Lumpiasty/llama.cpp
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

CUDA: fix crash on uneven context without FA (#16988)

Browse Source
This commit is contained in:
Johannes Gäßler
2025-11-06 14:05:47 +01:00
committed by GitHub
parent 5b180c3d60
commit aa374175c3
7 changed files with 44 additions and 38 deletions
Download Patch File Download Diff File Expand all files Collapse all files
ggml/src/ggml-cuda/mmf.cu
+9 -3
View File
@@ -119,15 +119,21 @@ void ggml_cuda_mul_mat_f(ggml_backend_cuda_context & ctx, const ggml_tensor * sr
}
}
bool ggml_cuda_should_use_mmf(enum ggml_type type, int cc, int warp_size, const int64_t * src0_ne, const int src1_ncols, bool mul_mat_id) {
bool ggml_cuda_should_use_mmf(enum ggml_type type, int cc, int warp_size, const int64_t * src0_ne,
const size_t * src0_nb, const int src1_ncols, bool mul_mat_id) {
if (ggml_is_quantized(type)) {
return false;
}
if (src0_ne[0] % (warp_size * (4/ggml_type_size(type))) != 0) {
const size_t ts = ggml_type_size(type);
if (src0_ne[0] % (warp_size * (4/ts)) != 0) {
return false;
}
for (size_t i = 0; i < GGML_MAX_DIMS; ++i) {
if (src0_nb[i] % (2*ts) != 0) {
return false;
}
}
if (src0_ne[1] % MMF_ROWS_PER_BLOCK != 0) {
return false;
}