70B still worth the pain with TensorRT-LLM

[oxiderunner]

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

It feels like the conversation has shifted hard toward inference and serving details lately, and honestly that makes sense once you start dealing with router policy.

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

I care more about stability than squeezing the last possible tokens/sec out of it.

[kernelmike]

Thanks, that helps. I don't think your conclusion is far off. I got better results once I stopped chasing one metric and just looked at whether the workflow felt usable.

[cobalttinker]

I think the bottleneck is somewhere else. 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.

[audionode8]

My first guess would be this: If you're on vLLM, I'd confirm the version first and then reduce variables until the behavior is obvious.

[edgeriver]

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

[orbitbench6]

Interesting, I hadn't tried that. 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.

[latentcraft]

Good point. That matches what I saw on 4090. The biggest change for me was simplifying the stack instead of adding one more optimization layer. The logs were somehow calmer than I was.

[kernelmike]

My first guess would be this: Start by verifying the driver/runtime combo, then test with a smaller prompt, then look at small local stacks. A lot of these threads end up being one bad config plus a pile of conclusions drawn from it.

[cudaline]

I hit almost the same issue last night.

[fastbench]

Maybe, but I had the opposite result. I wouldn't optimize around that first. On my side the bigger swing came from the runtime and not the model. Not saying you're wrong, just that I wouldn't lock that in as the lesson yet.