llamaperf
Calculator
GPUs
Engines
Guides
Blog
Community
MCP
Advertise
Submit
Sign in
Submit
llamaperf
/
NVIDIA RTX 5070
/
report
Qwen3.8 27B
on
NVIDIA RTX 5070
Tone: mixed
NVIDIA hardware
Sep 7, 2026
Share
X
Reddit
LinkedIn
Bluesky
Copy
link
Throughput
3.0 t/s gen
Summary
User asks about Intel Arc B60 or B65 for Qwen 3.8 27B, currently getting 3 t/s on an RTX 5070. User mentions a custom vLLM fork for Intel Arc reaching about 20 t/s on the larger B70.
View source ↗
Feedback