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

CUDA: use arch list for compatibility check (#11775)

Browse Source 
* CUDA: use arch list for feature availability check

---------

Co-authored-by: Diego Devesa <slarengh@gmail.com>
This commit is contained in:
Johannes Gäßler
2025-02-11 00:17:22 +01:00
committed by GitHub
co-authored by Diego Devesa
parent 7b891bdc86
commit b9ab0a4d0b
6 changed files with 80 additions and 24 deletions
Download Patch File Download Diff File Expand all files Collapse all files
ggml/src/ggml-cuda/convert.cu
+1 -1
View File
@@ -599,7 +599,7 @@ to_fp16_cuda_t ggml_get_to_fp16_cuda(ggml_type type) {
case GGML_TYPE_Q5_1:
return dequantize_block_cuda<QK5_1, QR5_1, dequantize_q5_1>;
case GGML_TYPE_Q8_0:
if (ggml_cuda_info().devices[ggml_cuda_get_device()].cc >= GGML_CUDA_CC_PASCAL) {
if (fp16_available(ggml_cuda_info().devices[ggml_cuda_get_device()].cc)) {
return dequantize_block_q8_0_f16_cuda;
}
return dequantize_block_cuda<QK8_0, QR8_0, dequantize_q8_0>;