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

[SYCL] Support q2 mul_mat (#26231)

Browse Source 
* support q2_0 in mul_mat

* support more q2_0 case
This commit is contained in:
Neo Zhang
2026-07-31 09:19:41 +03:00
committed by GitHub
parent 1553725965
commit a2be61dc87
4 changed files with 174 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files
ggml/src/ggml-sycl/convert.cpp
+4
View File
@@ -644,6 +644,8 @@ to_fp16_sycl_t ggml_get_to_fp16_sycl(ggml_type type, ggml_tensor * dst) {
switch (type) {
case GGML_TYPE_Q1_0:
return dequantize_block_sycl<QK1_0, QR1_0, dequantize_q1_0>;
case GGML_TYPE_Q2_0:
return dequantize_block_sycl<QK2_0, QR2_0, dequantize_q2_0>;
case GGML_TYPE_Q4_0:
if (dst->src[0]->extra &&
((ggml_tensor_extra_gpu*)dst->src[0]->extra)->optimized_feature.reorder) {
@@ -728,6 +730,8 @@ to_fp32_sycl_t ggml_get_to_fp32_sycl(ggml_type type, ggml_tensor *dst) {
switch (type) {
case GGML_TYPE_Q1_0:
return dequantize_block_sycl<QK1_0, QR1_0, dequantize_q1_0>;
case GGML_TYPE_Q2_0:
return dequantize_block_sycl<QK2_0, QR2_0, dequantize_q2_0>;
case GGML_TYPE_Q4_0:
if (dst->src[0]->extra &&
((ggml_tensor_extra_gpu*)dst->src[0]->extra)->optimized_feature.reorder) {