EngineX-Ascend/enginex-ascend-910-llama.cpp
10
0
Fork 0
Code Issues Pull Requests Actions 4 Projects Releases Wiki Activity

vulkan: fix noncontig check for mat_mul_id splitting (#14683)

Browse Source 
* vulkan: fix noncontig check for mat_mul_id splitting

Remove supports_op check for > 4096 (splitting fixes this)

* vulkan: fix batched matmul dequant for Q*_K
This commit is contained in:
Jeff Bolz
2025-07-15 14:51:09 -05:00
committed by GitHub
parent 10a0351a97
commit ba1ceb3456
6 changed files with 6 additions and 10 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
ggml/src/ggml-vulkan/vulkan-shaders/dequant_q5_k.comp
View File

@@ -10,7 +10,7 @@ layout (binding = 1) writeonly buffer D {D_TYPE data_b[];};
void main() {
[[unroll]] for (uint wgy = 0; wgy < 256; wgy++) {
const uint ib = gl_WorkGroupID.x * 256 + wgy;
if (ib >= p.M * p.K / QUANT_K) {
if (ib >= p.nel / QUANT_K) {
return;
}