sycl: 2-3x prompt processing speedup by dequantizing PTQ1_0 and PQ2_0 to FP16 - #278

sycl: 2-3x prompt processing speedup by dequantizing PTQ1_0 and PQ2_0 to FP16#278

bri-prism merged 1 commit into

PrismML-Eng:prismPrismML-Eng/llama.cpp:prismfrom

dwymark:sycl-fp16-prompt-gemmdwymark/llama.cpp:sycl-fp16-prompt-gemmCopy head branch name to clipboard