; ++j) {
+              vec_ptr[i * N + j] = {vec_tmp[i * N + j], 0};
+            }
+          }
+        }
+        vec_ptr += N * N;
+      }
+      a_ptr += N * N;
+      eig_ptr += N;
+      if (info != 0) {
+        std::stringstream msg;
+        msg << "[Eig::eval_cpu] Eigenvalue decomposition failed with error code "
+            << info;
+        throw std::runtime_error(msg.str());
+      }
+    }
+  });
+  encoder.add_temporary(a);
+}
+
+} // namespace
+
+void Eig::eval_cpu(
+    const std::vector<array>& inputs,
+    std::vector<array>& outputs) {
+  const auto& a = inputs[0];
+  auto& values = outputs[0];
+
+  auto vectors = compute_eigenvectors_
+      ? outputs[1]
+      : array(a.shape(), complex64, nullptr, {});
+
+  auto a_copy = array(a.shape(), a.dtype(), nullptr, {});
+  copy_cpu(
+      a,
+      a_copy,
+      a.flags().row_contiguous ? CopyType::Vector : CopyType::General,
+      stream());
+
+  values.set_data(allocator::malloc(values.nbytes()));
+
+  if (compute_eigenvectors_) {
+    // Set the strides and flags so the eigenvectors
+    // are in the columns of the output
+    auto flags = vectors.flags();
+    auto strides = vectors.strides();
+    auto ndim = a.ndim();
+    std::swap(strides[ndim - 1], strides[ndim - 2]);
+
+    if (a.size() > 1) {
+      flags.row_contiguous = false;
+      if (ndim > 2) {
+        flags.col_contiguous = false;
+      } else {
+        flags.col_contiguous = true;
+      }
+    }
+    vectors.set_data(
+        allocator::malloc(vectors.nbytes()), vectors.size(), strides, flags);
+  }
+  switch (a.dtype()) {
+    case float32:
+      eig_impl<float>(a_copy, vectors, values, compute_eigenvectors_, stream());
+      break;
+    default:
+      throw std::runtime_error("[Eig::eval_cpu] only supports float32.");
+  }
+}
+
+} // namespace mlx::core

mlx/backend/cpu/eigh.cpp

@@ -3,6 +3,7 @@
 #include "mlx/allocator.h"
 #include "mlx/array.h"
 #include "mlx/backend/cpu/copy.h"
+#include "mlx/backend/cpu/encoder.h"
 #include "mlx/backend/cpu/lapack.h"
 #include "mlx/linalg.h"
 #include "mlx/primitives.h"
@@ -11,35 +12,173 @@ namespace mlx::core {
 
