Inference › Inference Engines › Other Engines v
NVFP4 / INT4 tradeoffs

Thread Rating:
  • 0 Vote(s) - 0 Average
  • 1
  • 2
  • 3
  • 4
  • 5
Linear Mode
NVFP4 / INT4 tradeoffs
kernelmike Offline
Member
***
Posts: 184
Threads: 23
Joined: Sep 2026
Reputation: 0
#4
10-18-2026, 06:33 AM
My first guess would be this: Start by verifying the driver/runtime combo, then test with a smaller prompt, then look at reranking. Once you know the plain path works, then add the faster path back in piece by piece.
Mostly CUDA, sometimes wrong, usually measured
Find
Reply


Messages In This Thread
NVFP4 / INT4 tradeoffs - by deltagrid - 10-15-2026, 04:52 AM
Re: NVFP4 / INT4 tradeoffs - by oxidestream - 10-15-2026, 05:41 AM
Re: NVFP4 / INT4 tradeoffs - by vectorport - 10-18-2026, 06:26 AM
Re: NVFP4 / INT4 tradeoffs - by kernelmike - 10-18-2026, 06:33 AM
Re: NVFP4 / INT4 tradeoffs - by kernelmike - 10-18-2026, 11:30 PM
Re: NVFP4 / INT4 tradeoffs - by sanescope - 10-18-2026, 07:34 AM
Re: NVFP4 / INT4 tradeoffs - by rocmmixer - 10-18-2026, 08:28 AM
Re: NVFP4 / INT4 tradeoffs - by phaseport36 - 10-18-2026, 09:14 AM
Re: NVFP4 / INT4 tradeoffs - by amberplane - 10-18-2026, 09:02 AM
Re: NVFP4 / INT4 tradeoffs - by mellowwire98 - 10-18-2026, 09:54 AM
Re: NVFP4 / INT4 tradeoffs - by localpatch78 - 10-18-2026, 09:17 PM
Re: NVFP4 / INT4 tradeoffs - by sparrowcraft - 10-18-2026, 09:51 PM
Re: NVFP4 / INT4 tradeoffs - by sparrowhall - 10-19-2026, 03:33 AM
Re: NVFP4 / INT4 tradeoffs - by quantyard - 10-19-2026, 05:43 AM

Forum Jump:


Users browsing this thread: 1 Guest(s)