Help needed with PC issue / new GTX 970

[Daniel Ward]

Hi all

Apologies if this is in the wrong place, but I could not find an official support thread.

I started having issues with windows 10, it would black screen and then crash to desktop during random tasks, and give me a Nvidia driver has stopped responding and has recovered error, apparently a fairly common error with windows 10.

I wanted to upgrade my card in any case, and got a second hand msi 970 gaming edition. Card was working in the seller's machine, tested before I bought.

Worked for about 24 hours and then started crashing, with the same errors. This is when I started toubleshooting. To name a few, I have tried the following, and still have the same problem:

- Reverted to Windows 7
- Changed power plan of PC to high performance
- Changed power plan on nvidia to high performance
- Increased power supply to card to max on MSI afterburner
- Removed drivers with DDU ( safe mode ) and installed previous versions
- Tried a different PSU

The most success I had was with driver version 347.88. With this version I was able to have a stable PC, and run The Division game.

I then ran into another problem - My FPS would drop from 60 down to 25 during graphic intensive parts, and during these parts, my CPU usage would be at 100% and 82 degrees celcius, while my GPU usage then dropped to 40%. GPU never went above 60 degrees. Perhaps the CPU was causing a problem with the GPU, not sure.

With this version the nvidia control panel would not work, nor geforce experience, perhaps because the driver is too old.

I then tried to run Heaven benchmark 4.0 to test the frame rates under load and the programme would not launch due to it not detecting that I was in 1920x1080 res. After that I went back to the latest version of the nvidia drivers and maxed the max power supply to the card through afterburner.

Everything seemed good for a while, Heaven benchmark gave me the expected frame rates and score for my card, and my PC was working. I then ran Heaven again and it went downhill. PC crashing again, display errors. When running The Division after a few seconds the screen would go bright pink and crash.

At this point I am going to take my PC in to be checked, but thought I would first see if anyone can assist. Willing to answer any questions.

PC specs

Gigabyte DH67CL mobo
8 Gigs ram
i5 2320 3Ghz cpu
Adata SSD drive
550w Corsair PSU
GTX msi 970 gaming edition

Thanks all,

 

[xfia]

dont max out the power limit on your gpu

i had to tell you that.. there is more going on but i have a feeling this one is going to span for pages and pages

 

[FreedomEclipse]

you said you tried a different PSU.... Which brand and model of PSU did you try exactly?

the VS550 is a budget PSU and I wouldnt use it for anything more then that as quite a few reviewers have mentioned that Corsair use cheap chinese capacitors in the VS550. It would not surprise me if they did this for all of the VS product line. Best to stay away from the VS -- take a gamble with a CS/CX range --- I have had multiple bad experiences with the CS line up but for a lot of people it seems to work fine so YMMV.

XFX core power PSUs are great and quite often discounted at retailers where I am. Seasonics are good, Corsair RM line is good afaik, Corsair HX series is what i would call 'the standard' for a mid to high to higher end tier PC. I have used HX PSUs for years and used them with almost every build i make and they tend to last a bloody long time. They come with a 5 or 7 year warranty so they are definitely worth the investment.


Since you did mention that you plan to go SLi with 970s its best to get a solid 750-850w PSU (750w minimum) it will save you having to buy a new PSU in the long run if you want to add more hardware.

 

[BiggieShady]

dont max out the power limit on your gpu

This. Setting max power in nvidia control panel is not the same thing as setting the max power limit in afterburner. Former is prevention of automatic down clocking with less demanding games and latter is used when overclocking.

 

[Daniel Ward]

Currently using Corsair VX550W, tested with a Corsair TX750W.

I'm having a slightly different error at the moment. I can't even properly install nvidia drivers now, it goes through the entire installation process, restarts PC and then in device manager has an exclamation mark next to my GPU, Geforce experience tries to download the driver again, and Nvidia control panel just error's when I try to open. Also get the same error message when opening The Division.

Alright, but at the moment afterburner won't even detect that I have a GPU installed :/

 

[95Viper]

Check your temps... cpu and gpu.
Test your memory.
Turn off any OC.

It is possible both cards have rolled snake eyes.

 

[Daniel Ward]

GPU temps never about 60c, 40c when idling. CPU 50c idling, 80c under load. Never tried to OC the card.

Really stumped with this new error now. Device manager lists the latest driver but displays an error. Nvidia control panel wont open and Geforce experience tries to re download driver.

