Unverified50% confidenceBenchmarkExact time
Compared with the full-rank (no low-rank approximation) approach, VPU at dim=256 reduced 200 steps from 4249ms to 160ms, achieving a 26.44x speedup.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/12/2026
First Seen
Valid until: 1/10/2027
Sources
Related Entities
Related Claims
UnverifiedIn measurements at dim=256 and dim=512, VPU's average latency is about 0.32ms in both cases, with equivalent FPS of about 3088-3099.78% similarUnverifiedVPU latency is about 0.32 ms/step, with throughput exceeding 3000 FPS.71% similarUnverifiedThe typical generation speed of an NVIDIA RTX 4090 running a similarly scaled Q4 quantized model is in the range of 40–60 TPS70% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1008098API
curl https://kongchang.com/api/v1/knowledge/claims/1008098MCP
get_claim(id=1008098)