ggml-hrx: fence global F32 output before Qwen Q8 packing - #174

ggml-hrx: fence global F32 output before Qwen Q8 packing#174

AaronStGeorge merged 4 commits into

hrx-graph-develop-v2AMD-Ecosystem/llama.cpp:hrx-graph-develop-v2from

users/astgeorg/pr-qwen3-coder-q8-publicationAMD-Ecosystem/llama.cpp:users/astgeorg/pr-qwen3-coder-q8-publicationCopy head branch name to clipboard