rackmount vs tower for homelab inference: setup advice

[batchyard]

I figured I'd ask here before changing everything.

My current setup is SGLang with dual 3090, and I keep running into workstation thermals.

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

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

[sanevault]

I figured I'd ask here before changing everything.

What worked for me was pretty simple. On B200, 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.

[vectorcrate98]

That matches what I was seeing. That matches what I saw on 4090.

[edgequill]

That matches what I was seeing. The pattern around embeddings has been pretty similar for me. Worth rechecking your defaults though, because one bad assumption can make the rest of the thread look confusing.

[plainscope50]

That matches what I was seeing. 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.

There are a few different conversations hiding inside this thread, so being explicit about the goal helps.

[llm_wren]

I'd approach it in a boring order. On Threadripper box, the easiest win is often to back off the aggressive settings and see whether the problem survives. A lot of these threads end up being one bad config plus a pile of conclusions drawn from it.

[batchwire]

What worked for me was pretty simple. If you're on MLX, I'd confirm the version first and then reduce variables until the behavior is obvious.

[fastquill74]

subscribed, curious where this lands.

[vectorcrate98]

Not sure I agree there. I've seen people blame KV cache pressure when the real issue was much more mundane.

[cozygrid12]

Not sure I agree there. I've seen people blame KV cache pressure when the real issue was much more mundane.

I'd approach it in a boring order. Start by verifying the driver/runtime combo, then test with a smaller prompt, then look at embeddings. Once you know the plain path works, then add the faster path back in piece by piece.