PC stuttering issue

[Giu176]

Computer Type: Custom Build.

GPU: MSI RTX 3070 Ventus 2x OC, no manual OC, I installed MSI dragon center and set Extreme Performance scenario, in normal conditions the GPU runs at 1755MHz and boosts up to 2010MHz.

CPU: AMD Ryzen 5 5600x, (OLD OC settings: Before encountering this issue I followed a guide and find out a stable configuration, I can report my BIOS settings because I'm not an expert, PBO: enabled (advanced), PBO limits: motherboard,Precision Boost Overdrive Scalar: 4X, Max CPU Boost Clock Override: 200MHz, Curve Optimizer: All Cores negative 5. With this configuration I used the PC for several months without any problems, after this problem started I tried to restore the defaults values but nothing changed). CURRENT OC settings: only D.O.C.P. Enabled.

Motherboard: ASUS ROG STRIX X570-F GAMING; BIOS version 4021 (08/09/2021).

RAM: Crucial Ballistix BL2K8G36C16U4W 3600 MHz, DDR4, 32GB (8GB x4), CL16, D.O.C.P. Enabled.

PSU: Seasonic Focus GX-750, 750W, 80 plus gold.

Operating System & Version: Windows 10 Pro, Version 21H2, 64 bit, OS build 19044.1526, clean install.

GPU Drivers: GeFrorce Game Ready Driver Version 511.79, upgrade using Geforce Experience (latest avilable drivers).

I bought all new parts and built the PC exactly one year ago.

 

Description of the problem: During the last 6/7 months I noticed a strange behavior from my new PC, sometimes, the whole system stutters (video and audio) for like a second and then comes back to normal. I noticed this behavior independently of the load (happened in gaming and while browsing, it seems to be random) it happens like one/two times a day.

I check constantly the temperatures, power loads and clock of my devices using AIDA64 extreme sensor panel on an external monitor inside the case so I can troubleshoot thing like this easier. I can say that the temperatures are good, the GPU reach max 72°C in games (38/40 in idle) but as I said before the load doesn't matter, I stutters randomly and I can't reproduce the issue in a controlled manner.

 

Troubleshooting I did:

Introduction:

I run Malwarebytes Premimum scan daily and the system is clean.

I updated all the drivers and the BIOS but nothing changed.

I always keep Windows 10 up to date and the issue is present in every version.

Initially I thought that this problem could be caused by the 3070 because the times it happened while gaming I've seen the GPU usage and power usage going to zero while the CPU power load was unchanged, I hope I was wrong, after all it still can be due to something else and the GPU usage drops because of the stutter and not the other way around. Usually the problem is harmless, everything behave well even after the stuttering.

Complete formatting:

I performed a clean Windows installation erasing everything but the issue remained.

Stuttering makes games crash:

The worst event happened three months ago when I was playing Overwatch and it stuttered very hard (like for two/three seconds), the system didn't come back to normal and the game crashed, HDMI monitor went dark while display ports output remained active. After this event the GPU clock was "misbehaving", it came down to 210 MHz with casual boosts to 350MHz, I went to Nvidia Control Panel to check the Power Management Mode and it was set to Maximum Performance as usual. I ran an AIDA64 extreme stress test and the GPU couldn't reach anything above 400MHz, GPU usage remained lower than 4%. I rebooted the system and everything was back to normal (1755MHz clock in idle, few minutes of stress test in AIDA64 ran perfectly). I think the GPU entered a self-protection state. Windows Event View screenshots of the events reported after Overwatch crash: https://imgur.com/a/PF7daUR

Never experienced anything like this even if the stutter still persists in games (as already said without any noticeable increase in frequency as compared to other applications).

Some minor troubleshoots:

Tried to bypass the ups and powering the PC directly from the wall but the issue remained.

Tried using MSI Dragoncenter to manage GPU performance, it didn't work.

Removed the 500GB secondary NVMe drive, initially I installed it on the PC for extra storage but I removed it because it was causing problems when was my main drive on my laptop, but that's another story.

Adding more RAM:

Doubled my ram storage passing from 16 to 32 GB, performed a memtest86 checkup and the memory passed all the tests.

Changing the motherboard:

When I opened the case for upgrading the RAM I inspected my hardware: the motherboard turned out to be damaged. My cooler backplate was interfering with some components. The cooler (coolermaster hyper evo 212 black edition) is rated as compatible with the motherboard (Rog Strix X570-f) but it uses a custom backplate that is too big and touches some capacitor (or resistance idk) on the back of the motherboard. Photos of the damaged board: https://imgur.com/a/gAVMjrd . I replaced the motherboard with a newer one (same model) and installed a new cooler (arctic freezer 34) that is mounted directly on the original rog backplate. I reinstalled the same bios version I had before, reactivate windows, kept all my data. The issue still persist.

