I should just replace my GPU, right? (sanity check)

Jul 24, 2017
93
25
61
So, in late June I upgraded to a Ryzen 7 1700 w/Asus PRIME X370 Pro and Corsair Vengeance LPX DDR4-3000 (from an Ivy Bridge system). I kept using my old GTX 970 because it was doing everything I needed it to do.

When I first ran this system together, everything ran great and I had no issues in either games or productivity software (using Premiere, After Effects, and Handbrake extensively).

Then, a couple weeks ago, I started seeing a significant increase in driver crashes while playing games (as well as stuttering during heavy alpha-effect scenes, like explosions in Doom 2016), and my PC started locking up entirely whenever I tried to do any CUDA-accelerated rendering in Adobe Media Encoder.

I checked my GPU temperatures and I noticed that I was getting temps that are a bit higher than normal - not danger zone, but for example I'm getting 75C when running at only about 60% load (playing Dragon Age II with Vsync on, by no means a demanding workload for a 970). 75C seems more appropriate for a 100% load temperature. I also noticed that one of my GPU fans was starting to rattle, which is surely related.

So obviously there's an issue with my GPU. I have independently tested the CPU, RAM, and HDD the games are on, and they all reported no issues. But I'm paranoid that I might spend hundreds to replace my GPU and then end up with the same problems because the issue was something else.

So essentially the point of this post is to confirm that the problems are, indeed, just a matter of the GPU dying a slow death, and not potentially caused by some other system issue with my new parts. I'm OK if I have to replace my GPU (as it's 3 years old now) but I'd be angry if I have to replace any of my new components.
 

eddman

Senior member
Dec 28, 2010
239
87
101
Maybe you could test the 970 on another system to make 100% sure, or maybe another card on your main system.
 

Tweak155

Lifer
Sep 23, 2003
11,449
264
126
You can replace the fan most likely on the card, also maybe try underclocking? Lose a little performance but see if the problem goes away?
 
Jul 24, 2017
93
25
61
You can replace the fan most likely on the card, also maybe try underclocking? Lose a little performance but see if the problem goes away?

Yeah, I dialed back my overclock from about 1400 to 1200, but the problem persists. I'll see if taking it down to 1100 or so fixes my issues.
 

eddman

Senior member
Dec 28, 2010
239
87
101
Yeah I should note - I don't have any other cards or systems (or friends :( ). Thanks for the suggestion though!
Aren't there any PC repair shops where you live? I usually take any part that I can't test myself to them and they do it, for a fee, obviously. Not much though.
 

RadiclDreamer

Diamond Member
Aug 8, 2004
8,622
40
91
I'd honestly check the power supply, your card sucks a lot of juice and if your power supply is getting old or just getting weak/going out it will cause these issues also.
 
  • Like
Reactions: TurtleCrusher

psolord

Platinum Member
Sep 16, 2009
2,142
1,265
136
Would it cause my card to actually run *hot*, though?

If you live on the Northern Hemisphere, the increased summer temperatures could certainly affect the overall gpu temps. As the rest of the system of course.

However if you have strict temperature control in the computer room, this should be a non issue.

Look up on the net if there are similar issues with the specific model you have. The 970 is not that power hungry or hot, in order to deteriorate so fast, due to increased delta in its heat/chill cycles.

A clean install of the latest driver wouldn't be a bad idea either.

Memtest too. The system being new does not guarantee much.

Also a check if there's any background malware running mining behind your back. Utorrent has been accused of such stuff in the past too. Haven't used such apps in years though.

PS Make sure there isn't a rogue cable affecting the fans operation. I am sure you've checked, but you never know. Not treating you like an idiot. Just trying to help.
 
Jul 24, 2017
93
25
61
If you live on the Northern Hemisphere, the increased summer temperatures could certainly affect the overall gpu temps. As the rest of the system of course.

However if you have strict temperature control in the computer room, this should be a non issue.

Look up on the net if there are similar issues with the specific model you have. The 970 is not that power hungry or hot, in order to deteriorate so fast, due to increased delta in its heat/chill cycles.

A clean install of the latest driver wouldn't be a bad idea either.

Memtest too. The system being new does not guarantee much.

Also a check if there's any background malware running mining behind your back. Utorrent has been accused of such stuff in the past too. Haven't used such apps in years though.

PS Make sure there isn't a rogue cable affecting the fans operation. I am sure you've checked, but you never know. Not treating you like an idiot. Just trying to help.

