is RDNA worth the hassle yet with SGLang

[thetaember38]

Did a bit of testing and thought I'd post the result here.

After a few runs on M3 Max, I ended up with decent performance, but the setup only really made sense once I adjusted around continuous batching.

Happy to post more details if anyone wants them.

[quantshed]

Did a small comparison this afternoon. I couldn't reproduce the slowdown until I pushed context much higher. I keep promising myself this is the last rebuild.

[cozykite]

A couple questions before I guess too hard. Do you see the same thing with a smaller batch or shorter prompt?

[rocmmixer]

I'd approach it in a boring order. 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.

[deltapoint17]

What worked for me was pretty simple. On A6000, the easiest win is often to back off the aggressive settings and see whether the problem survives.

[northhall88]

A couple things I'd check first. If you're on vLLM, I'd confirm the version first and then reduce variables until the behavior is obvious.

[papermixer4]

What worked for me was pretty simple. If you're on vLLM, I'd confirm the version first and then reduce variables until the behavior is obvious.

[rocmmixer]

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

[tensorworks]

That matches what I was seeing. I ended up in about the same place using SGLang. Worth rechecking your defaults though, because one bad assumption can make the rest of the thread look confusing.

[mintquill]

Ran a quick check on my box. On M3 Max with vLLM, I saw basically the same behavior. The surprise was that memory pressure showed up before throughput collapsed.