Unverified50% confidenceBenchmarkExact time
NVIDIA H100的屋脊点约为597 FLOP/Byte（约2000 TFLOPS ÷ 3.35 TB/s）
1
Sources
50%
Confidence
Long-term
Relevance
7/15/2026
First Seen
Sources
Related Claims
VerifiedNVIDIA H100单卡FP64双精度浮点性能达到约34 TFLOPS80% similarVerifiedNVIDIA A100单卡拥有80GB HBM2e显存和312 TFLOPS的FP16算力78% similarUnverifiedNVIDIA A100、H100系列GPU单卡功耗可达300-700瓦75% similarUnverifiedNVIDIA H100单卡热设计功耗（TDP）约700W，一个万卡集群的理论峰值功耗超过7兆瓦73% similarUnverifiedNVIDIA RTX 5090基于Blackwell架构，配备32GB GDDR7显存，理论FP16算力超过100 TFLOPS72% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/512116API
curl https://kongchang.com/api/v1/knowledge/claims/512116MCP
get_claim(id=512116)