Summer temps are actually cooler than normal around here, so I have had the windows open some, but the temp in the computer room usually stays around 75C, so I don't think that's the problem. I have run Memtest/chkdsk/Prime95 in an attempt to rule out other components as the source of the problem, and I've also run a full computer scan with ESET NOD32 and Malwarebytes to make sure there's nothing untoward going on with software.

The card is a Zotac base model 970, which I know doesn't have the highest-quality cooling system, so I figured it was just deteriorating after 3 years of heavy use.

I'll double-check on the cable situation - the "do'h" answer is frequently the correct one in my expericne...
 

psolord

Platinum Member
Sep 16, 2009
2,142
1,265
136
75C in the room sounds like you found out the culprit. Get out of there now!

j/k :p

You meant 75F I guess which would be around 24C which is great indeed.
 
  • Like
Reactions: Christopher Bohling
Jul 24, 2017
93
25
61
75C in the room sounds like you found out the culprit. Get out of there now!

j/k :p

You meant 75F I guess which would be around 24C which is great indeed.

LEL

Checked the cabling situation and the fan isn't hitting a cable. I also checked the PSU using speccy and all my voltages were within spec, so I don't think it's the PSU. At this point I think I'm going to take tential's advice and just buy a 1080 lol, I could try to clean the thermal paste but I just kinda want to make it into an excuse for an upgrade.
 

Guru

Senior member
May 5, 2017
830
361
106
I mean if you want an upgrade get an upgrade, but if the card hasn't died yet and it serves your gaming needs, then I see no reason why not to try and fix your problem.

1. You mentioned driver crashes, this 100% means you have corrupt driver installation. Even uninstalling the current driver won't fix it, because you have traces from your old driver left and are causing corruption. Best advice I could give you is to download Revo uninstaller pro, do an uninstall through there and use its feature to find all related traces of the driver and delete them.
Then do a "forced uninstall" just type in "nvme" or "nvidia" and see if it finds more traces, delete them as well.

Another option is to go to safe mode, use the complete removal tool to uninstall the driver, do a scan with Ccleaner, clean up temp files, restart and reinstall the newest driver.

2. Reset your GPU to factory settings. (you mentioned OC and only lowering it, RESET the card to factory settings). See if this works.

3. Summer days will result in increased temperatures in your PC, UNLESS you have a custom water cooling system, even a well built system with big in and out air fans will get increased temperatures in the summer.

4. Remove the gpu, unscrew the screws and take out the fan, see if there is something stuck there or anything causing a problem, check to see if the wires are all okay.

5. Reseat the card again on the pci-e slot, sometimes when replacing other parts you can accidentally slightly bulge the card a bit and not even notice it and just by it sitting wrong it can cause issues. So reseat it, make sure the screws are all screwed nice and tight.
 

RadiclDreamer

Diamond Member
Aug 8, 2004
8,622
40
91
LEL

Checked the cabling situation and the fan isn't hitting a cable. I also checked the PSU using speccy and all my voltages were within spec, so I don't think it's the PSU. At this point I think I'm going to take tential's advice and just buy a 1080 lol, I could try to clean the thermal paste but I just kinda want to make it into an excuse for an upgrade.

How are you checking the PSU? They should always be checked under load, furmark+prime95 etc. Otherwise you are basically saying your broken leg doesnt hurt when you arent using it.
 
Apr 20, 2008
10,067
990
126
I'm looking towards the PSU to be honest. I had an Antec Truepower Trio 550w that unexpectedly took a dump weeks after I upgraded from a heavily OCed C2Q to my fx8350 usually kept at stock. Same symptoms (gpu driver crashes mostly) and my GPU temps were always 80-90C under load. I put in what I though was an inferior Thermaltake TR2 and my system has been good ever since.

Also if your PSU has multiple 12V circuits I'd suggest checking to make sure one isn't overloaded.
 
Jul 24, 2017
93
25
61
I'm looking towards the PSU to be honest. I had an Antec Truepower Trio 550w that unexpectedly took a dump weeks after I upgraded from a heavily OCed C2Q to my fx8350 usually kept at stock. Same symptoms (gpu driver crashes mostly) and my GPU temps were always 80-90C under load. I put in what I though was an inferior Thermaltake TR2 and my system has been good ever since.

Also if your PSU has multiple 12V circuits I'd suggest checking to make sure one isn't overloaded.

Thanks for the help! Ultimately I did just replace my GPU with a GTX 1080 and that solved the problems...though I mean I admit it is just nice to have a 1080 lol.

I charted voltages on the 12V/5V/3.3V rails while under heavy load for 30 minutes (Prime95 + Superposition on loop) and I didn't see more than a +/-2% voltage variance on any of them, so I'm pretty sure my PSU is OK.