NVIDIA GTX 970 getting horrible FPS on games it used to run perfect. [SOLVED]

Page 2 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.

sbkspyder

Reputable
Feb 17, 2015
22
0
4,510
(((((Too Long - Don't Want To Read - Problem and Solution)))))

Problem: Video card was showing display but not using the graphics card for during gaming or really anything. Core and Memory Clocks were at a constant rate (Usually 135/165). Same results after using another power supply, using another system, wiping windows, wiping drivers and bios flash.

Solution: GPU-Z was used to monitor card, as suggested by cst1992. TDP was at 232%+, indicating that there was a problem with the card. New card was purchased and works properly. Old card was sent off for RMA. PSU is suspected cause but not confirmed.

--------------------------------
Original Post
-------------------------------

Hello All,


I am here because I am truly done trying to search, test, reset and search again. I have a GTX 970 SSC from EVGA that I bought in April I believe. When I first got it the thing ran perfectly. All my games were set to high/ultra/ultra high and ran without issue. Skyrim, Planetside, Max Payne 3, ARK, BF4, CS:GO, Sims 3, Arma 3, Cities : Skylines and so on. After a few months of playing without issue I even found the receipt for it and realized that it came with Witcher 3 for free, so I downloaded and installed that. It played perfect on the highest settings. I thought I was set until about a month ago. I messed up by dual booting Linux and then accidentally deleting linux (Which happened to be using that partition with Grub as the primary bootloader). I was able to get all back together (Minus the Linux install) and my games still seemed to run great but it seems like out of nowhere it just went to junk.

In Skyrim I was seeing an FPS of around 90+ almost constantly. ARK was around 60 usually. I am now at the point where I am seeing pretty much 20 FPS on Skyrim on High and Ultra settings, 18 FPS on ARK with the lowest settings possible and 16 FPS on Witcher 3 with the lowest settings possible. I've always made sure my drivers were up to date, always keep the case and internals clean, run MalwareBytes when I am not gaming, did disk defrags on the HDD that the games are stored on and I even used DDU to wipe it all off and install again. None of these things have helped. I've followed numerous posts on this site about changing NVidia Control settings, lowering gaming settings and so on with no luck. Mind you that I am playing Witcher 3 on the lowest settings possible and getting 18.

My rig is as follows.

AMD FX 9590
GTX 970 SSC
ASUS SABERTOOTH 990FX R2.0
120 GB Kingston SSD
1TB WD Caviar Blue
Corsair H75 Cooler
16GB Corsair Vengence 1600 DDR3
EVGA SuperNOVA NEX 650W 80+ Gold

Dual Monitors at 1920x1080 each.
Windows 7 Ultimate

My GPU temps are usually right at 34c and CPU are usually 35c - 45c (Lots of fans and sits in a cold room).
GPU does show full usage. CPU shows about 30 on 3 cores depending on the game (Witcher has put a few of them at 50 usually not higher as confirmed by MSI.


I know something is surely wrong here and ran Firestrike again to prove it. My previous score (4/25/2015) was 9093. My most recent score (12/13/2015) was 1109. If that doesn't scream PROBLEM at the top of it's lungs then I don't know what does.

Please help people. I built the thing myself so I am not afraid to rip it back apart (Actually did that last night for cable managements sake). Let me know what I need to do. So far the next things on my list are re-seating the CPU (Which it should have no reason for) and if that doesn't do it then I am prepared to wipe it clean (Or at least put Windows on another HDD and try again).
 
Solution
Before you do that, do one thing. Put the 970 in a system that can handle it(500W or more PSU rating, a good brand), and repost the GPU-Z screenshots.

It seems the OCZ(Firepower is the name of the new owner of their PSU business) PSU caused something to blow internally in the GPU over those 7 months you had it. 232% TDP is NOT normal. If that's consistent, RMA the card and don't use the new one on the OCZ or the CX.
Just picked up a CX750M (Best I could do locally). Will try it once I get home and see if helps. If not, I will be throwing the card in an older PC and seeing if that gives me any results.
 


