Thread Rating:
  • 0 Vote(s) - 0 Average
  • 1
  • 2
  • 3
  • 4
  • 5
NVFP4 / INT4 tradeoffs
#10
Ran a quick check on my box. I couldn't reproduce the slowdown until I pushed context much higher. The surprise was that memory pressure showed up before throughput collapsed.
Reply


Messages In This Thread
NVFP4 / INT4 tradeoffs - by oxidegrove40 - 10-18-2026, 04:15 PM
Re: NVFP4 / INT4 tradeoffs - by mintforge - 10-18-2026, 05:10 PM
Re: NVFP4 / INT4 tradeoffs - by batchyard - 10-19-2026, 02:35 AM
Re: NVFP4 / INT4 tradeoffs - by thetapath - 10-18-2026, 06:16 PM
Re: NVFP4 / INT4 tradeoffs - by thetagrid15 - 10-18-2026, 09:13 PM
Re: NVFP4 / INT4 tradeoffs - by modbench - 10-18-2026, 06:49 PM
Re: NVFP4 / INT4 tradeoffs - by llmtrack - 10-18-2026, 07:04 PM
Re: NVFP4 / INT4 tradeoffs - by kernelghost - 10-18-2026, 09:23 PM
Re: NVFP4 / INT4 tradeoffs - by llmarch83 - 10-18-2026, 11:08 PM
Re: NVFP4 / INT4 tradeoffs - by batchshed - 10-19-2026, 12:08 AM

Forum Jump:


Users browsing this thread: 1 Guest(s)