CS348Project/llama.cpp
5
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-10-30 08:42:00 +00:00
Code Issues Packages Projects Releases Wiki Activity

CUDA: fix broken oob check for FA vec f32 kernel (#7904)

Browse Source
This commit is contained in:
Johannes Gäßler
2024-06-12 17:41:51 +02:00
committed by GitHub
parent a9cae48003
commit 963552903f
1 changed files with 1 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
ggml-cuda/fattn-vec-f32.cuh
View File

@@ -149,7 +149,7 @@ static __global__ void flash_attn_vec_ext_f32(
for (int i0 = 0; i0 < D/2; i0 += WARP_SIZE) { for (int i0 = 0; i0 < D/2; i0 += WARP_SIZE) {
const int i = i0 + threadIdx.x; const int i = i0 + threadIdx.x;
Q_f2[j][i0/WARP_SIZE] = ncols <= 2 || ic0 + j ? Q_f2_j[i] : make_float2(0.0f, 0.0f); Q_f2[j][i0/WARP_SIZE] = ncols <= 2 || ic0 + j < ne01 ? Q_f2_j[i] : make_float2(0.0f, 0.0f);
Q_f2[j][i0/WARP_SIZE].x *= scale; Q_f2[j][i0/WARP_SIZE].x *= scale;
Q_f2[j][i0/WARP_SIZE].y *= scale; Q_f2[j][i0/WARP_SIZE].y *= scale;
} }