Worse? I'm confused now. What makes this worse than the one I already have? I don't plan on keeping this thing but I just want to know for future reference.
 
li5n5.jpg


zluo9d.jpg


These are what GPU-Z is showing me. The CX750M doesn't seem to be changing anything. Everything is pretty much the same. Going to toss it in my other system as soon as the wife gets off of it.
 
Omg dont listen to them, there is nothing wrong with your power supply or the new power supply, they are both good but not ideal or excellent. There should be something wrong with your graphics card, since I dont beleive the power supply doesnt get enough power to it, because it was all working fine until that moment... If it doesnt make any difference with the new CX power supply, then the power supply is a 100% not guilty!
 
There are People on this forum Just hating everything they can idk why anybody would do that it seems unfair that he spends money on useless things... I have a LC Power 850w chinese power supply, which runs two gtx970's Just fine for nearly 1 year, just dont hate please... It was actually bought used 4 years ago andand was in two of builds.
 
Slocenia, If you think people on this forum just hate, then you have clearly been on the wrong side.


The PSU Is most LIKELY the issue. AMD recommends a PROPER, HIGH QUALITY, 1000 watt PSU. The Cx750 failed HardOCP's testing, and the NEX is not a very good unit. A bit better than the cx, but not enough. The GPU is most likely not the issue. However, I do see something.

My recommendation to you. Take the MOST demanding game you have.

Open MSI After burner and then play that game for about 20 minutes. Exit out, and look at the charts in MSI after burner. Take a picture so we can see if there is something wrong.

Your GPU should be at a consistent clock rate, somewhat cool temps, and be having 100% usage.

Your CPU usage may be high, but make sure there is NO throttling. If, at anytime, the game crashes, I'd suspect over heating or PSU failure.
 
Before you do that, do one thing. Put the 970 in a system that can handle it(500W or more PSU rating, a good brand), and repost the GPU-Z screenshots.

It seems the OCZ(Firepower is the name of the new owner of their PSU business) PSU caused something to blow internally in the GPU over those 7 months you had it. 232% TDP is NOT normal. If that's consistent, RMA the card and don't use the new one on the OCZ or the CX.
 
Solution


So your definition of 'good' is 'running upto 700W load for 1 year and nothing blowing'? I'm sorry but I need a better definiton than that.
 


Swapped the card into another machine. Same results. High TDP, no movement of core/memory. Seems that it is the card. Just put in a request to RMA the thing (Sadly, probably won't take place until after the Christmas season ends so I will be stuck with my old GPU for a bit). I guess I will be waiting to see what they respond with from my ticket but I'm pretty sure that is it.

I will keep updating as info comes my way until the issue is completely resolved. In the mean time, thank you all for your time and help. It was very much appreciated.
 
@cst1992 Well as I can tell everything still works properly in my Pc, obviously my power supply has more quality than I thought. As for @sbkspyder it is already to late. Just return the CX power supply and idk buy something like a seasonic or xfx that everybody is recommmending, but you probably have ruined ur system anyway so... Obviously the LC Power 850w power supply is better than everyone thinks since at me everything works correctly for nearly 1 year in this current build and 2 years in a an old build and 1 year in the one's build I bought it from
 


RMA should definitely help performance. My earlier guess was that the card was going into fallback mode(405 MHz GPU, 405 MHz memory) due to an unstable overclock(happens even on factory-overclocked cards). But this is something new.

Even if it's a faulty sensor, shouldn't take the risk. You can also monitor power draw of your system when you're gaming, if you have one of those Kill-A-Watt meters or a multimeter, but I'm guessing it doesn't matter.
It sucks that you have to wait that much. Hopefully the new GPU will solve things.

 


Yeah thought it sounded like it was the graphics card at fault.