Things I found on the LTT Forum:

Tried following this guide on the forum: https://linustechtips.com/topic/1353904-amd-ftpm-causing-random-stuttering/ about fTPM causing the issue, I tried disabling it but nothing changed.

The french guy in the second video posted in this topic is describing the same exact issue: https://linustechtips.com/topic/1177546-windows-10-micro-stutters-with-ryzen/ , I tried uninstalling those devices drivers as shown in the video and the problem still persists, I was very happy after finding this video because I had the same exact problem but the solution didn't work. The other video shows how to use a tool for switching the GPU mode to "msi" (I don't know what it is but mine appeared to be already in that mode, the check-box in the software was already marked, I don't know if I have to do something different with that program).

 

I need your help to troubleshoot more. I have a friend that has a Ryzen 7 5800x and we can try to swap our CPUs or in alternative I can search for another Ryzen 5. But after watching the video about the drivers I really start to think that it's fixable in some way I can't figure out.

 

Thanks for reading.

[gymleader]

Just to make sure. How did you go about disabling fTPM? Lots of people confuse this thinking they turned it off when they didn't.

[Giu176]

3 minutes ago, gymleader said:

Just to make sure. How did you go about disabling fTPM? Lots of people confuse this thinking they turned it off when they didn't.

Disabled it in the BIOS, if I remember correctly I searched it using the search bar and changed it from enabled to disabled.

[SquintyG33Rs]

I would guess that the GPU entered the 2D lock mode. Usually it shouldn't but it can be turned on by multiple things:
 - GPU overheating (or GPU VRM or GPU vRAM too) unlikely in this case
 - mining software typically forces lock state
 - Driver bug
 - detecting an error in power measurement (for example what happens with liquid metal mods)
 - GPU bios bug (you can try and fix this with the bios flash tool)

 

It's hard to say what could fix it because I can't try each fix for you.

[gymleader]

To make sure go to Advanced > Trusted Computing > Security Device Support.

 

Make sure it says Disabled there.

[Giu176]

58 minutes ago, SquintyG33Rs said:

I would guess that the GPU entered the 2D lock mode. Usually it shouldn't but it can be turned on by multiple things:
 - GPU overheating (or GPU VRM or GPU vRAM too) unlikely in this case
 - mining software typically forces lock state
 - Driver bug
 - detecting an error in power measurement (for example what happens with liquid metal mods)
 - GPU bios bug (you can try and fix this with the bios flash tool)

 

It's hard to say what could fix it because I can't try each fix for you.

I will exclude the first two on the list. Can you suggest something I can try for the other points? Driver bug is still the main suspect in my opinion

[SquintyG33Rs]

35 minutes ago, Giu176 said:

I will exclude the first two on the list. Can you suggest something I can try for the other points? Driver bug is still the main suspect in my opinion

actually it could be Dragon centre itself causing the issue. because the 2D lock does happen when a bad OC is applied. and when you click those quiet or performance buttons it applies an OC profile and it's not always all that smart about it. I don't really like dragon center so i don't have much experience using it but it would be much less annoying if it just modified the powerlimit slider.
 

[Giu176]

10 hours ago, SquintyG33Rs said:

actually it could be Dragon centre itself causing the issue.

I forgot to write that I've already uninstalled after testing it.

[SquintyG33Rs]

6 hours ago, Giu176 said:

I forgot to write that I've already uninstalled after testing it.

for driver you use DDU to wipe it clean and install it back again

the other two come down to BIOS flash (if it's not a defect physically) you have to find your card's bios in archives (just google your card + bios) then you get the nvidia bios flash tool and you try and flash it. if you don't add any force flags it won't let you flash a wrong bios on your card but it's just like your motherboard if anything stops the process midway you will be left with an expensive paperweight.

 

it could also be something else being dumb and just a windows restart will clear it (btw clicking restart is not the same as shutdown and then power-on button).

you should also take into account that the daily stutter is probably not linked to the 2D lock in overwatch at all.
that's the kind of behaviour you see from antivirus software doing it's scan in the background (or other autoscanning shit). unless you're pirating games allot and you don't trust yourself in choosing good sources for the cracks there's no way you will ever need any antivirus on top of windows defender.