tqcq/llama.cpp
0
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-06-26 19:55:04 +00:00
Code Issues Packages Projects Releases Wiki Activity

CUDA: fix race condition in FA vector kernels (#13742)

Browse Source
This commit is contained in:
Johannes Gäßler
2025-05-24 11:46:19 +02:00
committed by GitHub
parent b775345d78
commit ffd0eae60b
2 changed files with 2 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

1
ggml/src/ggml-cuda/fattn-vec-f16.cuh
View File

@ -212,6 +212,7 @@ static __global__ void flash_attn_vec_ext_f16(
}
}
if (__all_sync(0xFFFFFFFF, skip)) {
__syncthreads();
continue;
}
#endif // GGML_USE_HIP

1
ggml/src/ggml-cuda/fattn-vec-f32.cuh
View File

@ -217,6 +217,7 @@ static __global__ void flash_attn_vec_ext_f32(
}
}
if (__all_sync(0xFFFFFFFF, skip)) {
__syncthreads();
continue;
}
#endif // GGML_USE_HIP