+- Inference (https://www.datasculpting.ai/inference)
+-- Forum: LLM Models (https://www.datasculpting.ai/inference/forumdisplay.php?fid=12)
+--- Forum: Open Weight (https://www.datasculpting.ai/inference/forumdisplay.php?fid=49)
+---- Forum: Recipes & Configurations (https://www.datasculpting.ai/inference/forumdisplay.php?fid=52)
+---- Thread: simple router policy ideas for Phi (/showthread.php?tid=215)
Pages:
1
2
Re: simple router policy ideas for Phi - batchpeak - 09-17-2026
Quote:[quote][quote]A couple things I'd check first. Before rebuilding anything, I'd test one clean baseline with Phi and compare from there. Not glamorous
Tried to reproduce it just now. I couldn't reproduce the slowdown until I pushed context much higher. The surprise was that memory pressure showed up before throughput collapsed.
Re: simple router policy ideas for Phi - embershed99 - 09-18-2026
That matches what I was seeing. That matches what I saw on A6000. The biggest change for me was simplifying the stack instead of adding one more optimization layer.
Re: simple router policy ideas for Phi - kernelrunner - 09-18-2026
Good point. The pattern around consumer GPUs has been pretty similar for me. I got better results once I stopped chasing one metric and just looked at whether the workflow felt usable.