benchmark result I don't trust: early impressions

[kernelarch]

I'm trying to sanity-check my setup.

I'm mostly working with DeepSeek in vLLM on GB200. The part I'm trying to understand is tokens/sec.

Not looking for the perfect answer, mostly what worked in real use.

[plainline]

My first guess would be this: On mini PC, the easiest win is often to back off the aggressive settings and see whether the problem survives. Once you know the plain path works, then add the faster path back in piece by piece.

[batchpeak]

worth checking the driver first.

[batchsize]

subscribed, curious where this lands.

If you post your exact settings, people here can probably narrow it down pretty fast.

[cacheline]

My first guess would be this: On A6000, 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.

[routearch]

That matches what I was seeing. I don't think your conclusion is far off. The biggest change for me was simplifying the stack instead of adding one more optimization layer.

[batchsize]

Need a little more detail here. Are you testing this on A6000 or something else?

[deltastation]

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.

[kernelghost]

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.

Maybe, but I had the opposite result. The reported numbers can look compelling, but in ordinary usage I didn't see the same benefit. Not saying you're wrong, just that I wouldn't lock that in as the lesson yet.

[cacheline]

A couple things I'd check first. If you're on Ollama, I'd confirm the version first and then reduce variables until the behavior is obvious. Not glamorous advice, but repeatable baselines save a lot of time.