+- Inference (https://www.datasculpting.ai/inference)
+-- Forum: Inference Engines (https://www.datasculpting.ai/inference/forumdisplay.php?fid=9)
+--- Forum: Ollama / LM Studio / Desktop Runtimes (https://www.datasculpting.ai/inference/forumdisplay.php?fid=38)
+--- Thread: Ollama for power users on 5090 (/showthread.php?tid=782)
Ollama for power users on 5090 - fusedhall - 10-05-2026
Did a bit of testing and thought I'd post the result here.
Short version: TensorRT-LLM behaved well for me, but only after I stopped ignoring structured output.
Happy to post more details if anyone wants them.
Re: Ollama for power users on 5090 - mangofield33 - 10-06-2026
Can you fill in a bit more context? Are you testing this on dual 3090 or something else?
Re: Ollama for power users on 5090 - localcraft17 - 10-06-2026
A couple things I'd check first. Start by verifying the driver/runtime combo, then test with a smaller prompt, then look at tail latency.
Re: Ollama for power users on 5090 - cudapeak - 10-07-2026
I don't think that's the whole story. The reported numbers can look compelling, but in ordinary usage I didn't see the same benefit.
Re: Ollama for power users on 5090 - cacheline - 10-08-2026
worth checking the driver first.
Re: Ollama for power users on 5090 - edgekite - 10-08-2026
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.
Re: Ollama for power users on 5090 - vectorhouse71 - 10-10-2026
That matches what I was seeing. I ended up in about the same place using MLX. Worth rechecking your defaults though, because one bad assumption can make the rest of the thread look confusing.
Re: Ollama for power users on 5090 - latentscope35 - 10-11-2026
I hit almost the same issue last night.