zhangyiss/mlx
1
0
Fork 0
mirror of https://github.com/ml-explore/mlx.git synced 2025-12-16 01:49:05 +08:00
Code Issues Actions Packages Projects Releases Wiki Activity

[CUDA] Use cuda::std::complex in place of cuComplex (#2372)

Browse Source
This commit is contained in:
Cheng
2025-07-15 16:36:13 +09:00
committed by GitHub
parent f0a0b077a0
commit cb349a291c
15 changed files with 169 additions and 460 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
mlx/backend/cuda/device/atomic_ops.cuh
View File

@@ -2,7 +2,7 @@
#pragma once
#include "mlx/backend/cuda/device/cucomplex_math.cuh"
#include "mlx/backend/cuda/device/complex.cuh"
#include "mlx/backend/cuda/device/fp16_math.cuh"
#include <cuda/atomic>
@@ -48,7 +48,7 @@ inline __device__ void atomic_add(__half* out, __half val) {
atomicAdd(out, val);
}
inline __device__ void atomic_add(cuComplex* out, cuComplex val) {
inline __device__ void atomic_add(complex64_t* out, complex64_t val) {
#if __CUDA_ARCH__ < 900
atomic_add_general(out, val);
#else