• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

Can GPU cause system instability?

***F1ZZY***

***F1ZZY***

***F1ZZY***

Associate
Joined
5 Jan 2004
Posts
1,110
Location
Wiltshire
I recently got one of the 3070 from the forum deals and upgraded from 1080ti and I think the 3070 has been causing strange issues. Strange because they are not normally related to GPU. Mostly file related issues and missing files. Although sometimes games just crash to desktop. Always the same games, others are fine. Occasionally blue screen on boot up - winload.exe missing.

Normally you would expect this to be RAM or HDD or maybe CPU related. However, benches run ok. Prime, cinebench memtest all ok. It seems that once loaded, things works. It's the opening and loading stuff that causes the problem.

I have tried different SSD for OS and different RAM and the problem still persisted - admittedly it's quite intermittent so difficult to troubleshoot. I was playing games last night no problem, yet it bluescreened this morning and wouldn't even boot. I must have spent 30 hours in the past few days troubleshooting - latest BIOS, New OS on different drive, different RAM.

I have had the 1080ti back in a couple of times just to see but it doesn't seem right to me that it's caused by the GPU, so have swapped it back to 3070 once things appear to be stable. Only for it to fail again later. I'm currently on the 1080ti again and cant cause any issues at all. Even running prime and 3D mark together or cinebench and superposition together is ok. Will give the 1080ti a good innings this time.

I was convinced it was a RAM or SSD issue but the faults have continued with replaced parts. Maybe it is mobo/cpu/BIOS but I dont have alternative spare parts to test that. Anyhow memtest and prime seem ok.

Rest of system is:
Ryzen 2700
B450I Auros wifi miniITX
16GB GSkill 3200
Corsair SF750

Everything is at stock and PSU volts seem fine.

Can it be GPU related?
How to RMA it when it runs and benches fine? (most of the time)
 
was it a fresh install ? if not theres your problem.. reload windows .. if not ????? try a wall meter but cant see it doing over a 750w psu but it is a corsair ..
 
Tried checking your PSU? Some of the recent Nvida GPUs seem to have issues with power spikes that some PSUs can't handle. Maybe it's just drawing a lot of power and leaving your PSU borderline for a stable system when a load and power spike happens?
 
I have 2 fresh installs of win10 on 2 separate drives, one is m.2 and one is SATA SSD. I have switched between them many times. I was convinced it was my main m.2 for a long time but have disproved that.

The things that would break mostly would be files not opening up or installing (not much power draw at that point). Maybe the files were corrupted during a moment of instability???
Running benchmarks (even 2 at once) hasn't provoked anything to fail.

1080ti pulls about 300w so not far behind the 3070. I don't have a wall mater to check the draw.
If software monitoring is accurate, I haven't noticed any spikes in watts. but am aware the 30 series do spike.

It's SFF with not many components, total power draw less than 450W I'd say. My CPU pulls only 80W. Can the GPU spike to 500W or more?

Thanks for inputs thus far
 
I ran cinebench and superposition 8k together several times on the 1080ti. Swapped to 3070 and it black screened and disabled the keyboard and mouse lights during same tests.

3070 was sagging a bit so propped the far corner up and I’ve run those tests again ok.

can GPU sag cause these strange issues?

Edit. The 3070 is pulling 240w vs 300w 1080ti according to hwinfo64
 
Well, from what you have said and done so far I think it's obvious that the GPU is causing the issues. Why it is causing the issues is another matter. It could be a defective unit, some conflict with the mobo, a power issue etc as others have suggested. Is there any way you can get hands on another PSU just to try for a while? I hate this type of issue. You end up changing every component.

GPU sag is not good, obviously. There's a lot of strain on that PCI slot. You could support it with a suitable sized prop temporarily or a proper gpu support. At least that investment would not go to waste as all cards going forward are likely to be heavy due to HS/Cooling requirements.

EDIT: Oops didn't read your last post correctly and I now see that you have already battled the sag.
If it's working well now then it looks a pretty likely cause. You might also try running the PC on its side so there is no strain on the slot - just to confirm.
 
