Unverified50% confidenceBenchmarkExact time
In this int8 workload, both 24 GB cards and the L40S took 3.97 seconds per step, while the H100 PCIe required only 2.99 seconds
1
Sources
50%
Confidence
Long-term
Relevance
9/12/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedIn real-world testing on an RTX 3060 (6GB VRAM) with 32GB RAM running the pre-quantized Krea2 Turbo, 1024×1024 resolution with 8-step Euler sampling achieved a speed of 1.78 s/it and a total elapsed time of 17.64 seconds.72% similarUnverifiedCollapse推理速度：CPU上单条10-token上下文约0.23毫秒，MPS在batch 1024下约230万token/秒，最近邻查询约0.48毫秒72% similarUnverified对于70B模型（序列长度2048、隐藏层维度8192），每层激活值约32MB，在InfiniBand(400Gbps)中约需0.6毫秒，而在50Mbps家庭宽带下需超过5秒71% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/916657API
curl https://kongchang.com/api/v1/knowledge/claims/916657MCP
get_claim(id=916657)