 namespace {
 
-void ssyevd(
-    char jobz,
-    char uplo,
-    float* a,
-    int N,
-    float* w,
-    float* work,
-    int lwork,
-    int* iwork,
-    int liwork) {
+template <typename T, class Enable = void>
+struct EighWork {};
+
+template <typename T>
+struct EighWork<
+    T,
+    typename std::enable_if<std::is_floating_point<T>::value>::type> {
+  using R = T;
+
+  char jobz;
+  char uplo;
+  int N;
+  int lwork;
+  int liwork;
   int info;
-  MLX_LAPACK_FUNC(ssyevd)
-  (
-      /* jobz = */ &jobz,
-      /* uplo = */ &uplo,
-      /* n = */ &N,
-      /* a = */ a,
-      /* lda = */ &N,
-      /* w = */ w,
-      /* work = */ work,
-      /* lwork = */ &lwork,
-      /* iwork = */ iwork,
-      /* liwork = */ &liwork,
-      /* info = */ &info);
-  if (info != 0) {
-    std::stringstream msg;
-    msg << "[Eigh::eval_cpu] Eigenvalue decomposition failed with error code "
-        << info;
-    throw std::runtime_error(msg.str());
+  std::vector<array::Data> buffers;
+
+  EighWork(char jobz_, char uplo_, int N_)
+      : jobz(jobz_), uplo(uplo_), N(N_), lwork(-1), liwork(-1) {
+    T work;
+    int iwork;
+    syevd<T>(
+        &jobz,
+        &uplo,
+        &N,
+        nullptr,
+        &N,
+        nullptr,
+        &work,
+        &lwork,
+        &iwork,
+        &liwork,
+        &info);
+    lwork = static_cast<int>(work);
+    liwork = iwork;
+    buffers.emplace_back(allocator::malloc(sizeof(T) * lwork));
+    buffers.emplace_back(allocator::malloc(sizeof(int) * liwork));
+  }
+
+  void run(T* vectors, T* values) {
+    syevd<T>(
+        &jobz,
+        &uplo,
+        &N,
+        vectors,
+        &N,
+        values,
+        static_cast<T*>(buffers[0].buffer.raw_ptr()),
+        &lwork,
+        static_cast<int*>(buffers[1].buffer.raw_ptr()),
+        &liwork,
+        &info);
+  }
+};
+
+template <>
+struct EighWork<std::complex<float>> {
+  using T = std::complex<float>;
+  using R = float;
+
+  char jobz;
+  char uplo;
+  int N;
+  int lwork;
+  int lrwork;
+  int liwork;
+  int info;
+  std::vector<array::Data> buffers;
+
+  EighWork(char jobz_, char uplo_, int N_)
+      : jobz(jobz_), uplo(uplo_), N(N_), lwork(-1), lrwork(-1), liwork(-1) {
+    T work;
+    R rwork;
+    int iwork;
+    heevd<T>(
+        &jobz,
+        &uplo,
+        &N,
+        nullptr,
+        &N,
+        nullptr,
+        &work,
+        &lwork,
+        &rwork,
+        &lrwork,
+        &iwork,
+        &liwork,
+        &info);
+    lwork = static_cast<int>(work.real());
+    lrwork = static_cast<int>(rwork);
+    liwork = iwork;
+    buffers.emplace_back(allocator::malloc(sizeof(T) * lwork));
+    buffers.emplace_back(allocator::malloc(sizeof(R) * lrwork));
+    buffers.emplace_back(allocator::malloc(sizeof(int) * liwork));
+  }
+
+  void run(T* vectors, R* values) {
+    heevd<T>(
+        &jobz,
+        &uplo,
+        &N,
+        vectors,
+        &N,
+        values,
+        static_cast<T*>(buffers[0].buffer.raw_ptr()),
+        &lwork,
+        static_cast<R*>(buffers[1].buffer.raw_ptr()),
+        &lrwork,
+        static_cast<int*>(buffers[2].buffer.raw_ptr()),
+        &liwork,
+        &info);
+    if (jobz == 'V') {
+      // We have pre-transposed the vectors but we also must conjugate them
+      // when they are complex.
+      //
+      // We could vectorize this but it is so fast in comparison to heevd that
+      // it doesn't really matter.
+      for (int i = 0; i < N; i++) {
+        for (int j = 0; j < N; j++) {
+          *vectors = std::conj(*vectors);
+          vectors++;
+        }
+      }
+    }
+  }
+};
+
+template <typename T>
+void eigh_impl(
+    array& vectors,
+    array& values,
+    const std::string& uplo,
+    bool compute_eigenvectors,
+    Stream stream) {
+  using R = typename EighWork<T>::R;
+
+  auto vec_ptr = vectors.data<T>();
+  auto eig_ptr = values.data<R>();
+  char jobz = compute_eigenvectors ? 'V' : 'N';
+
+  auto& encoder = cpu::get_command_encoder(stream);
+  encoder.set_output_array(vectors);
+  encoder.set_output_array(values);
+  encoder.dispatch([vec_ptr,
+                    eig_ptr,
+                    jobz,
+                    uplo = uplo[0],
+                    N = vectors.shape(-1),
+                    size = vectors.size()]() mutable {
+    // Work query
+    EighWork<T> work(jobz, uplo, N);
+
+    // Work loop
+    for (size_t i = 0; i < size / (N * N); ++i) {
+      work.run(vec_ptr, eig_ptr);
+      vec_ptr += N * N;
+      eig_ptr += N;
+      if (work.info != 0) {
+        std::stringstream msg;
+        msg << "[Eigh::eval_cpu] Eigenvalue decomposition failed with error code "
+            << work.info;
+        throw std::runtime_error(msg.str());
+      }
+    }
+  });
+  if (!compute_eigenvectors) {
+    encoder.add_temporary(vectors);
   }
 }
 
@@ -55,12 +194,13 @@ void Eigh::eval_cpu(
       ? outputs[1]
       : array(a.shape(), a.dtype(), nullptr, {});
 
