working config for 32k context for Llama

[rocmhouse90]

Appreciate the detail. That matches what I saw on 7900 XTX.

I tested this tonight. I couldn't reproduce the slowdown until I pushed context much higher. The main thing that moved was latency, not quality.

[relayforge]

Appreciate the detail. That matches what I saw on 7900 XTX.

Interesting, I hadn't tried that. I ended up in about the same place using vLLM.

[cozysmith]

Good point. I ended up in about the same place using llama.cpp.

[llmpeak]

Good point. I ended up in about the same place using llama.cpp.

Need a little more detail here. Is this local-only or are you routing anything externally?

[bytehall]

Did a small comparison this afternoon. The issue went away once I changed one setting and restarted the stack cleanly. The surprise was that memory pressure showed up before throughput collapsed.