gufo Ships Pre-packaged Windows Build for AMD Strix Halo, 40 tps on Agentic Workloads
hiImMate · reddit · 2026-10-03
A Reddit user released an unofficial Windows port of gufo for AMD Strix Halo (gfx1151), now with a pre-packaged library requiring no build step, plus a start.cmd that autodiscovers supported quants. Supported: 3.8 Flash Next (UDQ4XL), 27b (UDQ4XL), and 35BA3B (UDQ8XL + TeilCoder). Flash Next reportedly holds 40 tps on agentic workloads at high context. Key tip: set VRAM to 96GB to avoid Windows shared-memory overhead. Changes are validated with logit checks for bit-identical output.
More from Infra
- Dev submits PR adding periodic checkpoints to GKE sandboxes so agents don't lose progress — yoimnotkesku · 2026-11-19
- AWS Lambda Adds GPU Support via Managed Instances for Inference and Media Workloads — DavidWells · 2026-11-19
- Nvidia Allegedly Made GPUs Artificially Scarce to Keep Prices Sky-High — GaryMarcus · 2026-11-19
- Leak Hints at Classified 1M-GPU Training Cluster in China That Skips HBM — teortaxesTex · 2026-11-19
- Tencent leaned on Japanese neocloud for AI compute; DeepSeek, Zhipu best positioned — teortaxesTex · 2026-11-19
- Open-source prefill engine for GLM-5.3-Flash beats vLLM by 3.2x on 4x B200 — CShorten30 · 2026-11-19