Inference › Optimization & Performance › Context / KV Cache v
speculative decoding actually worth it on 5090

Thread Rating:
  • 0 Vote(s) - 0 Average
  • 1
  • 2
  • 3
  • 4
  • 5
Linear Mode
speculative decoding actually worth it on 5090
cacheline Offline
Senior Member
****
Posts: 341
Threads: 51
Joined: Sep 2026
Reputation: 0
#3
09-25-2026, 11:06 AM
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.
measure twice, benchmark three times
Find
Reply


Messages In This Thread
Re: speculative decoding actually worth it on 5090 - by cacheline - 09-25-2026, 11:06 AM

Forum Jump:


Users browsing this thread: 1 Guest(s)