llama.cpp patch reports 25% Qwen3.5 prefill gain in MI100 test
The merged change raised Q5_K_M throughput from 422.29 to 529.34 tokens per second at a 512-token microbatch in the contributor's single-GPU test.
By RuntimeWire Staff · Published

The merged change raised Q5_K_M throughput from 422.29 to 529.34 tokens per second at a 512-token microbatch in the contributor's single-GPU test.