diff options
-rw-r--r-- | ggml/src/ggml-cuda/fattn-new-mma.cu | 2 |
1 files changed, 2 insertions, 0 deletions
diff --git a/ggml/src/ggml-cuda/fattn-new-mma.cu b/ggml/src/ggml-cuda/fattn-new-mma.cu index d1484451..8da96370 100644 --- a/ggml/src/ggml-cuda/fattn-new-mma.cu +++ b/ggml/src/ggml-cuda/fattn-new-mma.cu @@ -898,6 +898,8 @@ static __device__ __forceinline__ void flash_attn_ext_f16_process_tile( KQ_crs += __shfl_xor_sync(0xFFFFFFFF, KQ_crs, offset, WARP_SIZE); } + __syncthreads(); + // Write back combined meta data: #pragma unroll for (int imeta = 0; imeta < nmeta; ++imeta) { |