tqcq/llama.cpp
0
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-08-11 11:05:39 -04:00
Code Issues Packages Projects Releases Wiki Activity

cuda: Add Q5_1, Q5_0, Q4_1 and Q4_0 to F32 conversion support. (#12000)

Browse Source
This commit is contained in:
Gian-Carlo Pascutto
2025-02-22 09:43:24 +01:00
committed by GitHub
parent de8b5a3624
commit d70908421f
2 changed files with 103 additions and 6 deletions
Download Patch File Download Diff File Expand all files Collapse all files

12
ggml/src/ggml-cuda/ggml-cuda.cu
View File

@@ -3075,15 +3075,27 @@ static bool ggml_backend_cuda_device_supports_op(ggml_backend_dev_t dev, const g
if (src0_type == GGML_TYPE_F32 && src1_type == GGML_TYPE_Q4_0) {
return true;
}
if (src0_type == GGML_TYPE_Q4_0 && src1_type == GGML_TYPE_F32) {
return true;
}
if (src0_type == GGML_TYPE_F32 && src1_type == GGML_TYPE_Q4_1) {
return true;
}
if (src0_type == GGML_TYPE_Q4_1 && src1_type == GGML_TYPE_F32) {
return true;
}
if (src0_type == GGML_TYPE_F32 && src1_type == GGML_TYPE_Q5_0) {
return true;
}
if (src0_type == GGML_TYPE_Q5_0 && src1_type == GGML_TYPE_F32) {
return true;
}
if (src0_type == GGML_TYPE_F32 && src1_type == GGML_TYPE_Q5_1) {
return true;
}
if (src0_type == GGML_TYPE_Q5_1 && src1_type == GGML_TYPE_F32) {
return true;
}
if (src0_type == GGML_TYPE_F32 && src1_type == GGML_TYPE_IQ4_NL) {
return true;
}