Unverified50% confidenceFactExact time
NVIDIA H100/H200系列GPU基于Hopper架构，集成专为Transformer模型优化的Tensor Core，单卡售价超过3万美元
1
Sources
50%
Confidence
Long-term
Relevance
7/19/2026
First Seen
Sources
Related Claims
UnverifiedAfter 2020, NVIDIA launched the A100 and H100 series GPUs, whose tensor core matrix multiplication acceleration design is well-suited to Transformer workloads.77% similarUnverifiedThe H100 GPU is based on the Hopper architecture; its Transformer Engine can dynamically switch between FP8 and FP16 precision, boosting throughput to approximately 2–3 times that of the A100.73% similarPartially VerifiedNVIDIA's MIG technology enables hardware-level resource isolation on high-end GPUs such as the A100/H100.72% similarVerifiedThe H100 GPU adopts NVIDIA's Hopper architecture, with a single-card peak compute of 3,958 TFLOPS (FP16 precision).72% similarUnverifiedH100芯片由英伟达于2022年推出，采用Hopper架构，相比A100在大规模Transformer训练上吞吐量提升约3倍，单卡售价曾高达4万美元71% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/561932API
curl https://kongchang.com/api/v1/knowledge/claims/561932MCP
get_claim(id=561932)