Last edited:
If the gpu is causing your PC to do weird stuff and. Rash at idle then it's most likely nothing to do with power draw.

a gpu that loses connection while the pc is on will cause a blue screen
 
The problem is that it is quite intermittent. It will run the harshest of benches fine and then not boot up windows or start a simple program.

I have been convinced it was HDD and then disproved it.
I was convinced it was RAM and then disproved it (after ordering and cancelling replacement from OCUK)

I hope the minion eraser holding up the GPU earns its keep. Fingers crossed.

I hate intermittent faults, they are the worst!!
 
Incidentally, may be slightly off topic but I made some DIY GPU supports using the following components:
Rubber Nuts
Threaded Bar
Decorative Nuts (Optional)
By cutting the bar into short lengths, putting a rubber nut at each end and extending if needed with a decorative nut in the middle, I made 2 rods, one for each end of the card but if there is nothing between the card and the computer case base you could add any number of these. The screw nuts ensure you can make the support as tight or slack as you want. Just a cheap and (I think) visibly pleasing solution.
 
They can cause system instability, any hardware misbehaving can do this, but typically it will result in a BSOD on modern versions of windows. The only way it can affect files on disk is if you crash during some kind of write operation and it's interrupted. I would definitely be graphing both load and temps for the CPU and GPU during games that crash, see if anything is overheating, probably also clockspeeds to see if anything is being throttled.

The absolute best test is to simply swap components out if you can source spares, friends who are willing to lend you a GPU for a few hours or even a CPU if you can be bothered to re-seat the cooling etc.

I assume you've done the usual, done all windows updates, installed the latest drivers for everything, primarily the GPU, made sure to virus scan the PC for malware/viruses, and made sure nothing dodgy is running in the background. And tried running these games with more or less everything else software disabled.

You say specific games seem to crash with more regularity, what are those games, is there a pattern with them (new/old/graphically intensive or not, etc)
 
Whilst PC running i gently moved the gpu pcie power cables and got black screen crash instantly. Rebooted, did same again.
Then PSU broke completely.
Replaced PSU with a spare 650w and it's up and running again.

OK, so the pcie power connectors on the 2 graphics cards are in different orientations. Retaining clip uppermost on 3070 and down on the 1080ti.
Theory is that the PSU and/or cables were dodgy and the manipulation required to plug each card in was aggravating things until it finally gave up the ghost.

Either that or the 3070 has some dodgy circuitry and its only a matter of time until this PSU also goes pop..!!!!

minion still holding in there........
 
When it crashes, does it tend to be a black-screen reboot?

I had these, and like you spent hours and hours disproving every other component in the system.
In my case it was a faulty RM750i PSU. I 'proved' this by swapping to an older HX750 I had, and the issue went away.

Try running the system with your spare PSU and 3070. See how you get on.
Be sure that the PCIe cables are firmly in place, not overly twisted or pulling on the card in anyway.

Good luck. These kinda issues are enough to drive one to tears.
 
Mine wasn't due to power consumption though - 3070 draws less than the 1080ti it replaced. PSU was an almost new 750w corsair platinum unit.

Just bad luck I guess.
 
***F1ZZY*** said:
I think it was psu as it has been ok since the swap thanks.

Corsair RMA is cr4p. 2 weeks nearly and no response from them.
Click to expand...
Really? I RMA'ed my RM750i back in March, and Corsair where brilliant. Usually responded within the day, and provided helpful advice for things to test before going straight to approving the RMA and sending me a shipping label.
They then upgraded me to the just refreshed RM850x. Which had to come from HK for some reason... so the longest bit was waiting the 5~ days for that to arrive.
 
xPETEZx said:
Really? I RMA'ed my RM750i back in March, and Corsair where brilliant. Usually responded within the day, and provided helpful advice for things to test before going straight to approving the RMA and sending me a shipping label.
They then upgraded me to the just refreshed RM850x. Which had to come from HK for some reason... so the longest bit was waiting the 5~ days for that to arrive.
Click to expand...

I have always found this as well. Corsair are brilliant. I guess that they must be having Covid problems at the moment or something.
 
You must log in or register to reply here.
Back
Top Bottom