public_git/whisper.cpp
1
0
Fork 0
mirror of https://github.com/ggml-org/whisper.cpp.git synced 2026-10-02 20:41:54 +02:00
Code Issues Packages Projects Releases Wiki Activity

vulkan: fix flash attention dot product precision (llama/20589)

Browse Source
This commit is contained in:
Ruben Ortlam
2026-03-29 15:04:36 +03:00
committed by Georgi Gerganov
parent dae7781052
commit 724ea71cf9
1 changed files with 2 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files
ggml/src/ggml-vulkan/vulkan-shaders/flash_attn.comp
+2 -2
View File
@@ -245,7 +245,7 @@ void main() {
#endif
}
[[unroll]] for (uint32_t r = 0; r < rows_per_thread; ++r) {
Sf[r][c] += ACC_TYPE(dot(Q_cache[r], K_Tf));
Sf[r][c] += dot(ACC_TYPEV4(Q_cache[r]), ACC_TYPEV4(K_Tf));
}
}
}
@@ -270,7 +270,7 @@ void main() {
#endif
}
[[unroll]] for (uint32_t r = 0; r < rows_per_thread; ++r) {
Sf[r][c] += ACC_TYPE(dot(Qf[tile_row(r) * qf_stride + d * D_split + d_tid], K_Tf));
Sf[r][c] += dot(ACC_TYPEV4(Qf[tile_row(r) * qf_stride + d * D_split + d_tid]), ACC_TYPEV4(K_Tf));
}
}
}