-  values.set_data(allocator::malloc_or_wait(values.nbytes()));
+  values.set_data(allocator::malloc(values.nbytes()));
 
-  copy(
+  copy_cpu(
       a,
       vectors,
-      a.flags().row_contiguous ? CopyType::Vector : CopyType::General);
+      a.flags().row_contiguous ? CopyType::Vector : CopyType::General,
+      stream());
 
   if (compute_eigenvectors_) {
     // Set the strides and flags so the eigenvectors
@@ -78,41 +218,23 @@ void Eigh::eval_cpu(
         flags.col_contiguous = true;
       }
     }
-    vectors.move_shared_buffer(vectors, strides, flags, vectors.data_size());
+    vectors.copy_shared_buffer(vectors, strides, flags, vectors.data_size());
   }
-
-  auto vec_ptr = vectors.data<float>();
-  auto eig_ptr = values.data<float>();
-
-  char jobz = compute_eigenvectors_ ? 'V' : 'N';
-  auto N = a.shape(-1);
-
-  // Work query
-  int lwork;
-  int liwork;
-  {
-    float work;
-    int iwork;
-    ssyevd(jobz, uplo_[0], nullptr, N, nullptr, &work, -1, &iwork, -1);
-    lwork = static_cast<int>(work);
-    liwork = iwork;
-  }
-
-  auto work_buf = array::Data{allocator::malloc_or_wait(sizeof(float) * lwork)};
-  auto iwork_buf = array::Data{allocator::malloc_or_wait(sizeof(int) * liwork)};
-  for (size_t i = 0; i < a.size() / (N * N); ++i) {
-    ssyevd(
-        jobz,
-        uplo_[0],
-        vec_ptr,
-        N,
-        eig_ptr,
-        static_cast<float*>(work_buf.buffer.raw_ptr()),
-        lwork,
-        static_cast<int*>(iwork_buf.buffer.raw_ptr()),
-        liwork);
-    vec_ptr += N * N;
-    eig_ptr += N;
+  switch (a.dtype()) {
+    case float32:
+      eigh_impl<float>(vectors, values, uplo_, compute_eigenvectors_, stream());
+      break;
+    case float64:
+      eigh_impl<double>(
+          vectors, values, uplo_, compute_eigenvectors_, stream());
+      break;
+    case complex64:
+      eigh_impl<std::complex<float>>(
+          vectors, values, uplo_, compute_eigenvectors_, stream());
+      break;
+    default:
+      throw std::runtime_error(
+          "[Eigh::eval_cpu] only supports float32 or float64.");
   }
 }
 

mlx/backend/cpu/encoder.cpp

@@ -0,0 +1,16 @@
+// Copyright © 2025 Apple Inc.
+
+#include "mlx/backend/cpu/encoder.h"
+
+namespace mlx::core::cpu {
+
+CommandEncoder& get_command_encoder(Stream stream) {
+  static std::unordered_map<int, CommandEncoder> encoder_map;
+  auto it = encoder_map.find(stream.index);
+  if (it == encoder_map.end()) {
+    it = encoder_map.emplace(stream.index, stream).first;
+  }
+  return it->second;
+}
+
+} // namespace mlx::core::cpu

mlx/backend/cpu/encoder.h

@@ -0,0 +1,67 @@
+// Copyright © 2025 Apple Inc.
+
+#pragma once
+
+#include <unordered_map>
+
+#include "mlx/array.h"
+#include "mlx/scheduler.h"
+
+namespace mlx::core::cpu {
+
+// Number of dispatches per scheduler task
+constexpr int DISPATCHES_PER_TASK = 10;
+
+struct CommandEncoder {
+  CommandEncoder(Stream stream) : stream_(stream) {}
+
+  CommandEncoder(const CommandEncoder&) = delete;
+  CommandEncoder& operator=(const CommandEncoder&) = delete;
+  CommandEncoder(CommandEncoder&&) = delete;
+  CommandEncoder& operator=(CommandEncoder&&) = delete;
+
+  void set_input_array(const array& a) {}
+  void set_output_array(array& a) {}
+
+  // Hold onto a temporary until any already scheduled tasks which use it as
+  // an input are complete.
+  void add_temporary(array arr) {
+    temporaries_.push_back(std::move(arr));
+  }
+
+  void add_temporaries(std::vector<array> arrays) {
+    temporaries_.insert(
+        temporaries_.end(),
+        std::make_move_iterator(arrays.begin()),
+        std::make_move_iterator(arrays.end()));
+  }
+
+  std::vector<array>& temporaries() {
+    return temporaries_;
+  }
+
+  template <class F, class... Args>
+  void dispatch(F&& f, Args&&... args) {
+    num_ops_ = (num_ops_ + 1) % DISPATCHES_PER_TASK;
+    auto task = std::bind(std::forward<F>(f), std::forward<Args>(args)...);
+    if (num_ops_ == 0) {
+      scheduler::notify_new_task(stream_);
+      auto task_wrap = [s = stream_, task = std::move(task)]() mutable {
+        task();
+        scheduler::notify_task_completion(s);
+      };
+      scheduler::enqueue(stream_, std::move(task_wrap));
+    } else {
+      scheduler::enqueue(stream_, std::move(task));
+    }
+  }
+
+ private:
+  Stream stream_;
+  std::vector<array> temporaries_;
+  int num_ops_{0};
+};
+
+CommandEncoder& get_command_encoder(Stream stream);
+
+} // namespace mlx::core::cpu