Sorry some people convinced you it was the PSU... FYI PSU faults do not normally cause these kind of symptoms. Faulty/bad quality psu would usually result in a crash or the system powering off.
 


I think that the card being bad is one of those things that people don't want to accept right away. Deep down I knew that it had to be the card simply because the display was still working and nothing was happening other than that. I just wish I would have accepted that earlier before I wiped my boot drive. lol. Live and learn though. Might go grab an AMD card and sell the RMA when it gets back. NVidia has been letting me down lately.

No problem on the PSU part either. Going to just go take it back to the store and get my refund.
 
@Rob, though I believe the card was not at fault from the beginning. Remember how sbkspyder said the score dropped from ~9000 to ~2000 suddenly, and for a sizable interval, was working normally? I still think it's not just the card, something else is also the problem.
 


He tried the card in another computer and it displayed the same problem... How can that not be the card at fault?

Just because the card worked correctly before doesnt mean it cannot go faulty.
 
The card ran perfectly well for 7 months, and now it's gone faulty. So it was not a manufacturing defect, and something went wrong after a decent time interval. So I'm saying it's not just the card, the PSU could be at fault as well(The Firepower one I'm talking about). Since the card is under warranty, it can be sent for an RMA, but I'd suggest him not to use the new one with that PSU.
 


Very highly unlikely to the a fault of the PSU in my opinion. The symptoms dont match any issues I have seen with PSUs, where as its not totally uncommon for cards to fail in this fashion.

Again this forum points blame at lower quality PSUs. Sure its not the greatest unit, it doesnt get a brilliant review, but its not dangerous or likely to blow a component in normal conditions. The main failure of the unit is it cannot maintain a 700w load, well the op is no where near 700w, more like 450w.
 
It is indeed the card that went bad. Bought another 970 while I send the other in for RMA. Working great (Not quite better than the last one but I think I still have some fine tuning to do to get it back to normal). Loaded up Witcher 3 with all of the highest settings possible and cannot make that thing flinch after trying for about an hour. Core and memory clocks are moving as they should. Going to go on Amazon after Christmas is over and grab an 850w EVGA, as suggested here. Figure I'd rather not take the chance on it actually being the OCZ, though I highly doubt that was the case here as it probably should have been all or nothing if the PSU was to blow something (Meaning I probably shouldn't have been able to still get a display with it, adjust the fans, view it's temps/clocks/power and shouldn't have been able to flash it). I will just toss that over into my wife's machine and see if it wrecks anything.


Again, I want to thank you all for your time and effort that you put into helping me with this issue. It is highly appreciated. I wish you all the best for the holidays, the new year and the future. Have a great one.
 
Ran into one more issue after swapping the cards. Kept randomly freezing while I was playing my games. I thought it was the new GPU due to it only happening when I was playing my game. After troubleshooting it a bit I found that it was actually the CPU. I cleaned the CPU/CPU Cooler when I was troubleshooting the previous problems and I apparently must have been butterfingering it as it had paste everywhere but the center. Cleaned it up one more time and put it all back together and have not had any issues since (After playing 2 hours of CS:GO, 3 hours of Skyrim, 1 hour of Witcher 3, 1 hour of ARK and not to mention downloading all of those games as well as all of my other games). Seems all is well now. over 7 hours of GPU intensive gaming, 5 hours of downloading a few hours of driver installs/updates/configurations and other gaming related tasks. New PSU is on its way from Amazon. Hopefully all stays good.
 
I finally found the problem for the computer freezing. The card was the original problem and it was solved after getting a new one. The new problem, which was the computer freezing and turning off, was due to the CPU cooler. The H75 just wasn't cutting it for the TDP. I swapped over to a Lepa Aquachanger 240 and have not had one single problem yet. The computer has been running, playing games, streaming and just normal every day use and it has held up without issue. Was able to turn my settings all the way up on all games and no issues. Hope everyone with this 9590 know that you have to have something big (220+ tdp) to cool this down.