 

[xfia]

using windows 10 again?

 

[Daniel Ward]

No, still on 7 Ultimate, SP1 x64.

Trying a different driver now from last month.

I have been trying both express and custom installation. Clean installs all the time.

Possibly my SSD causing the issue?

 

[95Viper]

run ddu again, along with a registry cleaning with CCleaner, reboot, then try installing the drivers.

Oh and run them with administrative rights... you know right click and choose the option.

EDIT: You my want to disable any Antivirus/Antimalware and firewall, for grins and giggles.

Run windows update, too.

 

[Daniel Ward]

Ok, just saw that my pci-e slot on my mobo was out of date. I did update everything on the mobo after the Windows 7 installation. The card was also working after that, so not sure why the slot stopped working / lost its drivers? Weird.

ran DDU again and installed an older driver, installation was successful and device manager OK. Back to where I was now.

Tried to run Heaven benchmark and the screen went plain orange and then black after about 15 seconds. GPU temp was not above 70c. Had to hard reboot PC.

 

[eidairaman1]

Reseat the vid card

 

[FreedomEclipse]

Sounds like the GPu could be faulty then. Since you said it was new. cant you send it back for a replacement?

::EDIT::

Maybe the heatsink isnt making good contact with the die. try contacting MSI to see if they are willing to do an RMA or contact the previous owner and see if he is will to help with RMA.

Unless he's your friend or someone you know instead of a random stranger - you might not get any help or contact from him. Hes got your money, he doesnt care what happens

 

[Daniel Ward]

Reseated the card as well^^

Looks like it is just crashing after any kind of load. GPU temp is fine though, wonder if it's a CPU or PSU issue.

 

[Bill_Bright]

apparently a fairly common error with windows 10.

It is a fairly common problem but not more so with W10, or any other OS. I note you went back to W7 and got the same errors.

wanted to upgrade my card in any case, and got a second hand msi 970 gaming edition.

So it did this with your first graphics card too? Or just this second hand card?

I agree with the comments about the VS series of Corsair PSUs. They don't even have a basic 80 PLUS certifications. But trying a TX750 (assuming it s working properly) would suggest it is not a PSU problem.

80°C for your CPU may not be "hot", but it is pretty warm as the Tcase spec is 72.6°C. Not knowing where the sensor is for that 80°C reading means not sure if you are exceeding specs. Typically, the CPU would throttle down in speed, but stability could still be an issue. You could try opening the side panel and blasting a desk fan in there to see if it holds. If it does, you might consider improving your case cooling.

CCleaner is one of my favorite programs and I use it frequently and have for years. But even as a fan, I hold my breath when running the registry cleaner feature the first time on a system that has been around a while. So be sure to follow the prompts to backup up the registry. And regardless, ensure you have a good backup of any data you don't want to lose. And not to divert this discussion into yet another endless and heated debate about registry cleaners, FTR, CCleaner is only registry cleaner I ever recommend and IMHO, CCleaner should not be lumped in with all registry cleaners (especially by those who automatically attempt to discredit them) because (1) it is very conservative (where others are very aggressive) and (2) it always prompts to backup the registry first - and the restore process works. The people at Piriform have been around for many years and know what they are doing. In all my years using it, I have never - not once - had CCleaner put a computer into a irrecoverable state.

Since you said it was new

Actually, he said it was "second hand".

 

[FreedomEclipse]

Actually, he said it was "second hand".

My bad - I didnt read the entire post. he said it was new so i assumed he bought it from a retailer.

 

[Daniel Ward]

So it did this with your first graphics card too? Or just this second hand card?

The older card was crashing to the desktop and giving me the nvidia stopped responding and has recovered. I never ran any benchmarks on that card so could not tell if it would crash there or not.

The 970 is now crashing on the benchmarks but not giving me any driver errors, well not in the past few minutes. It's also crashing in games.

Yeah card is second hand, not sure what the seller would agree to, probably lost money at this point. Sigh

 

[eidairaman1]

Reseated the card as well^^

Looks like it is just crashing after any kind of load. GPU temp is fine though, wonder if it's a CPU or PSU issue.

Since the problem only started after you got the card, id say its the card.

 

[Daniel Ward]

Ok, probably, just don't want to believe it.

Any ideas why my GPU usage drops to about 40% under graphics load during a game? Temp is about 65. CPU usage 100% at 88c.

