CS348Project/llama.cpp
5
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-11-04 09:32:00 +00:00
Code Issues Packages Projects Releases Wiki Activity

CUDA: fix shared memory access condition for mmv (#10740)

Browse Source
This commit is contained in:
Johannes Gäßler
2024-12-09 20:07:12 +01:00
committed by GitHub
parent c37fb4cf62
commit 26a8406ba9
1 changed files with 1 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
ggml/src/ggml-cuda/mmv.cu
View File

@@ -57,7 +57,7 @@ static __global__ void mul_mat_vec(
if (block_size > WARP_SIZE) { if (block_size > WARP_SIZE) {
buf_iw[tid/WARP_SIZE] = sumf; buf_iw[tid/WARP_SIZE] = sumf;
__syncthreads(); __syncthreads();
if (tid > WARP_SIZE) { if (tid >= WARP_SIZE) {
return; return;
} }
sumf = buf_iw[tid]; sumf = buf_iw[tid];