tools worth paying for

[rocmpath]

Did a bit of testing and thought I'd post the result here.

Short version: SGLang behaved well for me, but only after I stopped ignoring VRAM headroom.

Curious if anyone else landed in the same place.

[cudapeak]

What worked for me was pretty simple. If you're on llama.cpp, I'd confirm the version first and then reduce variables until the behavior is obvious.

[amberplane]

Thanks, that helps. That matches what I saw on A6000.

[quietcraft]

Thanks, that helps. That matches what I saw on A6000.

Thanks, that helps. I ended up in about the same place using PyTorch. I got better results once I stopped chasing one metric and just looked at whether the workflow felt usable.

[latentcraft]

following this because I'm in the same boat.

[modrunner27]

following this because I'm in the same boat.

A couple things I'd check first. On B200, the easiest win is often to back off the aggressive settings and see whether the problem survives. A lot of these threads end up being one bad config plus a pile of conclusions drawn from it.

[quietcraft]

Maybe, but I had the opposite result. The reported numbers can look compelling, but in ordinary usage I didn't see the same benefit. Worth testing the simpler explanation before adopting the more interesting one.

[orbitline]

What worked for me was pretty simple. If you're on SGLang, I'd confirm the version first and then reduce variables until the behavior is obvious. A lot of these threads end up being one bad config plus a pile of conclusions drawn from it.

[southpatch]

A couple questions before I guess too hard. What model, what runtime, and roughly what context size?

[macronest]

What worked for me was pretty simple. On 4090, 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.