Hello all,
I am folding with a RTX 4090 on Linux (also true on Win10) and I see a GPU usage close to 100% when folding but a memory usage of the card of only 3-4%.
I thought that the advantage of the RTX 4090 was to have more memory on the card than others.
Do you witness equivalent usage? Is there a way to increase GPU RAM usage to fold more or the limitation is on the GPU itself?
thanks
RTX 4090 High GPU usage low memory usage
Moderator: Site Moderators
Forum rules
Please read the forum rules before posting.
Please read the forum rules before posting.
-
- Posts: 100
- Joined: Wed Jan 05, 2022 1:06 am
- Hardware configuration: 4080 / 12700F, 3090Ti/12900KS, 3090/12900K, 3090/10940X, 3080Ti/12700K, 3080Ti/9900X, 3080Ti/9900X
Re: RTX 4090 High GPU usage low memory usage
I am not seeing this on my 4090 on Win10 and Intel 12900KS.
-
- Site Admin
- Posts: 7926
- Joined: Tue Apr 21, 2009 4:41 pm
- Hardware configuration: Mac Pro 2.8 quad 12 GB smp4
MacBook Pro 2.9 i7 8 GB smp2 - Location: W. MA
Re: RTX 4090 High GPU usage low memory usage
No, the advantage for the RTX 4090 is the large number of shaders on the GPU, 16384 as compared to 9728 on a RTX 4080 or 7680 on the RTX 4070 TI. Th card could have half the video RAM and would still process WUs as fast as long as they did not change other aspects of the card. Nvidia has released different RAM amounts on cards without changing other specs in the past and no detectable difference in processing speed was seen.
This does not apply where other specs were also changed. Examples that I can recall are the GTX 1060 which was released in 3 and 6 GB configurations and the RTX 2060 in 6 and 12 GB. The GTX 1060 3 GB card had fewer shaders, 1152 vs 1280. The 12 GB version of the RTX 2060 had more shaders and a higher base core clock compared to the 6 GB card.
iMac 2.8 i7 12 GB smp8, Mac Pro 2.8 quad 12 GB smp6
MacBook Pro 2.9 i7 8 GB smp3
-
- Site Moderator
- Posts: 6349
- Joined: Sun Dec 02, 2007 10:38 am
- Location: Bordeaux, France
- Contact:
Re: RTX 4090 High GPU usage low memory usage
FAH barely uses the VRAM on the GPU board, this is how OpenMM has been designed.