is RDNA worth the hassle yet for FLUX

[quantghost]

Thought I'd post a small recipe because a lot of threads ask similar things.

Maybe useful as a starting point: Ollama on M3 Max, keeping reranking in mind from the start.

Feel free to tear it apart if you see something dumb.

My use case is pretty ordinary: local chat, some document work, and occasional batch jobs.

I expected the biggest difference to come from the model, but the stack around it mattered at least as much.

[cozykite]

I think the bottleneck is somewhere else. I get why that seems right, but on 4090 the bottleneck felt elsewhere. Could be a workload difference, but I wouldn't generalize too fast from one setup.

[batchyard]

Can you fill in a bit more context? Is this local-only or are you routing anything externally?

[mangofield33]

Can you fill in a bit more context? Do you see the same thing with a smaller batch or shorter prompt?

If you post your exact settings, people here can probably narrow it down pretty fast.

[rocmriley]

Can you fill in a bit more context? Do you see the same thing with a smaller batch or shorter prompt?

Good point. The pattern around prefill/decode split has been pretty similar for me. The biggest change for me was simplifying the stack instead of adding one more optimization layer.

[phasetrack]

A couple things I'd check first. Before rebuilding anything, I'd test one clean baseline with Mistral and compare from there. A lot of these threads end up being one bad config plus a pile of conclusions drawn from it.

[northhall88]

Need a little more detail here. Are you testing this on 5090 or something else?

[mangofield33]

Tried to reproduce it just now. I couldn't reproduce the slowdown until I pushed context much higher. The main thing that moved was latency, not quality.

[tensorworks]

Tried to reproduce it just now. I couldn't reproduce the slowdown until I pushed context much higher. The main thing that moved was latency, not quali

Good point. 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.

[torchsmith25]

that fixed it for me too.