best local runtime for mixed models

[vectorember]

Posting this in case it saves someone else a little time.

If you were starting from scratch today for this use case, what would you choose first and what would you ignore?

Simple and boring is a valid answer.

[oxidemixer]

A couple questions before I guess too hard. What model, what runtime, and roughly what context size?

[edgeriver]

What exact setup are you running? Is this local-only or are you routing anything externally?

[edgeriver]

A couple questions before I guess too hard. What model, what runtime, and roughly what context size?

[vectorhouse71]

What worked for me was pretty simple. On GB200, the easiest win is often to back off the aggressive settings and see whether the problem survives. Once you know the plain path works, then add the faster path back in piece by piece.

[edgeriver]

What exact setup are you running? What model, what runtime, and roughly what context size?

[ggufwire22]

I'd approach it in a boring order. On GB200, 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.

[pixelspark]

Tried to reproduce it just now. On Threadripper box with MLX, I saw basically the same behavior. The surprise was that memory pressure showed up before throughput collapsed.

[edgeriver]

My first guess would be this: If you're on LM Studio, I'd confirm the version first and then reduce variables until the behavior is obvious. A lot of these threads end up being one bad config plus a pile of conclusions drawn from it.

[cacheworks]

My first guess would be this: If you're on LM Studio, I'd confirm the version first and then reduce variables until the behavior is obvious. A lot of

Not sure I agree there. The reported numbers can look compelling, but in ordinary usage I didn't see the same benefit. Worth testing the simpler explanation before adopting the more interesting one.