70B still worth the pain

[fusedhall]

Thanks, that helps. I ended up in about the same place using TensorRT-LLM. Worth rechecking your defaults though, because one bad assumption can make the rest of the thread look confusing.

[llmtrack26]

Thanks, that helps. I ended up in about the same place using TensorRT-LLM. Worth rechecking your defaults though, because one bad assumption can make

Ran a quick check on my box. The issue went away once I changed one setting and restarted the stack cleanly. I expected the model choice to matter more than it did.

[nanopatch]

That matches what I was seeing. I don't think your conclusion is far off.

[relaynest82]

Interesting, I hadn't tried that. The pattern around batch size has been pretty similar for me. Worth rechecking your defaults though, because one bad assumption can make the rest of the thread look confusing.

[labnest]

What worked for me was pretty simple. Start by verifying the driver/runtime combo, then test with a smaller prompt, then look at MoE serving. A lot of these threads end up being one bad config plus a pile of conclusions drawn from it.

[macrotinker]

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

[paperrunner]

Appreciate the detail. The pattern around HBM constraints has been pretty similar for me. I got better results once I stopped chasing one metric and just looked at whether the workflow felt usable.

[dockkite69]

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

I tried this with Mixtral using LM Studio on Threadripper box. The headline result was better than I expected, but VRAM headroom still ended up being the limiting factor.

Nothing scientific here, just practical notes from a real setup.

[dockworks23]

What worked for me was pretty simple. Before rebuilding anything, I'd test one clean baseline with Gemma and compare from there. A lot of these threads end up being one bad config plus a pile of conclusions drawn from it.

[latentscope35]

following this because I'm in the same boat.