[SOLVED] Mini freezes in game with drops in GPU Core Load ?

[Lokart]

Hi,

I'm experiencing issues with my computer (specs below) in the last few months. When playing games, I get some short, max 1-sec long, freezes. They happen every minute or so. As you can imagine, this is really annoying.

• They happened in several different games, like WoW, LoL, TESO
• The sound does not seem affected and carries on normally
• I've tried to reduce graphic quality in game, but it changes nothing

I've been investigating with the help of friends. I have downloaded Hw monitor and OCCT to track CPU GPU usage, temperatures etc. I find OCCT to be quite useful with its graphic option.
We thought it might be due to the graphic card having a problem, so I changed from GTX 970 to GTX 1660, but the problem persists!
Here are some things I noticed.

When I freeze, here is what I noticed happens (with GTX 970) :
- Drop in GPU core Load
- Increase in CPU usage
- Drop in GPU power, but nothing changes for CPU power
- Slight decrease in GPU temperature - which is not overly high on normal times (65-70°C)

Below with GTX 970

Here with GTX 970 - The spike on the left is the freezing moment

Power supply screenshot taken slightly after the image above

After changing to GTX 1660, here what's happening
- Drop in GPU usage
- Increase in CPU usage
- Nothing noticeable for GPU power (drops dissapeared). Slight increase in CPU power usage
- Slight increase in GPU and CPU temperature (one spike +5-10°). Was decrease with GTX 970

Below with GTX1660

What I already tried to do (without sucess):

• Countless times uninstall and install drivers, even using DDU uninstaller
• Uninstall and install windows 10
• Uninstall and install WoW
• Change the power cable that powers the GPU to another unused cable from the PSU
• Play on a single monitor that I have used for a long time (I recently upgraded to 27'' monitor and I thought it could be the problem)
• Run OCCT 3D stress test for 5 minutes : No problems detected
• Run OCCT Power supply stress test : Computer shuts down shortly after starting the test (like 1-3 seconds) (like black screen, as if you'd remove power supply)
• Change graphic card to GTX1660

I'm desperate Any help would be greatly appreciated. Also, I am not very familiar with computer so please don't use too much acronyms that I wouldn't understand.
My only guess would be that my PSU is failing because of the stress test problem.

Thanks.

My specs and date of acquisition are the following :
CPU: Intel(R) Core(TM) i7-4790K CPU @4.00GHz 4.00 GHz (2014)
Motherboard: ASUS ROG Maximus VII Ranger (2014)
CPU Cooler: Corsair Cooling hydro series H60 (2016)
GPU: Nvidia GeForce GTX 970 (2014)
Newly bought GPU : Nvidia GeForce GTX 1660 Super (2020)
RAM: Corsair Vengeance DDR3 16Gb (2x8 GB) CCR1600 (PC3-12800) (2014)
SSD1: Samsung 840 Evo Basic SATA 6 Gb/s 2.5'' - 500 GB (2014)
SSD2 : SSD 480 Gb SATA Kingston HyperX Savage SSD (2016)
PSU: Corsair CX Series 750W (2014)
Windows 10

 

[tennis2]

Oversimplified description of CPU & GPU roles in games:

• CPU determines what should be shown in each frame and sends the "draw call" to the GPU
  • What should be in the scene, where it should be, etc
• The GPU renders the scene in the frame (generates a picture essentially) and sends it down the pipeline to the monitor
• As you'd imagine, the GPU doesn't know what to draw until the draw call is received from the CPU. If the CPU isn't able to complete the draw call for the next successive frame before the GPU finishes the previous one, the GPU must "wait". This would result in <100% GPU usage for that duration. This generally happens when many things are changing from one frame to the next, and/or when the graphical complexity is comparatively low (think high FPS and low video quality settings).
• If the graphical complexity is high, the GPU will not be able to finish rendering each frame before the CPU sends it the next frame's draw call. In that case, the GPU is pegged at 100% and the CPU usage may dip.
• Becuase of high core count CPUs, it's not always easy to monitor CPU usage %. If a game can only utilize 4 threads on a 4c/8t CPU, then you're probably not going to see 100% CPU usage, but you could still be having <100% GPU usage which would indicate a CPU IPC bottleneck.

I presume that, since you mentioned reinstalling Win10, you've made efforts to limit/eliminate background processes? Do you have any Asus software for your mobo installed? (for example)

 

[Lokart]

Thanks for replying and explaining how it works. So it might be the CPU failing to give info to the GPU, which stops for a while ?

I have not made any particular effort to reduce background processes.
I don't know if I have any particular Asus software on my mobo, thought I just cleared the CMOS as suggested by a friend, and it did not help with the issue.

 

[tennis2]

A couple things to try

• Disable/Uninstall any/all Asus software.
• Check your Task Manager (Ctl Alt Delete) Startup tab and disable any/all software which isn't absolutely necessary for your usage/testing. Really just want to eliminate all the variables which could chew up CPU clocks other than your game for now. Then you can enable them one at a time, testing each time to see if/what software is causing these issues.
• Check for (and perform) a BIOS update.

 

[Lokart]

Hi,
Thanks again,
I have tried an OCCT test with the CPU. There are two tests avaiable, "OCCT" and "Linpack".
Both test ran for 1 minute or so and then black screen (as if I unplugged the computer)
I took some screenshot during the Linpack test. Temperature rose to ~80°C spiking up and down. Then, shortly before turning off, it started to spike heavily between 65° and 93°C as I've tried to show on this image :

Another thing that I have not mentioned : My CPU cooler system broke down 2 years after installing it (2016) and I had high temperature issue on my CPU which made my computer crash when playing some games. So it might have damaged the CPU at that time, though I have not noticed any issue since changing the cooling system 4 years ago.

Background processes : Google chrome uses the most processor and memory so it should be fine
Asus software. I haven't found any on my PC.
BIOS : I have a version from 2014, might be time to upgrade it I will try this
Edit: Updated, but it does not work :/

 

[Lokart]

A little update :

I figured I had to change CPU so I bought a new CPU, motherboard and RAM. To my amasement the problem is still present.


My specs are now :
CPU: AMD Radeon 5600X - brand new
MB: MSI X570 Gaming pro carbon wifi - brand new
RAM: G-Skill Ripjaw V (2x, 8GB, DDR4-3200, DIMM 288) - brand new
GPU: GeForce 1660X Super - brand new
PSU : Seasonic focus GX 750 W - brand new

The only hardware that remains from the previous pc is the water cooler - which I guess can't be involved in freeze - and my SSD hard drives.
When playing different games installed in both my drives there is the problem.

I'm really lost here...

Thanks for your help