Inference › Hardware › GPUs v
4090 vs 5090 for 70B quants

Thread Rating:
  • 0 Vote(s) - 0 Average
  • 1
  • 2
  • 3
  • 4
  • 5
Linear Mode
4090 vs 5090 for 70B quants
turbogrid11 Offline
Junior Member
**
Posts: 48
Threads: 7
Joined: Sep 2026
Reputation: 0
#5
10-17-2026, 09:49 AM
Quote:That matches what I was seeing. I ended up in about the same place using vLLM. I got better results once I stopped chasing one metric and just looked

I tested this tonight. I couldn't reproduce the slowdown until I pushed context much higher.
retrieval before vibes
Find
Reply


Messages In This Thread
4090 vs 5090 for 70B quants - by batchwire - 10-16-2026, 02:20 AM
Re: 4090 vs 5090 for 70B quants - by cudapoint - 10-16-2026, 03:41 AM
Re: 4090 vs 5090 for 70B quants - by fastscope - 10-17-2026, 02:02 PM
Re: 4090 vs 5090 for 70B quants - by rocmriley - 10-17-2026, 03:53 AM
Re: 4090 vs 5090 for 70B quants - by turbogrid11 - 10-17-2026, 09:49 AM
Re: 4090 vs 5090 for 70B quants - by turbogrid11 - 10-17-2026, 10:40 PM
Re: 4090 vs 5090 for 70B quants - by southspark - 10-18-2026, 11:05 PM
Re: 4090 vs 5090 for 70B quants - by rocmriley - 10-18-2026, 11:45 PM
Re: 4090 vs 5090 for 70B quants - by batchshed - 10-19-2026, 01:15 AM
Re: 4090 vs 5090 for 70B quants - by batchwire - 10-19-2026, 05:43 AM

Forum Jump:


Users browsing this thread: 1 Guest(s)