mlx/backend/cpu/eval.cpp

@@ -0,0 +1,40 @@
+// Copyright © 2025 Apple Inc.
+#include "mlx/backend/cpu/eval.h"
+#include "mlx/backend/cpu/encoder.h"
+#include "mlx/primitives.h"
+#include "mlx/scheduler.h"
+#include "mlx/utils.h"
+
+namespace mlx::core::cpu {
+
+void eval(array& arr) {
+  auto s = arr.primitive().stream();
+
+  auto outputs = arr.outputs();
+  {
+    // If the array is a tracer hold a reference
+    // to its inputs so they don't get donated
+    std::vector<array> inputs;
+    if (arr.is_tracer()) {
+      inputs = arr.inputs();
+    }
+    arr.primitive().eval_cpu(arr.inputs(), outputs);
+  }
+
+  std::unordered_set<std::shared_ptr<array::Data>> buffers;
+  for (auto& in : arr.inputs()) {
+    buffers.insert(in.data_shared_ptr());
+  }
+  for (auto& s : arr.siblings()) {
+    buffers.insert(s.data_shared_ptr());
+  }
+  // Remove the output if it was donated to by an input
+  if (auto it = buffers.find(arr.data_shared_ptr()); it != buffers.end()) {
+    buffers.erase(it);
+  }
+  auto& encoder = cpu::get_command_encoder(s);
+  encoder.dispatch([buffers = std::move(buffers),
+                    temps = std::move(encoder.temporaries())]() {});
+}
+
+} // namespace mlx::core::cpu

mlx/backend/cpu/eval.h

@@ -0,0 +1,12 @@
+// Copyright © 2025 Apple Inc.
+
+#pragma once
+
+#include "mlx/array.h"
+#include "mlx/stream.h"
+
+namespace mlx::core::cpu {
+
+void eval(array& arr);
+
+} // namespace mlx::core::cpu

mlx/backend/cpu/fft.cpp

@@ -4,6 +4,7 @@
 
 #include "mlx/3rdparty/pocketfft.h"
 #include "mlx/allocator.h"
+#include "mlx/backend/cpu/encoder.h"
 #include "mlx/primitives.h"
 
 namespace mlx::core {
@@ -21,7 +22,7 @@ void FFT::eval_cpu(const std::vector<array>& inputs, array& out) {
     s *= out.itemsize();
   }
 
-  out.set_data(allocator::malloc_or_wait(out.nbytes()));
+  out.set_data(allocator::malloc(out.nbytes()));
 
