MiniMax says AMD MI355X is now close to Nvidia B200 in model serving
hongyangzh · x · 2026-07-24
MiniMax says AMD MI355X is nearing Nvidia B200 in serving
A repost from MiniMax claims AMD MI355X is now close to Nvidia B200 for serving MiniMax-M3.
The company says it published two technical blog posts describing a full-stack co-optimization of its ATOM + ATOMesh serving system for a 428B sparse MoE multimodal base model.
The post highlights several engineering changes:
- EAGLE3 speculative decoding to reduce generation latency
- MSA-aligned Page-16 KV cache for 1M ultra-long context
- Layer-wise FP8 quantization to avoid precision loss in vision/MoE components
- Distributed P/D separation for large-scale clusters
The main takeaway is that AMD hardware is getting closer to Nvidia performance in a real large-scale serving setup.
Related event: MiniMax Claims AMD MI355X Serving Performance Nears Nvidia B200(2 posts)→
More from Infra
- Engram's random reads don't suit SSDs; CPU-memory over NVLink could serve all 72 GPUs — bookwormengr · 2026-09-11
- 80% of the DIY LLM inference hype posters have already quit — it's brutally hard systems work — abhijithneil · 2026-09-11
- Hugging Face's Ultra Scale Playbook: a free book on training LLMs on GPU clusters — mdancho84 · 2026-09-11
- Is inference latency becoming the biggest bottleneck for production AI agents? — Euphoric_Sea632 · 2026-09-11
- LLM Serving Metrics Thread: Why TPOT and Uptime Make or Break User Experience — abhijithneil · 2026-09-11
- PlanetScale launches sharded Postgres: 768 servers acting as one, 1PB scale — dhruv2038 · 2026-09-11