突發訊息:MiniMax M3.1 Flash 推出不到 3 週,NVIDIA 和 @Inferact 已提交首份 vLLM Rubin NVL72 成果,達到最先進水準,且每單位 TCO 的效能比 GB300 NVL72 高 6.5x。
很高興看到 Rubin 已透過與 @vllm_project 合作,逐漸成熟並完成更多整合!
這份 Rubin 成果採用 CUDA 13.4 全新的 locality-aware MoE 功能,可降低延遲,並將實際 HBM 頻寬提高約 1.2x。
AI 伴隨問答
針對這篇報導有疑問?可直接點選快捷問題或向 AI 追問背景脈絡與技術細節:
BREAKING NEWS: Less than 3 weeks since the MiniMax M3.1 Flash launch, NVIDIA and @Inferact have already submitted an initial vLLM Rubin NVL72 submission that is state-of-the-art and delivers 6.5x better perf per TCO than GB300 NVL72.
Great to see that Rubin is already becoming more and more mature and integrated in partnership with @vllm_project!
This Rubin submission uses the new locality-aware MoE feature in CUDA 13.4, which reduces latency and increases realized HBM bandwidth by ~1.2x.