   std::vector<size_t> shape;
   if (out.dtype() == float32) {
@@ -38,46 +39,78 @@ void FFT::eval_cpu(const std::vector<array>& inputs, array& out) {
         });
     scale /= nelem;
   }
+
+  auto& encoder = cpu::get_command_encoder(stream());
+  encoder.set_input_array(in);
+  encoder.set_output_array(out);
+
   if (in.dtype() == complex64 && out.dtype() == complex64) {
     auto in_ptr =
         reinterpret_cast<const std::complex<float>*>(in.data<complex64_t>());
     auto out_ptr =
         reinterpret_cast<std::complex<float>*>(out.data<complex64_t>());
-    pocketfft::c2c(
-        shape,
-        strides_in,
-        strides_out,
-        axes_,
-        !inverse_,
-        in_ptr,
-        out_ptr,
-        scale);
+    encoder.dispatch([shape = std::move(shape),
+                      strides_in = std::move(strides_in),
+                      strides_out = std::move(strides_out),
+                      axes = axes_,
+                      inverse = inverse_,
+                      in_ptr,
+                      out_ptr,
+                      scale]() {
+      pocketfft::c2c(
+          shape,
+          strides_in,
+          strides_out,
+          axes,
+          !inverse,
+          in_ptr,
+          out_ptr,
+          scale);
+    });
   } else if (in.dtype() == float32 && out.dtype() == complex64) {
     auto in_ptr = in.data<float>();
     auto out_ptr =
         reinterpret_cast<std::complex<float>*>(out.data<complex64_t>());
-    pocketfft::r2c(
-        shape,
-        strides_in,
-        strides_out,
-        axes_,
-        !inverse_,
-        in_ptr,
-        out_ptr,
-        scale);
+    encoder.dispatch([shape = std::move(shape),
+                      strides_in = std::move(strides_in),
+                      strides_out = std::move(strides_out),
+                      axes = axes_,
+                      inverse = inverse_,
+                      in_ptr,
+                      out_ptr,
+                      scale]() {
+      pocketfft::r2c(
+          shape,
+          strides_in,
+          strides_out,
+          axes,
+          !inverse,
+          in_ptr,
+          out_ptr,
+          scale);
+    });
   } else if (in.dtype() == complex64 && out.dtype() == float32) {
     auto in_ptr =
         reinterpret_cast<const std::complex<float>*>(in.data<complex64_t>());
     auto out_ptr = out.data<float>();
-    pocketfft::c2r(
-        shape,
-        strides_in,
-        strides_out,
-        axes_,
-        !inverse_,
-        in_ptr,
-        out_ptr,
-        scale);
+    encoder.dispatch([shape = std::move(shape),
+                      strides_in = std::move(strides_in),
+                      strides_out = std::move(strides_out),
+                      axes = axes_,
+                      inverse = inverse_,
+                      in_ptr,
+                      out_ptr,
+                      scale]() {
+      pocketfft::c2r(
+          shape,
+          strides_in,
+          strides_out,
+          axes,
+          !inverse,
+          in_ptr,
+          out_ptr,
+          scale);
+    });
   } else {
     throw std::runtime_error(
         "[FFT] Received unexpected input and output type combination.");

mlx/backend/cpu/gemm.h

@@ -7,14 +7,20 @@ namespace mlx::core {
 
 template <typename T>
 void matmul(
-    const array& a,
-    const array& b,
-    array& out,
+    const T* a,
+    const T* b,
+    T* out,
     bool a_transposed,
     bool b_transposed,
     size_t lda,
     size_t ldb,
+    size_t ldc,
     float alpha,
-    float beta);
+    float beta,
+    size_t batch_size,
+    const Shape& a_shape,
+    const Strides& a_strides,
+    const Shape& b_shape,
+    const Strides& b_strides);
 
 } // namespace mlx::core

mlx/backend/cpu/gemms/bnns.cpp

@@ -9,39 +9,46 @@
 
 namespace mlx::core {
 
-BNNSDataType to_bnns_dtype(Dtype mlx_dtype) {
-  uint32_t size_bits = size_of(mlx_dtype) * 8;
-  switch (kindof(mlx_dtype)) {
-    case Dtype::Kind::b:
-      return BNNSDataTypeBoolean;
-    case Dtype::Kind::u:
-      return BNNSDataType(BNNSDataTypeUIntBit | size_bits);
-    case Dtype::Kind::i:
-      return BNNSDataType(BNNSDataTypeIntBit | size_bits);
-    case Dtype::Kind::f:
-      return BNNSDataType(BNNSDataTypeFloatBit | size_bits);
-    case Dtype::Kind::V:
-      return BNNSDataTypeBFloat16;
-    case Dtype::Kind::c:
-      throw std::invalid_argument("BNNS does not support complex types");
-  }
+template <typename T>
+constexpr BNNSDataType to_bnns_dtype();
+
+template <>
+constexpr BNNSDataType to_bnns_dtype<float>() {
+  return BNNSDataType(BNNSDataTypeFloatBit | 32);
+}
+template <>
+constexpr BNNSDataType to_bnns_dtype<float16_t>() {
+  return BNNSDataType(BNNSDataTypeFloatBit | 16);
 }
 
