From 74eccbf3fae68e6bcb3cb554b9b241f304334ef9 Mon Sep 17 00:00:00 2001
From: Awni Hannun <awni@apple.com>
Date: Tue, 22 Jul 2025 07:00:53 -0700
Subject: [PATCH] use size option in binary (#2399)

---
 mlx/backend/cuda/CMakeLists.txt | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/mlx/backend/cuda/CMakeLists.txt b/mlx/backend/cuda/CMakeLists.txt
index 9f236b4ea2..a98308044b 100644
--- a/mlx/backend/cuda/CMakeLists.txt
+++ b/mlx/backend/cuda/CMakeLists.txt
@@ -87,6 +87,11 @@ endif()
 target_compile_options(
   mlx PRIVATE "$<$<COMPILE_LANGUAGE:CUDA>:--Wno-deprecated-gpu-targets>")
 
+if(CMAKE_CUDA_COMPILER_VERSION VERSION_GREATER_EQUAL 12.4.0)
+  target_compile_options(
+    mlx PRIVATE "$<$<COMPILE_LANGUAGE:CUDA>:--compress-mode=size>")
+endif()
+
 # Compute capability 7 is required for synchronization between CPU/GPU with
 # managed memory. TODO: Add more architectures for potential performance gain.
 set(MLX_CUDA_ARCHITECTURES