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

One thing I keep noticing is that people underestimate router policy until they actually try to run this stuff for more than one afternoon.

Wondering if this is just where the whole space is heading now.

Has anyone else run into this?

My current setup is llama.cpp with 4090, and I keep running into consumer GPUs.

If you've tested something similar, what did you keep and what did you stop bothering with?

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

The part that surprised me was how much the boring details mattered more than the model choice.

Spent some time on this today and wanted to compare notes.

I'm comparing two approaches here: keep it simple on Threadripper box, or squeeze more out of the stack and deal with the complexity.

Interested in tradeoffs more than fan loyalty.

Posting this in case it saves someone else a little time.

If you were starting from scratch today for this use case, what would you choose first and what would you ignore?

Would rather learn from your scars than collect my own all at once.

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 prefix caching.

Maybe obvious to some of you, but it clicked for me this week.

I figured I'd ask here before changing everything.

I'm mostly working with FLUX in MLX on 5090. The part I'm trying to understand is tokens/sec.

If you've tested something similar, what did you keep and what did you stop bothering with?

Has anyone else run into this?

I'm mostly working with Mixtral in PyTorch on Threadripper box. The part I'm trying to understand is token cost.

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

Posting this in case it saves someone else a little time.

I'm helping a small team set up a local workflow. We don't need a giant stack, but we do care about small local stacks.

Simple and boring is a valid answer.

Spent some time on this today and wanted to compare notes.

I've gone back and forth between PyTorch and TensorRT-LLM for Mixtral. The interesting difference for me has been structured output.

Interested in tradeoffs more than fan loyalty.

Posting this in case it saves someone else a little time.

I'm not trying to build a perfect architecture diagram. I just want a sane starting point around PyTorch and 4090.

Simple and boring is a valid answer.