GPU usage then goes back to 90% and 60FPS when not under a hectic scenario / load in game.

 

[95Viper]

Is that an Intel motherboard or a Gigabyte board? DH67CL is a Intel board.

Are you using the latest bios?

 

[eidairaman1]

Look up cpubenchmark by passmark and youll see a general idea of where your cpu stands against a 2500K even

 

[Daniel Ward]

Is that an Intel motherboard or a Gigabyte board? DH67CL is a Intel board.

Are you using the latest bios?

It's an intel mobo, sorry.

I did not update the bios after reinstalling, i'm not sure how, will go to youtube and figure it out.

I have managed to play The Division / use my PC for about 30 mins now without crashing, seems stable.

Just have the FPS issue where it drops and my GPU sits at 40% usage. I guess it still could be the card. I did read up that my CPU may not pass info onto the GPU fast enough, and the CPU is on 100% usage. Will keep investigating.

 

[Bill_Bright]

Look up cpubenchmark by passmark and youll see a general idea of where your cpu stands against a 2500K even

But that really does not help with the card crashing.

If this were me and I was 100% certain I was supplying quality power (and all power connections were secure) and heat was not an issue, I would uninstall all the NVIDIA drivers, shutdown the computer and flip the master power switch on the back of the PSU (or unplug it from the wall). Then I would pull that graphics card and install an AMD card, boot, install the AMD drivers and make sure it works - no need for benchmarking or testing, other than to ensure the card is working.

Then I would repeat and reverse that process and return the NVIDIA card. This will absolutely ensure all graphics related settings and drivers are truly reset and fully overwritten. This is an old trick I learned many years ago with hardware devices, in particular, their drivers that are being stubborn for no apparent reason. Replacing with a totally different brand works for sound and network drivers too.

What I have found (or believe) sometimes happens when just reinstalling the same drivers is the installer sees a file or setting is present and simply moves on instead of overwriting or resetting it. If the file is corrupt or the setting is wrong, it stays corrupt or wrong. Changing GPU brands (from AMD to NVIDIA, or NVIDIA to AMD) forces every single associated file and setting to be overwritten and/or reset (hopefully correctly this time).

It does not always work, but surprisingly, often does. Of course it means you must have spare cards laying around and having a shop means I do. If you don't maybe a friend has an unused AMD card you can try.

If you end up with the same issue after this, and there is no other obvious issue (power, heat, malware, etc), then I am afraid your second hard card became second hand for a reason other than the first owner just wanted a new card.

 

[Daniel Ward]

But that really does not help with the card crashing.

If this were me and I was 100% certain I was supplying quality power (and all power connections were secure) and heat was not an issue, I would uninstall all the NVIDIA drivers, shutdown the computer and flip the master power switch on the back of the PSU (or unplug it from the wall). Then I would pull that graphics card and install an AMD card, boot, install the AMD drivers and make sure it works - no need for benchmarking or testing, other than to ensure the card is working.

Then I would repeat and reverse that process and return the NVIDIA card. This will absolutely ensure all graphics related settings and drivers are truly reset and fully overwritten. This is an old trick I learned many years ago with hardware devices, in particular, their drivers that are being stubborn for no apparent reason. Replacing with a totally different brand works for sound and network drivers too.

What I have found (or believe) sometimes happens when just reinstalling the same drivers is the installer sees a file or setting is present and simply moves on instead of overwriting or resetting it. If the file is corrupt or the setting is wrong, it stays corrupt or wrong. Changing GPU brands (from AMD to NVIDIA, or NVIDIA to AMD) forces every single associated file and setting to be overwritten and/or reset (hopefully correctly this time).

It does not always work, but surprisingly, often does. Of course it means you must have spare cards laying around and having a shop means I do. If you don't maybe a friend has an unused AMD card you can try.

If you end up with the same issue after this, and there is no other obvious issue (power, heat, malware, etc), then I am afraid your second hard card became second hand for a reason other than the first owner just wanted a new card.

Thanks. I have an old AMD, 7750 or something. Will give that a go.

Has not crashed for the past while, just seeing terrible FPS with low GPU usage, trying to figure out why the GPU usage is not increasing when needed.

I will try another quality PSU to rule that out completely as well.

 

[95Viper]

I did not update the bios after reinstalling, I'm not sure how, will go to youtube and figure it out.

BIOS Update Instructions for Boards and Kits

BIOS Update [BLH6710H.86A]
Version: 0160 (Latest)