Curious what people here are seeing.

I'm trying to decide whether to keep the current stack or change course. Main question is how people here are handling HBM constraints.

Would appreciate practical advice rather than theory.

Thought I'd post a small recipe because a lot of threads ask similar things.

Just sharing a setup that ended up working reasonably well: DeepSeek, SGLang, M3 Max, with a lot of attention on batch size.

Feel free to tear it apart if you see something dumb.

This is mostly for local work rather than public serving.

Curious what people here are seeing.

I'm trying to decide whether to keep the current stack or change course. Main question is how people here are handling NVFP4.

If you've tested something similar, what did you keep and what did you stop bothering with? At this point the real bottleneck is my patience.

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

One thing I keep noticing is that people underestimate long context 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.

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

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

I tried this with Llama using PyTorch on M3 Max. The headline result was better than I expected, but consumer GPUs still ended up being the limiting factor.

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

Thought I'd post a small recipe because a lot of threads ask similar things.

Maybe useful as a starting point: Ollama on 7900 XTX, keeping INT4 in mind from the start.

Feel free to tear it apart if you see something dumb.

My use case is pretty ordinary: local chat, some document work, and occasional batch jobs.

Wanted to compare notes with people doing similar things.

My current setup is LM Studio with 7900 XTX, and I keep running into token cost.

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

For context, I'm not chasing leaderboard numbers. I mostly want the system to feel consistent day to day.

I expected the biggest difference to come from the model, but the stack around it mattered at least as much.

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

Two paths both seem viable, but they break down differently once prefill/decode split becomes important.

Trying to avoid rebuilding the whole thing twice.

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 quant drift.

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

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

The more I work with local setups, the more it seems like small local stacks decides everything downstream.

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