Unverified85% confidenceFactTime unknown
Blackwell架构的第五代RT Core将光线追踪性能提升约2倍，Tensor Core支持FP4精度运算使AI推理吞吐量翻倍
1
Sources
85%
Confidence
Long-term
Relevance
6/1/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedNVIDIA Blackwell架构（B200/GB200）支持FP4计算，理论上将Tensor Core吞吐量再次翻倍74% similarUnverifiedBlackwell是NVIDIA继Hopper和Ada Lovelace之后的新一代微架构，引入第五代Tensor Core以提升FP4/FP6推理效率73% similarUnverifiedBlackwell架构支持FP4精度计算，FP4是目前商用AI芯片支持的最低精度格式71% similarUnverifiedBlackwell架构引入第五代张量核心，原生支持FP4×FP4矩阵乘加运算，累加器使用FP32或FP16精度71% similarUnverifiedFP32到FP16的转换在现代GPU的Tensor Core上可实现2-4倍的吞吐量提升68% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/25924API
curl https://kongchang.com/api/v1/knowledge/claims/25924MCP
get_claim(id=25924)