best small quant that still feels smart on 4090

[llmarch51]

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

After a few runs on 5090, I ended up with decent performance, but the setup only really made sense once I adjusted around embeddings.

Happy to post more details if anyone wants them.

[vrambandit]

I tested this tonight. On GB200 with MLX, I saw basically the same behavior. I expected the model choice to matter more than it did.

[llmarch51]

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

[bytehall]

Thanks, that helps. 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. I keep promising myself this is the last rebuild.

[marbletrack]

Thanks, that helps. I don't think your conclusion is far off. The biggest change for me was simplifying the stack instead of adding one more optimizat

What worked for me was pretty simple. On M3 Max, 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.

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

[cacheworks]

[quote]Thanks, that helps. I don't think your conclusion is far off. The biggest change for me was simplifying the stack instead of adding one more op

I'd start with the simpler stack.