speculative decoding actually worth it on 5090

[cacheline]

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

After a few runs on B200, I ended up with decent performance, but the setup only really made sense once I adjusted around speculative decoding.

Curious if anyone else landed in the same place.

I care more about stability than squeezing the last possible tokens/sec out of it.

[cacheline]

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

Good point. The pattern around batch size has been pretty similar for me. The biggest change for me was simplifying the stack instead of adding one more optimization layer. Turns out 'just one more test' was a lie.

[cacheline]

What worked for me was pretty simple. If you're on PyTorch, 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.

[thetatinker77]

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

What worked for me was pretty simple. On 7900 XTX, the easiest win is often to back off the aggressive settings and see whether the problem survives.

[cacheline]

I think the bottleneck is somewhere else. I wouldn't optimize around that first. On my side the bigger swing came from the runtime and not the model. Could be a workload difference, but I wouldn't generalize too fast from one setup.

[cacheline]

Can you fill in a bit more context? Are you testing this on 7900 XTX or something else?