prefill bottlenecks on longer prompts on A6000

[labline]

subscribed, curious where this lands.

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

[vrambandit]

Thanks, that helps. That matches what I saw on B200. I got better results once I stopped chasing one metric and just looked at whether the workflow felt usable.

[deltastation]

Did a small comparison this afternoon. I couldn't reproduce the slowdown until I pushed context much higher.

[rocmstack25]

subscribed, curious where this lands.

[homelabhenry]

This keeps coming up in conversation so I figured I'd start a thread.

One thing I keep noticing is that people underestimate speculative decoding until they actually try to run this stuff for more than one afternoon.

Wondering if this is just where the whole space is heading now.

For context, I'm not chasing leaderboard numbers. I mostly want the system to feel consistent day to day.

[quiettoken]

My first guess would be this: Before rebuilding anything, I'd test one clean baseline with Mixtral and compare from there. Once you know the plain path works, then add the faster path back in piece by piece.