Write-ups of what I built and how it went.
After benchmarking halogen-flash-server against Nathanw v0.7.5 on the same Strix Halo box, halogen wins cold prefill but loses warm decode at production agent depth. Community 2–4× claims compare against the wrong baseline.
read more →Following up on the hipCUB fix post, I built and benchmarked EngramHalo.cpp, a llama.cpp fork with RDNA 3.5 kernel patches and MTP speculative decoding. The result: 28-38 tok/s decode on the same hardware and quant, nearly doubling the previous 15-20 tok/s baseline.
read more →A complete guide to running Qwen 3.8 Flash Next (125B MoE, 6B active) on the AMD Strix Halo 128 GB platform. Covers quant selection, the critical hipCUB fix, chat template traps, and the exact llama-swap config to reproduce 15-20 t/s decode across long contexts.
read more →