vbatts/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

CUDA: fix MMQ for non-contiguous src0, add tests (#10021)

Browse source 
* CUDA: fix MMQ for non-contiguous src0, add tests

* revise test code
This commit is contained in:
Johannes Gäßler 2024-10-24 11:09:36 +02:00 • committed by GitHub
parent 0a1c750c80
commit c39665f589
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
4 changed files with 73 additions and 29 deletions
Download patch file Download diff file Expand all files Collapse all files

2
ggml/src/ggml.c
View file

@ -3464,7 +3464,7 @@ int64_t ggml_nrows(const struct ggml_tensor * tensor) {
size_t ggml_nbytes(const struct ggml_tensor * tensor) {
size_t nbytes;
size_t blck_size = ggml_blck_size(tensor->type);
const size_t blck_size = ggml_blck_size(tensor->type);
if (blck_size == 1) {
nbytes = ggml_type_size(tensor->type);
for (int i = 0; i < GGML_MAX_DIMS; ++i) {