tqcq/llama.cpp
0
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-08-14 20:29:41 -04:00
Code Issues Packages Projects Releases Wiki Activity

HIP: disable sync warp shuffel operators from clr amd_warp_sync_functions.h (#15273)

Browse Source
This commit is contained in:
uvos
2025-08-12 22:15:12 +02:00
committed by GitHub
parent f4586ee598
commit b0493156fa
2 changed files with 1 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

1
ggml/src/ggml-cuda/fattn-wmma-f16.cu
View File

@@ -15,7 +15,6 @@ namespace wmma = mtmusa::wmma;
namespace wmma = nvcuda::wmma;
#endif // GGML_USE_MUSA
#elif defined(GGML_HIP_ROCWMMA_FATTN) && defined(FP16_MMA_AVAILABLE)
#undef HIP_ENABLE_WARP_SYNC_BUILTINS // conflicts with rocWMMA headers
#include <rocwmma/rocwmma.hpp>
namespace wmma = rocwmma;
#endif // !defined(GGML_USE_HIP)

2
ggml/src/ggml-cuda/vendors/hip.h vendored
View File

@@ -1,6 +1,6 @@
#pragma once
#define HIP_ENABLE_WARP_SYNC_BUILTINS 1
#define HIP_DISABLE_WARP_SYNC_BUILTINS 1
#include <hip/hip_runtime.h>
#include <hipblas/hipblas.h>
#include <hip/hip_fp16.h>