Awni Hannun 68c5fa1c95 fix memory count bug (#2717)

2025-10-30 14:27:15 -07:00

..

binary

Faster general unary op (#2472)

2025-08-15 15:04:12 -07:00

conv

[CUDA] Add GEMM-based fallback convolution kernels (#2511)

2025-08-20 10:06:22 +09:00

copy

Faster general unary op (#2472)

2025-08-15 15:04:12 -07:00

device

Fp8 conversion (#2686)

2025-10-27 16:35:50 -07:00

gemms

Faster complex matmul (#2571)

2025-10-02 23:33:15 -07:00

quantized

Add quantize/dequantize for mxfp8 and nvfp4 (#2688)

2025-10-28 16:23:12 -07:00

reduce

Fix and refactor row-reduce (#2650)

2025-10-07 01:51:08 -07:00

steel

[CUDA] Matmul utils initial commit (#2441)

2025-08-01 14:22:25 -07:00

unary

Fp8 conversion (#2686)

2025-10-27 16:35:50 -07:00

allocator.cpp

fix memory count bug (#2717)

2025-10-30 14:27:15 -07:00

allocator.h

[CUDA] Simplify allocator (#2392)

2025-07-22 08:24:01 -07:00

arange.cu

[CUDA] Remove thrust in arange (#2535)

2025-08-24 16:22:36 +09:00

arg_reduce.cu

[CUDA] Matmul utils initial commit (#2441)

2025-08-01 14:22:25 -07:00

bin2h.cmake

CUDA backend: compile (#2276)

2025-06-12 17:08:39 -07:00

binary_two.cu

Faster general unary op (#2472)

2025-08-15 15:04:12 -07:00

CMakeLists.txt

Add quantize/dequantize for mxfp8 and nvfp4 (#2688)

2025-10-28 16:23:12 -07:00

compiled.cpp

fix for max block dim (#2631)

2025-09-29 08:59:25 -07:00

conv.cpp

Debug cuda conv (#2662)

2025-10-10 16:12:47 -07:00

copy.cu

[CUDA] Implement DynamicSlice/DynamicSliceUpdate (#2533)

2025-08-26 07:31:39 +09:00

cuda.cpp

start cuda circle config (#2256)

2025-06-10 21:19:47 -07:00

cuda.h

start cuda circle config (#2256)

2025-06-10 21:19:47 -07:00

cudnn_utils.cpp

Debug cuda conv (#2662)

2025-10-10 16:12:47 -07:00

cudnn_utils.h

Split cuDNN helpers into a separate header (#2491)

2025-08-20 09:29:28 +09:00

custom_kernel.cpp

Custom cuda kernel (#2517)

2025-08-20 17:20:22 -07:00

device.cpp

wait for tasks in cuda (#2636)

2025-09-30 16:08:46 -07:00

device.h

wait for tasks in cuda (#2636)

2025-09-30 16:08:46 -07:00

distributed.cu

Fix allocation bug in NCCL (#2530)

2025-08-22 14:39:43 -07:00

eval.cpp

wait for tasks in cuda (#2636)

2025-09-30 16:08:46 -07:00

event.cu

[CUDA] Make CudaEvent work with multi-device (#2614)

2025-09-27 11:27:17 +09:00

event.h

[CUDA] Make CudaEvent work with multi-device (#2614)

2025-09-27 11:27:17 +09:00

fence.cpp

[CUDA] Recycle CUDA events (#2604)

2025-09-23 10:42:03 +09:00

indexing.cpp

[CUDA] Implement DynamicSlice/DynamicSliceUpdate (#2533)

2025-08-26 07:31:39 +09:00

jit_module.cpp

Split name into directories for cuda jit (#2656)

2025-10-07 01:52:58 -07:00

jit_module.h

fix for max block dim (#2631)

2025-09-29 08:59:25 -07:00

kernel_utils.cu

fix for max block dim (#2631)

2025-09-29 08:59:25 -07:00

kernel_utils.cuh

Export with callback (#2612)

2025-10-08 19:24:33 -07:00

layer_norm.cu

[CUDA] Matmul utils initial commit (#2441)

2025-08-01 14:22:25 -07:00

logsumexp.cu

[CUDA] Matmul utils initial commit (#2441)

2025-08-01 14:22:25 -07:00

lru_cache.h

Detect cache thrashing in LRUCache (#2600)

2025-09-19 09:12:14 +09:00

matmul.cpp

Faster complex matmul (#2571)

2025-10-02 23:33:15 -07:00

no_cuda.cpp

Custom cuda kernel (#2517)

2025-08-20 17:20:22 -07:00

primitives.cpp

[CUDA] Implement DynamicSlice/DynamicSliceUpdate (#2533)

2025-08-26 07:31:39 +09:00

random.cu

[CUDA] Matmul utils initial commit (#2441)

2025-08-01 14:22:25 -07:00

reduce.cu

faster rms norm (#2433)

2025-07-29 13:12:00 -07:00

rms_norm.cu

[CUDA] Matmul utils initial commit (#2441)

2025-08-01 14:22:25 -07:00

rope.cu

Add batch offsets for mx.fast.rope (#2564)

2025-09-08 17:35:07 -07:00

scaled_dot_product_attention.cu

avoid producing NaN in attention (#2608)

2025-09-22 13:10:43 -07:00

scan.cu

[CUDA] Matmul utils initial commit (#2441)

2025-08-01 14:22:25 -07:00

slicing.cpp

[CUDA] Use ConcurrentContext in concatenate_gpu (#2549)

2025-08-28 09:30:08 +09:00

softmax.cu

[CUDA] Matmul utils initial commit (#2441)

2025-08-01 14:22:25 -07:00

sort.cu

[CUDA] fix sort (#2550)

2025-08-27 19:48:43 -07:00

ternary.cu

Faster triu, tril, where with scalar (#2644)

2025-10-02 12:21:27 -07:00

utils.cpp

Split cuDNN helpers into a separate header (#2491)

2025-08-20 09:29:28 +09:00

utils.h

Export with callback (#2612)

2025-10-08 19:24:33 -07:00

worker.cpp

[CUDA] Make CudaEvent work with multi-device (#2614)

2025-09-27 11:27:17 +09:00

worker.h

[CUDA] Make CudaEvent work with multi-device (#2614)

2025-09-27 11:27:17 +09:00