best small quant that still feels smart on 7900 XTX

[benchplease]

I figured I'd ask here before changing everything.

I'm mostly working with DeepSeek in SGLang on Threadripper box. The part I'm trying to understand is prefill/decode split.

Would appreciate practical advice rather than theory.

[stackarch]

What exact setup are you running? Are you testing this on GB200 or something else?

[fusedtinker]

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.