+template <>
+constexpr BNNSDataType to_bnns_dtype<bfloat16_t>() {
+  return BNNSDataTypeBFloat16;
+}
+
+template <typename T>
 void matmul_bnns(
-    const array& a,
-    const array& b,
-    array& out,
+    const T* a,
+    const T* b,
+    T* out,
     bool a_transposed,
     bool b_transposed,
     size_t lda,
     size_t ldb,
+    size_t ldc,
     float alpha,
-    float beta) {
-  size_t M = a.shape(-2);
-  size_t N = b.shape(-1);
-  size_t K = a.shape(-1);
+    float beta,
+    size_t batch_size,
+    const Shape& a_shape,
+    const Strides& a_strides,
+    const Shape& b_shape,
+    const Strides& b_strides) {
+  auto ndim = a_shape.size();
+  size_t M = a_shape[ndim - 2];
+  size_t N = b_shape[ndim - 1];
+  size_t K = a_shape[ndim - 1];
 
-  BNNSDataType bnns_dtype = to_bnns_dtype(out.dtype());
+  BNNSDataType bnns_dtype = to_bnns_dtype<T>();
 
 #pragma GCC diagnostic push
 #pragma GCC diagnostic ignored "-Wdeprecated-declarations"
@@ -115,14 +122,14 @@ void matmul_bnns(
   auto bnns_filter =
       BNNSFilterCreateLayerBroadcastMatMul(&gemm_params, nullptr);
 
-  for (int i = 0; i < (a.size() / (M * K)); ++i) {
+  for (int i = 0; i < batch_size; ++i) {
     BNNSFilterApplyTwoInput(
         bnns_filter,
-        a.data<uint8_t>() +
-            elem_to_loc(M * K * i, a.shape(), a.strides()) * a.itemsize(),
-        b.data<uint8_t>() +
-            elem_to_loc(K * N * i, b.shape(), b.strides()) * b.itemsize(),
-        out.data<uint8_t>() + M * N * i * out.itemsize());
+        reinterpret_cast<const uint8_t*>(
+            a + elem_to_loc(M * K * i, a_shape, a_strides)),
+        reinterpret_cast<const uint8_t*>(
+            b + elem_to_loc(K * N * i, b_shape, b_strides)),
+        reinterpret_cast<uint8_t*>(out + M * N * i));
   }
 
   BNNSFilterDestroy(bnns_filter);
@@ -131,30 +138,72 @@ void matmul_bnns(
 
 template <>
 void matmul<float16_t>(
-    const array& a,
-    const array& b,
-    array& out,
+    const float16_t* a,
+    const float16_t* b,
+    float16_t* out,
     bool a_transposed,
     bool b_transposed,
     size_t lda,
     size_t ldb,
+    size_t ldc,
     float alpha,
-    float beta) {
-  matmul_bnns(a, b, out, a_transposed, b_transposed, lda, ldb, alpha, beta);
+    float beta,
+    size_t batch_size,
+    const Shape& a_shape,
+    const Strides& a_strides,
+    const Shape& b_shape,
+    const Strides& b_strides) {
+  matmul_bnns(
+      a,
+      b,
+      out,
+      a_transposed,
+      b_transposed,
+      lda,
+      ldb,
+      ldc,
+      alpha,
+      beta,
+      batch_size,
+      a_shape,
+      a_strides,
+      b_shape,
+      b_strides);
 }
 
 template <>
 void matmul<bfloat16_t>(
-    const array& a,
-    const array& b,
-    array& out,
+    const bfloat16_t* a,
+    const bfloat16_t* b,
+    bfloat16_t* out,
     bool a_transposed,
     bool b_transposed,
     size_t lda,
     size_t ldb,
+    size_t ldc,
     float alpha,
-    float beta) {
-  matmul_bnns(a, b, out, a_transposed, b_transposed, lda, ldb, alpha, beta);
+    float beta,
+    size_t batch_size,
+    const Shape& a_shape,
+    const Strides& a_strides,
+    const Shape& b_shape,
+    const Strides& b_strides) {
+  matmul_bnns(
+      a,
+      b,
+      out,
+      a_transposed,
+      b_transposed,
+      lda,
+      ldb,
+      ldc,
+      alpha,
+      beta,
+      batch_size,
+      a_shape,
+      a_strides,
+      b_shape,
+      b_strides);
 }
 
