vulkan: Block-load Q3_K/Q6_K block data and subtract on 32b ints - #23056

vulkan: Block-load Q3_K/Q6_K block data and subtract on 32b ints#23056

0cc4m merged 1 commit into

ggml-org:masterggml-org/llama.cpp:masterfrom

TheBlueMatt:2026-05-intel-q3-q6-mmvqTheBlueMatt/llama.cpp:2026-05-intel-q3-q6-mmvqCopy head branch name to clipboard