power delivery for 24/7 serving with vLLM

[nanobrook]

This keeps coming up in conversation so I figured I'd start a thread.

The more I work with local setups, the more it seems like MoE serving decides everything downstream.

Wondering if this is just where the whole space is heading now.

[batchyard]

This keeps coming up in conversation so I figured I'd start a thread.

Tried to reproduce it just now. My result was mixed: first run looked bad, second run looked normal. The main thing that moved was latency, not quality.

[tensorworks]

worth checking the driver first.

[bytegarden73]

I don't think that's the whole story. I get why that seems right, but on Threadripper box the bottleneck felt elsewhere. Worth testing the simpler explanation before adopting the more interesting one.

[cozygrid12]

I'd approach it in a boring order. Start by verifying the driver/runtime combo, then test with a smaller prompt, then look at small local stacks. Not glamorous advice, but repeatable baselines save a lot of time.

[bytegarden73]

My first guess would be this: On 7900 XTX, the easiest win is often to back off the aggressive settings and see whether the problem survives. Not glamorous advice, but repeatable baselines save a lot of time.

[phasenode]

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

[batchyard]

I don't think that's the whole story. I wouldn't optimize around that first. On my side the bigger swing came from the runtime and not the model. Could be a workload difference, but I wouldn't generalize too fast from one setup.

[nanobrook]

Thanks, that helps. I don't think your conclusion is far off. Worth rechecking your defaults though, because one bad assumption can make the rest of the thread look confusing.

[homelabhenry]

My first guess would be this: Start by verifying the driver/runtime combo, then test with a smaller prompt, then look at router policy. Not glamorous advice, but repeatable baselines save a lot of time.