 } // namespace mlx::core

mlx/backend/cpu/gemms/cblas.cpp

@@ -8,20 +8,27 @@ namespace mlx::core {
 
 template <>
 void matmul<float>(
-    const array& a,
-    const array& b,
-    array& out,
+    const float* a,
+    const float* b,
+    float* out,
     bool a_transposed,
     bool b_transposed,
     size_t lda,
     size_t ldb,
+    size_t ldc,
     float alpha,
-    float beta) {
-  size_t M = a.shape(-2);
-  size_t N = b.shape(-1);
-  size_t K = a.shape(-1);
+    float beta,
+    size_t batch_size,
+    const Shape& a_shape,
+    const Strides& a_strides,
+    const Shape& b_shape,
+    const Strides& b_strides) {
+  auto ndim = a_shape.size();
+  size_t M = a_shape[ndim - 2];
+  size_t N = b_shape[ndim - 1];
+  size_t K = a_shape[ndim - 1];
 
-  for (int i = 0; i < (a.size() / (M * K)); ++i) {
+  for (int i = 0; i < batch_size; ++i) {
     cblas_sgemm(
         CblasRowMajor,
         a_transposed ? CblasTrans : CblasNoTrans, // transA
@@ -29,34 +36,40 @@ void matmul<float>(
         M,
         N,
         K,
-        alpha, // alpha
-        a.data<float>() + elem_to_loc(M * K * i, a.shape(), a.strides()),
+        alpha,
+        a + elem_to_loc(M * K * i, a_shape, a_strides),
         lda,
-        b.data<float>() + elem_to_loc(K * N * i, b.shape(), b.strides()),
+        b + elem_to_loc(K * N * i, b_shape, b_strides),
         ldb,
-        beta, // beta
-        out.data<float>() + M * N * i,
-        out.shape(-1) // ldc
-    );
+        beta,
+        out + M * N * i,
+        ldc);
   }
 }
 
 template <>
 void matmul<double>(
-    const array& a,
-    const array& b,
-    array& out,
+    const double* a,
+    const double* b,
+    double* out,
     bool a_transposed,
     bool b_transposed,
     size_t lda,
     size_t ldb,
+    size_t ldc,
     float alpha,
-    float beta) {
-  size_t M = a.shape(-2);
-  size_t N = b.shape(-1);
-  size_t K = a.shape(-1);
+    float beta,
+    size_t batch_size,
+    const Shape& a_shape,
+    const Strides& a_strides,
+    const Shape& b_shape,
+    const Strides& b_strides) {
+  auto ndim = a_shape.size();
+  size_t M = a_shape[ndim - 2];
+  size_t N = b_shape[ndim - 1];
+  size_t K = a_shape[ndim - 1];
 
-  for (int i = 0; i < (a.size() / (M * K)); ++i) {
+  for (int i = 0; i < batch_size; ++i) {
     cblas_dgemm(
         CblasRowMajor,
         a_transposed ? CblasTrans : CblasNoTrans, // transA
@@ -64,15 +77,14 @@ void matmul<double>(
         M,
         N,
         K,
-        alpha, // alpha
-        a.data<double>() + elem_to_loc(M * K * i, a.shape(), a.strides()),
+        alpha,
+        a + elem_to_loc(M * K * i, a_shape, a_strides),
         lda,
-        b.data<double>() + elem_to_loc(K * N * i, b.shape(), b.strides()),
+        b + elem_to_loc(K * N * i, b_shape, b_strides),
         ldb,
-        beta, // beta
-        out.data<double>() + M * N * i,
-        out.shape(-1) // ldc
-    );
+        beta,
+        out + M * N * i,
+        ldc);
   }
 }
 

mlx/backend/cpu/gemms/no_bf16.cpp

@@ -1,21 +0,0 @@
-// Copyright © 2025 Apple Inc.
-
-#include "mlx/backend/cpu/gemm.h"
-
-namespace mlx::core {
-
-template <>
-void matmul<bfloat16_t>(
-    const array&,
-    const array&,
-    array&,
-    bool,
-    bool,
-    size_t,
-    size_t,
-    float,
-    float) {
-  throw std::runtime_error("[Matmul::eval_cpu] bfloat16 not supported.");
-}
-
-} // namespace mlx::core