Unverified50% confidenceBenchmarkExact time
V100的Tensor Core在FP16精度下峰值算力可达125 TFLOPS,是前代Pascal架构的5倍以上
1
Sources
50%
Confidence
Long-term
Relevance
7/22/2026
First Seen
Sources
Related Claims
UnverifiedH100的第四代Tensor Core原生支持FP8,峰值算力达3958 TFLOPS,是FP16的两倍80% similarVerifiedH100 GPU采用英伟达Hopper架构,单卡峰值算力达3,958 TFLOPS(FP16精度)79% similarUnverified原生 FP8 硬件加速最早在 NVIDIA Hopper 架构(H100,2022年)的第四代 Tensor Core 中实现,峰值算力达 3958 TFLOPS77% similarUnverifiedcuBLAS库的矩阵乘法实现可以接近V100 GPU理论峰值的125 TFLOPS,与朴素三重循环实现相差超过100倍75% similarUnverifiedTensorRT支持FP32/FP16/INT8/FP8多种精度校准75% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/586043API
curl https://kongchang.com/api/v1/knowledge/claims/586043MCP
get_claim(id=586043)