Thread Rating:
  • 0 Vote(s) - 0 Average
  • 1
  • 2
  • 3
  • 4
  • 5
synthetic data for support bots: setup advice
#1
Wanted to compare notes with people doing similar things.

My current setup is LM Studio with dual 3090, and I keep running into HBM constraints.

Not looking for the perfect answer, mostly what worked in real use.
serving stack tinkerer
Reply
#2
that fixed it for me too.
Reply
#3
I'd approach it in a boring order. Start by verifying the driver/runtime combo, then test with a smaller prompt, then look at workstation thermals. Not glamorous advice, but repeatable baselines save a lot of time.
Reply
#4
A couple things I'd check first. Start by verifying the driver/runtime combo, then test with a smaller prompt, then look at batch size.
Reply
#5
Did a small comparison this afternoon. The issue went away once I changed one setting and restarted the stack cleanly.
curious and caffeinated
Reply
#6
Need a little more detail here. Is this local-only or are you routing anything externally? Turns out 'just one more test' was a lie.
routing beats arguing over one perfect model
Reply
#7
Quote:Need a little more detail here. Is this local-only or are you routing anything externally? Turns out 'just one more test' was a lie.

That matches what I was seeing. That matches what I saw on GB200. I got better results once I stopped chasing one metric and just looked at whether the workflow felt usable.
Reply
#8
Quote:
Quote:Need a little more detail here. Is this local-only or are you routing anything externally? Turns out 'just one more test' was a lie.

A couple things I'd check first. On 5090, 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.

There are a few different conversations hiding inside this thread, so being explicit about the goal helps.
throughput matters, but tail latency matters too
Reply
#9
worth checking the driver first.
curious and caffeinated
Reply
#10
Interesting, I hadn't tried that. 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. The logs were somehow calmer than I was.
Reply


Forum Jump:


Users browsing this thread: 1 Guest(s)