GPU-World Runs Near Optimal: MFU Above 30% Is Common, Argues cis_female
cis_female · x · 2026-10-30
In a debate over inference efficiency, cisfemale counters that CPU-world work is small and software complicated, so nothing runs near optimal — but GPU-world work is expensive and software simple, so things run quite close to optimal, with MFU above 30% being common. A hardware-level counterpoint to the local-vs-shared inference argument.
More from Infra
- You can fingerprint your exact GPU with a single torch.index_add_ call — CatAstro_Piyush · 2026-10-31
- SGLang boosts DeepSeek V4.1 Flash throughput 67% to 1.09M tok/s per GPU in 4 days — BanghuaZ · 2026-10-31
- llama.cpp ROCm regression: prompt processing drops from ~1200 to ~460 t/s on RX 9070XT — uber-linny · 2026-10-31
- Sub-scale neocloud warning: VC-backed GPU cloud startups reportedly struggling — AccBalanced · 2026-10-31
- Qwen4 with HC architecture could widen the Q4 vs Q6 quantization gap significantly — Ok-Shower7286 · 2026-10-31
- Too many inference providers, too little demand: a developer's take — AccBalanced · 2026-10-31