GTX 980 Ti Causing System Freezes/Frame Stuttering

[Chalenor]

So, I have this Gigabyte 980 Ti G1 Gaming that I've had for a while but haven't been using since it started giving me issues, and I bought a 1080 Ti to replace it. However, I'm trying to use it now in a second older system and am still having the same issues I had back then. At this point, the issues persist across entirely different hardware configurations (different PSUs, motherboards, CPUs, etc...).

 

I kind of really need this second system to work so my mate can use it, though, so I was wondering if anyone here had any suggestions before I have to figure out if there are even options for paying to ship it to somewhere to have it repaired. I've tried looking the issue up online, but couldn't find anything that worked for me.

 

Here's the specs of the current system the 980 Ti is installed in.

• Motherboard: ASRock Z97 Extreme6
• CPU: Intel Core i7-4790K
• CPU Cooler: Corsair H80i
• RAM: G.Skill Ripjaws Z 32 GB (8x4) 1866 Mhz CL9-9-9
• Graphics Card: Gigabyte GeForce GTX 980 Ti 6GB G1 Gaming OC Edition
• Power Supply: SilverStone Strider 700 Watt 80+ Titanium Efficiency
• Windows 10 Pro 20H2 19042.804
• Nvidia Driver 461.72

 

The Issues

The 980 Ti functions fine on the desktop, but the moment I go into a game or benchmark that puts significant load on the card, GPU-Z reports it hits power limit at around 25-40% TDP, and the card clocks down to as low as 500-600 MHz core and 405 MHz memory clock, which quickly halves the frame rate and causes stuttering as the card bounces up and down in clock rates. Eventually, the entire screen will turn white, which requires the system be fully powered down by holding the power button.

 

Some applications tested with this behavior are: GTA V, Killing Floor 2, Half-Life: Alyx (in VR), The Forest, Elite Dangerous, Heaven Benchmark, Furmark

 

Things I've Tried That Made No Difference

• Installing overclocking software to raise/lower the power, temp, and voltage limits
• Underclocking core and memory clocks
• Swapping the card into different computers with all different hardware
• Fresh install of Windows 10
• Different Nvidia drivers, including most current, with DDU in between them
• Making sure PCI-E slot is set to 3.0 instead of auto in BIOS
• Running fans at 100% (The card runs mid to high 40s Celsius on core during 100% fan speed)

 

Screenshot from GPU-Z during the issue: https://imgur.com/a/kGMSk3F

 

[WhitetailAni]

The VRMs might not be cooled properly, meaning that they're providing less power, forcing the GPU to underclock.

Or they could be aging (not sure if VRMs can do that).

Is the cooler securely mounted?

[Chalenor]

1 minute ago, FakeKGB said:

The VRMs might not be cooled properly, meaning that they're providing less power, forcing the GPU to underclock.

Or they could be aging (not sure if VRMs can do that).

Is the cooler securely mounted?

I've never messed around with the cooler, but it seems quite secure. Everything is stiffly in place. All three fans run fast and loud at 100% without causing anything to vibrate. I don't think it has VRM sensors listed on GPU-Z for this card.

[WhitetailAni]

1 minute ago, Chalenor said:

I've never messed around with the cooler, but it seems quite secure. Everything is stiffly in place. All three fans run fast and loud at 100% without causing anything to vibrate. I don't think it has VRM sensors listed on GPU-Z for this card.

Have you ever repasted the card or done any overclocking?
Not saying it's the cause, but it's a good place to start.

[Chalenor]

3 hours ago, FakeKGB said:

Have you ever repasted the card or done any overclocking?
Not saying it's the cause, but it's a good place to start.

As far as overclocking, only very mildly in the past. With the card's limiter on voltage and power even before it stopped going above 50% TDP, I maybe only got an extra 100-150 MHz out of the slider or so before it was unstable. No OC software even lets the voltage go above +800 MV or around there since I never did anything drastic like change the card's BIOS.

 

Haven't touched anything with the paste or heat sink.fan arrangement yet. It didn't seem to me like it should be a heat issue since the power limit issue occurs immediately. It's like as soon as TDP is requested to go above 50%, it snaps back down to 25% and lowers clocks on the dot, but if I do anything light where TDP floats around 20% (frame rate limiter to about 15 FPS in a game), everything stays steady.

 

I found a "Hot Spot" measurement in GPU-Z that wasn't displaying by default, but it only reaches up to around 67 Celsius with the core clock at 61 Celsius under Heaven Benchmark. It white-screened soon after I was reading these measurements, so I can't really have time to heat the card any further.

[WhitetailAni]

1 minute ago, Chalenor said:

As far as overclocking, only very mildly in the past. With the card's limiter on voltage and power even before it stopped going above 50% TDP, I maybe only got an extra 100-150 MHz out of the slider or so before it was unstable. No OC software even lets the voltage go above +800 MV or around there since I never did anything drastic like change the card's BIOS.

 

Haven't touched anything with the paste or heat sink.fan arrangement yet. It didn't seem to me like it should be a heat issue since the power limit issue occurs immediately. It's like as soon as TDP is requested to go above 50%, it snaps back down to 25% and lowers clocks on the dot, but if I do anything light where TDP floats around 20% (frame rate limiter to about 15 FPS in a game), everything stays steady.

 

I found a "Hot Spot" measurement in GPU-Z that wasn't displaying by default, but it only reaches up to around 67 Celsius with the core clock at 61 Celsius under Heaven Benchmark.

Sounds like the VRMs aren't able to provide enough power to me, though why I don't know.

Just to rule it out, can you repaste the card and remount the cooler?

[Chalenor]

22 minutes ago, FakeKGB said:

Sounds like the VRMs aren't able to provide enough power to me, though why I don't know.

Just to rule it out, can you repaste the card and remount the cooler?

I'd have to buy some paste first. Since every CPU cooler I've gotten comes with thermal paste pre-applied, I've never actually gotten any tubes of the stuff before.

[WhitetailAni]

Just now, Chalenor said:

I'd have to buy some paste first. Since every CPU cooler I've gotten comes with thermal paste pre-applied, I've never actually gotten any tubes of the stuff before.

Well, MX-4 is decent for not too much if you need some.

Though if you don't wipe off the paste and just remount it, it should be fine for testing.

Just make sure to screw down securely.

[Simon_1599]

Hi, i have the same problem..

 

Do you fixed it?