CS348Project/llama.cpp
5
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-11-02 09:12:03 +00:00
Code Issues Packages Projects Releases Wiki Activity

[CANN]MUL_MAT optimization (#12382)

Browse Source
This commit is contained in:
Chenguang Li
2025-03-15 09:31:08 +08:00
committed by GitHub
parent 9f2250ba72
commit 92a391327e
2 changed files with 6 additions and 7 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
ggml/src/ggml-cann/ggml-cann.cpp
View File

@@ -1689,11 +1689,6 @@ static bool ggml_backend_cann_supports_op(ggml_backend_dev_t dev,
case GGML_OP_MUL_MAT: {
switch (op->src[0]->type) {
case GGML_TYPE_Q8_0:
// Current groupsize should not be greater than k-1 in
// aclnnWeightQuantBatchMatmulV2GetWorkspaceSize
if (op->src[0]->ne[0] <= QK8_0) {
return false;
}
case GGML_TYPE_F16:
case GGML_TYPE_F32:
case GGML_TYPE_Q4_0: