llama.cpp
vulkan: use fp32 in coopmat2 q4_k dequant function
#12309

Merged

Commits

vulkan: Adjust coopmat2 tile sizes and selection heuristic

jeffbolznv committed 189 days ago
vulkan: Pad N dimension of B matrix for coopmat2 perf, to avoid bounds checking

jeffbolznv committed 189 days ago
vulkan: use fp32 in coopmat2 q4_k dequant function

jeffbolznv committed 189 days ago

Loading