setup notes worth sharing: early impressions

[benchplease]

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

One thing I keep noticing is that people underestimate prefill/decode split until they actually try to run this stuff for more than one afternoon.

Curious whether that lines up with what other people are seeing.

[deltagrid]

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

[mangoquill]

worth checking the driver first.

[cachebench]

that fixed it for me too.

[mlxgarden]

I'd approach it in a boring order. On B200, the easiest win is often to back off the aggressive settings and see whether the problem survives.

[audioriver]

Interesting, I hadn't tried that. That matches what I saw on GB200. The biggest change for me was simplifying the stack instead of adding one more optimization layer.

[vramstream]

Ran a quick check on my box. On GB200 with Ollama, I saw basically the same behavior.

[quiethouse]

My first guess would be this: Before rebuilding anything, I'd test one clean baseline with FLUX and compare from there. Once you know the plain path works, then add the faster path back in piece by piece.

[deltagrid]

Good point. The pattern around reranking has been pretty similar for me. The biggest change for me was simplifying the stack instead of adding one more optimization layer.

[rocmcrate26]

Appreciate the detail. The pattern around quant drift has been pretty similar for me. Worth rechecking your defaults though, because one bad assumption can make the rest of the thread look confusing.