Jump to content

Interesting behavior - FPS vs GPU Usage

Interesting little behavior I noticed while going through my backlog of games.

 

I fired up Rise of the Tomb Raider and limited the max fps to 80 in Nvidia Control Panel, and the GPU usage was 89% during game play. GPU temperatures hovered around 50C. I decided to limit the fps even further by reducing the cap to 60 fps. I noticed the temperatures dropped into the low 40s, but the GPU usage remained around 90 to 91%.

 

It's not really a problem. I just can't really explain it. I would expect usage to decrease with fps to some degree. No??

 

 

GPU: ASUS 3070 Ti

Monitoring UI: GPU Tweak 3

Link to comment
https://linustechtips.com/topic/1455475-interesting-behavior-fps-vs-gpu-usage/
Share on other sites

Link to post
Share on other sites

When really examining GPU usage like you're doing, you have to pay attention to clockspeed just as carefully. That usage graph, whether it be in Task Manager or Afterburner, its going to display relative usage based on clockspeed. 

 

For instance, if I'm idle at my desktop and I open a Youtube video with my 3090, I might see 20-25% "usage". Is it taking 25% of my 3090's power to play a Youtube video? No, absolutely not. But if you look at the clockspeed, its still at at the 210Mhz idle clock so all of a sudden that 25% usage makes sense. 

 

I might have gone a whole ramble for nothing if you've been also graphing core clockspeed and it hasn't changed but i'm willing to bet it did.

Ryzen 7 7800x3D -  Asus RTX4090 TUF OC- Asrock X670E Taichi - 32GB DDR5-6000CL30 - SuperFlower 1000W - Fractal Torrent - Assassin IV - 42" LG C2 - Windows 11 Pro

Link to post
Share on other sites

12 minutes ago, GuiltySpark_ said:

I might have gone a whole ramble for nothing

I made this thread just to ramble 😀

 

That's even more interesting to me that you're saying usage is based on clock speed. I searched the web for about a half hour and couldn't find anything definitive on how cpu or gpu usage is calculated. And I suppose that's rightfully so. There's several ways to consider the "usage" of a device. But there has to be some reporting standard, right?? Even if it's unofficial.

 

I've grown very nerdy so I might just do that... Use HWinfo64 to log GPU data for plotting. I might try a few different games just to see if the behavior is consistent.

Link to post
Share on other sites

Just now, johnt said:

 

 

I've grown very nerdy so I might just do that... Use HWinfo64 to log GPU data for plotting. I might try a few different games just to see if the behavior is consistent.

I've got a whole display dedicated to HWiNFO's graphs! (if you double click any sensor it opens a graph) Got a column for CPU stats and another for GPU stats and I watch this stuff a lot lol

Ryzen 7 7800x3D -  Asus RTX4090 TUF OC- Asrock X670E Taichi - 32GB DDR5-6000CL30 - SuperFlower 1000W - Fractal Torrent - Assassin IV - 42" LG C2 - Windows 11 Pro

Link to post
Share on other sites

Results are in! GPU usage is still a brain scratcher.

 

HWinfo defines core load (or usage) as a number reported by the hardware, so it's no wonder there isn't as much info on the usage. I looked at 60 fps, 80 fps, and max fps just to see the behavior differences. GPU clock, power consumption, and temperatures all behave as expected... they increase as the fps increases. The core load reports 100% for max fps, but based on the other three charts, I don't think it should be reporting 90%+ for 60 or 80 fps. For reference, max fps ranges from 108 to 145 fps.

 

CPU temps and average clocks were also higher as fps increased, which is expected.

 

 

 

gpu core load.png

gpu clock.png

gpu power.png

gpu temp.png

Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

×