jdelony/ik_llama.cpp
1
0
Fork 0
mirror of https://github.com/ikawrakow/ik_llama.cpp.git synced 2026-06-28 04:30:15 -05:00
Code Issues Packages Projects Releases Wiki Activity

CUDA FA: cover Gemma4-4B/2B assistant

Browse Source
This commit is contained in:
Kawrakow 2026-06-08 05:37:15 +00:00
parent 3c0f7b2f47
commit 9b4b9ca4ae
1 changed files with 8 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

9
ggml/src/ggml-cuda/fattn-new-mma.cu
View File

@ -2122,7 +2122,10 @@ template <int DKQ, int DV, int ncols2>
static void ggml_cuda_flash_attn_ext_mma_f16_switch_ncols1(ggml_backend_cuda_context & ctx, ggml_tensor * dst) {
const ggml_tensor * Q = dst->src[0];
if constexpr ((DKQ == 576 || DKQ == 512) && ncols2 <= 4) {
if constexpr (DKQ == 512 && ncols2 == 2) {
ggml_cuda_flash_attn_ext_mma_f16_case<DKQ, DV, 8, ncols2>(ctx, dst);
}
else if constexpr ((DKQ == 576 || DKQ == 512) && ncols2 <= 4) {
ggml_cuda_flash_attn_ext_mma_f16_case<DKQ, DV, 4, ncols2>(ctx, dst);
} else {
@ -2269,6 +2272,10 @@ void ggml_cuda_flash_attn_ext_mma_new(ggml_backend_cuda_context & ctx, ggml_tens
else if (gqa_ratio % 4 == 0) {
ggml_cuda_flash_attn_ext_mma_f16_switch_ncols1<512, 512, 4>(ctx, dst);
}
else if (gqa_ratio % 2 == 0) {
// Gemma4-4B assistant
ggml_cuda_flash_attn_ext_mma_f16_switch_ncols1<512, 512, 2>(ctx, dst);
}
else {
GGML_ABORT("Fatal error");
}