Cointime update: Nvidia unveiled a full-stack AI memory co-design approach spanning GPU, CPU, networking, storage and software, saying it aims to raise memory efficiency rather than simply adding GPU memory capacity as HBM supply tightens and prices rise. The company said that in SemiAnalysis AgentX tests, GB300 NVL72 delivered up to 10x the per-cost inference throughput of H200 on certain workloads, while Vera Rubin NVL72 delivers up to 30x that of GB300 NVL72 with only about 3% more HBM per GPU.
Nvidia: Vera Rubin NVL72 Boosts Inference Throughput 30x on 3% More HBM
Share on WeChat
Open WeChat → Scan → then tap "…" to send to a chat or Moments.
Tap "…" in the top-right corner to send to a chat or share to Moments.
Related News
2h ago
HKMA Explores Agentic AI to Monitor Crypto Trades for Money Laundering
3h ago
Anthropic Expected to Price IPO Above Range at $80+ Per Share
3h ago
Coinbase AI Scam Defense Blocked $4.7M in Suspicious Transfers
4h ago
Nvidia CEO Jensen Huang Agrees to Testify at Senate Banking Committee
4h ago
Anthropic Existing Investors Plan Billions in IPO Share Purchase
5h ago
Analyst: Nvidia’s core business solid, AI infrastructure demand supports guidance