+- Inference (https://www.datasculpting.ai/inference)
+-- Forum: Drivers / Frameworks / Toolkits (https://www.datasculpting.ai/inference/forumdisplay.php?fid=8)
+--- Forum: Apple / MLX (https://www.datasculpting.ai/inference/forumdisplay.php?fid=33)
+--- Thread: M3 Max for local use (/showthread.php?tid=332)
Pages:
1
2
M3 Max for local use - labnest - 09-17-2026
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 prefill/decode split decides everything downstream.
Wondering if this is just where the whole space is heading now.
The part that surprised me was how much the boring details mattered more than the model choice.
Re: M3 Max for local use - dockridge - 09-18-2026
Quote:This keeps coming up in conversation so I figured I'd start a thread.
Good point. I don't think your conclusion is far off.
Re: M3 Max for local use - thetaarch43 - 09-18-2026
I hit almost the same issue last night.
Re: M3 Max for local use - pixelcrate82 - 09-18-2026
My first guess would be this: If you're on Ollama, I'd confirm the version first and then reduce variables until the behavior is obvious. Once you know the plain path works, then add the faster path back in piece by piece.
Re: M3 Max for local use - amberhall - 09-18-2026
I tested this tonight. On 7900 XTX with LM Studio, I saw basically the same behavior. I expected the model choice to matter more than it did.
Re: M3 Max for local use - cudacrate - 09-18-2026
Appreciate the detail. That matches what I saw on M3 Max. I got better results once I stopped chasing one metric and just looked at whether the workflow felt usable.
Re: M3 Max for local use - circuitwire58 - 09-18-2026
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.
Re: M3 Max for local use - turboyard94 - 09-18-2026
Tried to reproduce it just now. I couldn't reproduce the slowdown until I pushed context much higher. I expected the model choice to matter more than it did.
Re: M3 Max for local use - sanescope - 09-20-2026
Quote:Tried to reproduce it just now. I couldn't reproduce the slowdown until I pushed context much higher. I expected the model choice to matter more than
Can you fill in a bit more context? Do you see the same thing with a smaller batch or shorter prompt?
Re: M3 Max for local use - deepseeked - 09-20-2026
Ran a quick check on my box. The issue went away once I changed one setting and restarted the stack cleanly.