ggml-cpu: remove the per-matmul global barrier in ggml_compute_forward_mul_mat by codspeed-hq[bot] · Pull Request #38 · AvalancheHQ/llama.cpp

Performance Gate Passed

✅ 28 untouched benchmarks

_{Comparing codspeed-optim-remove-the-per-matmul-global-barrier-in-ggml-compu-1785318592407 (5bb839b) with master (46819c9)}