hrx: Q4_K batches off the 256 grid can read the packed layout, so a weight keeps one device copy (engine #2137, opt-in) - #308

hrx: Q4_K batches off the 256 grid can read the packed layout, so a weight keeps one device copy (engine #2137, opt-in)#308

bong-water-water-bong merged 1 commit into

1bit/hrx-vulkan-patched1bit-MONSTER/llama.cpp:1bit/hrx-vulkan-patchedfrom

1bit/2137-q4k-packed-tail1bit-MONSTER/llama.cpp:1bit/2137-q4k-packed-tailCopy head branch name to clipboard