+- Inference (https://www.datasculpting.ai/inference)
+-- Forum: Evaluation & Testing (https://www.datasculpting.ai/inference/forumdisplay.php?fid=16)
+--- Forum: Benchmarks (https://www.datasculpting.ai/inference/forumdisplay.php?fid=65)
+--- Thread: comparing models without fooling yourself on Threadripper box (/showthread.php?tid=647)
Pages:
1
2
Re: comparing models without fooling yourself on Threadripper box - benchplease - 10-07-2026
Thanks, that helps. That matches what I saw on dual 3090.
Re: comparing models without fooling yourself on Threadripper box - sparrowworks88 - 10-07-2026
Tried to reproduce it just now. On mini PC with Ollama, I saw basically the same behavior.
Re: comparing models without fooling yourself on Threadripper box - batchsize - 10-07-2026
Good point. That matches what I saw on Threadripper box. I got better results once I stopped chasing one metric and just looked at whether the workflow felt usable.
I'm happy to test one more variant if you want a second data point.
Re: comparing models without fooling yourself on Threadripper box - kernelghost - 10-07-2026
Quote:Good point. That matches what I saw on Threadripper box. I got better results once I stopped chasing one metric and just looked at whether the workflo
Appreciate the detail. I ended up in about the same place using SGLang. Worth rechecking your defaults though, because one bad assumption can make the rest of the thread look confusing.
Re: comparing models without fooling yourself on Threadripper box - deltastation - 10-07-2026
Ran a quick check on my box. The issue went away once I changed one setting and restarted the stack cleanly.
Re: comparing models without fooling yourself on Threadripper box - plainline - 10-07-2026
Quote:Ran a quick check on my box. The issue went away once I changed one setting and restarted the stack cleanly.
What worked for me was pretty simple. Start by verifying the driver/runtime combo, then test with a smaller prompt, then look at HBM constraints. A lot of these threads end up being one bad config plus a pile of conclusions drawn from it.