ROCm version that finally behaves for Gemma

[mangofield33]

same here on my side.

[torchsmith25]

That matches what I was seeing. I ended up in about the same place using TensorRT-LLM.

[marbletrack]

I don't think that's the whole story. I get why that seems right, but on B200 the bottleneck felt elsewhere.

What worked for me was pretty simple. If you're on SGLang, I'd confirm the version first and then reduce variables until the behavior is obvious. Not glamorous advice, but repeatable baselines save a lot of time.

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