DeepSeek-V3: From Roofline to Reality - A Performance Analysis Series
ezyang · x · 2026-09-02
ezyang announces a blog series, "DeepSeek-V3: from roofline to reality", digging into roofline analysis specifically for DSv3. The series aims to progressively refine a simplified performance model with real-world correction factors to explain actual PyTorch profile traces. It highlights DSv3 as a clean exemplar of a mixture-of-experts model.
Related event: ezyang launches DeepSeek-V3 roofline-to-reality blog series(2 posts)→
More from Infra
- SGLang Summit sponsors grow from 11 to 36 in one week; Lip-Bu Tan, Lilian Weng to speak — BanghuaZ · 2026-10-18
- Yacine jokes his datacenter KPI is gigawatt-hours: get paged if not burning enough power — yacineMTB · 2026-10-18
- 1.68x inference speedup drops, directly accelerating GRPO post-training — yacineMTB · 2026-10-18
- OpenRouter is only a fraction of the total addressable market, poster argues — sloppenheimer · 2026-10-18
- Developer builds change alerts for LLM prices and retirements across Bedrock, Azure and Vertex — jsandhu0780 · 2026-10-18
- 12.3M tokens in 27 minutes for $0.14: agent session with 99.6% cache hit — teortaxesTex · 2026-10-18