• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.
  • The forums have been upgraded with support for dark mode. By default it will follow the setting on your system/browser. You may override it by scrolling to the end of the page and clicking the gears icon.

Strange GPU Load readings

lolmanfin

New Member
Joined
Mar 4, 2019
Messages
9 (0.00/day)
I use GPU-Z mainly to log data from my GPU to analyze it. I think there's something wrong with the readings I'm getting now.
This the graph from GPU-Z provided data when I play Rocket League:
1551726440919.png

Meanwhile I monitored the reading from Task Manager at the same time. There the readings were 50%-ish all the time. Other readings matched with HWMonitor, like temperatures and fan speeds.
Is GPU-Z's GPU load different than Windows Task Manager's GPU's utilization and HWMonitor's GPU Utilization or is there a problem, and how to solve it?
I have Gigabyte GV-R9290OC-4GD which is factory overclocked R9 290. I tried to update drivers and checked that I have the newest version of GPU-Z.
My OS is Windows 10.

By the way, I know my GPU is getting way too hot, already working on that...
 
Fix the heat problem first. Besides that, are you using vsync or any frame limiter?
 
Fix the heat problem first. Besides that, are you using vsync or any frame limiter?
I don't think there is any correlation between heat and GPU-Z functionality. Should work even despite the slight heat problems. The R9 290 is a hot card anyways.
I always use vsync (60Hz) or other limiter on games.
 
I don't think there is any correlation between heat and GPU-Z functionality. Should work even despite the slight heat problems. The R9 290 is a hot card anyways.
I always use vsync (60Hz) or other limiter on games.

The GA cards in general are hot, cheaply made.
 
The GA cards in general are hot, cheaply made.
I had way lower temperatures before my GPU fans came to their end and I replaced them with similar ones. I also reapplied thermal paste (for the first time) and I think I didn't do that well enough.
 
I had way lower temperatures before my GPU fans came to their end and I replaced them with similar ones. I also reapplied thermal paste (for the first time) and I think I didn't do that well enough.
Do it again, right :p though your temps are under 80c so it's not limiting are you playing with vsync on? if so your GPU usage will vary just to keep you at 60fps or W/E your refresh rate is
 
Do it again, right :p though your temps are under 80c so it's not limiting are you playing with vsync on? if so your GPU usage will vary just to keep you at 60fps or W/E your refresh rate is
Planning to reapply the thermal paste.
Why do both Task Manager and HWMonitor give a steady 50% GPU load, which seems to be more reliable than varying between 0-100% a lot?
 
Planning to reapply the thermal paste.
Why do both Task Manager and HWMonitor give a steady 50% GPU load, which seems to be more reliable than varying between 0-100% a lot?
On average it's probably around the 50% mark which is what taskmgr and hwinfo are reporting though in game it will vary up and down depending on how much GPU power it needs at any given time to keep your 60fps so GPU-Z is giving a more real time usage read out.
 
On average it's probably around the 50% mark which is what taskmgr and hwinfo are reporting though in game it will vary up and down depending on how much GPU power it needs at any given time to keep your 60fps so GPU-Z is giving a more real time usage read out.
1551731027042.png

Time-data is done by me (+10sec, as set in GPU-Z also) just so I get to make the graph easier.
As you can see, GPU-Z is mainly giving 0 or 100 values.
Here are my GPU-Z sensor settings
1551731217919.png

It simply can't be right. There's a 2 minutes long 0% load period during gaming. I've tried with 0.1 sec refresh rate also, results are similiar.
 
That period looks like you have exited the game as your memory usage goes right down also, perhaps @W1zzard can clarify
 
That period looks like you have exited the game as your memory usage goes right down also, perhaps @W1zzard can clarify
I wouldn't trust the memory controller load either. Memory Used (Dedicated), which is third column from right, shows that there is over 1GB VRAM being used during that session (from 2 to 4 minutes) and I was in-game the whole time.
 
To clarify what I'm after here, I would like GPU-Z to log the % of maximum potential performance used, like Windows Task Manager does it. But the "GPU Load" readings aren't indicating that at all. Are they supposed to, is it broken, or is there not any way to do it as I want?
 
There is a similar issue with polaris cards, in hwinfo you will have "Utilisation" and "D3D usage" Task manager reports the latter, which works afaik. The former goes up down up down up down up down on my 570 like a yoyo. not had the issue on vega, but i cant remember if i did on any hawaii cards i owned.
 
1551962498114.png

I got the GPU utilization, which seems to be the D3D Usage that I wanted, from HWiNFO. I'm not sure if my temps are as good as they should be, around 90 degrees Celsius on around 70% utilization. (Battlefield 4 ultra graphics 1080p with vsync 60Hz on)
 
It has been mentioned gigabytes cooling solutions are not the best and are pretty poor compared to other 3rd party cards, you can ventilate the case more, add more fans etc though it is what it is for the most part unfortunately.
 
It has been mentioned gigabytes cooling solutions are not the best and are pretty poor compared to other 3rd party cards, you can ventilate the case more, add more fans etc though it is what it is for the most part unfortunately.
As far as I remember the only good R9 290 was Sapphire.

View attachment 118169
I got the GPU utilization, which seems to be the D3D Usage that I wanted, from HWiNFO. I'm not sure if my temps are as good as they should be, around 90 degrees Celsius on around 70% utilization. (Battlefield 4 ultra graphics 1080p with vsync 60Hz on)
Oh man... thought firstly you said around 70 Celsius. You must somehow cool it down to 80 or less. What temp is it on BF4 when you set fans to max rotation speed?
 
As far as I remember the only good R9 290 was Sapphire.


Oh man... thought firstly you said around 70 Celsius. You must somehow cool it down to 80 or less. What temp is it on BF4 when you set fans to max rotation speed?
I said around 70-80 Celsius when playing Rocket League and GPU utilization is around 50%.
The fans are going max rotation speed (around 4500 RPM) and the temperatures hit maximum 91 Celsius when playing BF4 with vsync.
The GPU has Windforce 3x 75mm fans.
I have two 80mm fans on the front blowing in and one 80mm fan blowing out in the case. Do you think a fan or two blowing out on top would help the temperatures?
 
My opinion is, that it can not help that much, to cool down 10 degrees.
But it will help a little. Also try to change paste between graphics chip and heatspreader, if you haven´t done yet. Together with some air pressure cleaning of radiator.

Make sure all case fans are really spinning. And if GPU fans spinning.

Well if that won´t help then there are three options left. Let it be. Expensive Accelerro extreme. Or even more expensive new graphics - honestly this will probably solve problem with 0% GPU load X-D

edit:
https://i2.wp.com/www.funkykit.com/wp-content/uploads/2014/04/int_01-1.jpg?w=600&ssl=1 - doesn´t look that bad. Maybe if there was direct contact with pipes but... yea, r9 290 was tough MoFo for any manufacturer
 
Last edited:
You need to reapply thermal paste as you said you've done this but not sure if you done it correctly, do it correctly, watch a video if necessary or we could talk you through it as you go, that's the only thing I could suggest unless the fans you've replaced are really that bad. Also make sure you reapply the heatsink properly with enough pressure that it's making the correct contact with the gpu.
 
Back
Top