Should I RMA my GTX 980?

[skeedo]

Built my new system about 2 weeks ago on Friday the 13th, lucky me. I keep getting the dreaded 'Display driver Nvlddmkm stopped responding and has successfully recovered' error. Thing is, I only get this error playing Swtor and Swtor only. Not a single crash in FC4, Dragon Age, Futuremark Firestrike, nor Unigine Heaven Benchmark. Evga GTX 980 SC ACX.

Swtor using an extremely outdated game engine leads me to believe that there is no problem with the card, especially more so since it's the only game that's giving me crashes. However there's still that nagging feeling that there could be a problem with the card that's only causing an issue with this game, it seems that RMAing had solved the issue for others in the past. I only have another week to decide if I want to RMA. Was considering a different brand 980 as well, however I'm going to be liable for 15% restocking fee if I choose this route, and this is not a cheap card.

The hardest part about diagnosing the problem is that it mostly happens in graphic intense areas in Swtor, typically 8 or 16 man raids so I can't just jump into one of these on the fly. Furthermore sometimes it will take a couple hours of playing before I get a crash. I've reverted back to 344.75 drivers and since then I've got two nvlddkmkm stopped responding errors however Swtor did not crash with Apphang, but it did take a minute to reload. I've tried every available fix out there I could find. I've also tried messing with GPU clock, but I don't see why I should have to do this when FC4 and DA:I run without issue. The card is definitely not overheating, I haven't seen temps much over 63C with my Evga Precision fan curve.

So what's the consensus? a) the card is fine and Swtor is junk or b) It is possible something is wrong with the card and I should RMA before the 30 day period is up?

It just sucks that I drop $2700 on a new PC and instead of having to deal with 5fps during Swtor raids, now have to deal with crashes!

 

[DaveSimmons]

> Furthermore sometimes it will take a couple hours of playing before I get a crash.

That could just mean the game engine has "leaks" of RAM and resources that build up over time.

I've played many games where it's only safe to play them for 1-3 hours before you need to exit and restart.

 

[TemjinGold]

If there's only one game giving you trouble, it's not the card.

 

[96Firebird]

Roll back your CPU overclock and see if the problem persists.

 

[skeedo]

Highly doubtful it's the CPU, I was checking Event viewer and I did get the Nvlddmkm error before I even overclocked. Furthermore I've had some pretty long stress tests in AIDA64 as well without issue.

One thing I don't like is how the card will go boost mode with the other games, 1328Mhz GPU and 1218mV but for Swtor it stays at 1265Mhz and 1143mV. Wanted to just try running boost mode fulltime in Swtor...I activate Kboost in Evga Precision, it claims that GPU clockspeed and Voltage are boosted however GPU-Z says they're not. The Evga Precision interface is so counter intuitive I really don't want to bother and try adjusting GPU and voltage manually.

 

[Termie]

It's not the card. SWTOR is [edit] an older game. It can't possibly load the GPU sufficiently to activate full boost/voltage, and the problems you're having are clearly driver-related. That doesn't mean it's not a problem, it just means an RMA isn't going to fix it.

 

[Magic Carpet]

Try underclocking the gpu/memory/cpu slightly. Play some.

 

[ocre]

Memtest, memtest, memtest.

"Display driver _____ stopped responding" is often blamed on the card when it is usually caused by a memory issue, like unstable memory and errors. It just as well could be triggered by overclocking as this can produce errors too.

Others have suggested a memory leak. This is a memory issue too, but it is of software origin. Memtest will check your HW. I suggest checking it at least overnight or longer. You can leave your system overclocked when you do this.

But you also need to know....
It doesn't matter how stable your overclock seems, or that you haven't had an issue before. It doesn't matter if you can run hours of occt, prime95, or ibt. When you overclock, some will tell you there is no such thing as 100% stable. And the truth is, this isn't far from the truth. You can be stable in 99.9% of every stress test and app there is, but find one random program that finds and exploits errors until there is a crash. Many many people have experienced this, it is just the way it is.
Just because you have been fine in everything else you do, this doesn't rule out your overclock causing this crash. The only way to rule out your overclock is to put your chip back to 100% stock and see if your issue still exist.

But, I highly recommend you check your system ram really really good. I had some corsair ram go bad on me and it was the last freaking thing I expected to be my issue. This ram was a month old and all of a sudden I started having an issue.
Anyway, you may also want to stress test your GPU but I would check the system ram while overclocked first, if that checks out then return your chip to 100% stock, if you still have driver crashes, then check other things like stressing the GPU, etc.

I once had a bad hd that caused very odd things to happen, seagate tools found issues in the long and short test. I was having crashes and finally a blue screen code had a list of possible causes. Bad drive was way down on the list but It turned out to be the culprit

 

[VirtualLarry]

SWTOR is 14 years old.

The MMO is 14 years old? I thought it only came out 5 years ago or less.

 

[Termie]

The MMO is 14 years old? I thought it only came out 5 years ago or less.

Sorry, should have realized there aren't raids in the single player swtor rpg from 2001.

But the point still stands. Old engines will not load a 980 sufficiently to run at full boost/voltage.

Following up on other comments, the OP could skip all the stress testing to start and just run at stock. He has a huge overclock on that CPU, which could be negatively affecting memory stability and vice versa. He can then fix the overclock or investigate drivers further, depending on the outcome of his stock gameplay.

 

[Ranulf]

Check the swtor forums for similar issues? That game engine is... well, not optimized for any card. I remember many reports from many users having problems with high end stuff at launch + 1 year.

 

[MeldarthX]

It's not the card. SWTOR is 14 years old. It can't possibly load the GPU sufficiently to activate full boost/voltage, and the problems you're having are clearly driver-related. That doesn't mean it's not a problem, it just means an RMA isn't going to fix it.

ummm TOR isn't 14 years old maybe 2 you're thinking of Knight of the Old Republic with TOR is based off of

 

[MeldarthX]

Memtest, memtest, memtest.

"Display driver _____ stopped responding" is often blamed on the card when it is usually caused by a memory issue, like unstable memory and errors. It just as well could be triggered by overclocking as this can produce errors too.

Others have suggested a memory leak. This is a memory issue too, but it is of software origin. Memtest will check your HW. I suggest checking it at least overnight or longer. You can leave your system overclocked when you do this.

But you also need to know....
It doesn't matter how stable your overclock seems, or that you haven't had an issue before. It doesn't matter if you can run hours of occt, prime95, or ibt. When you overclock, some will tell you there is no such thing as 100% stable. And the truth is, this isn't far from the truth. You can be stable in 99.9% of every stress test and app there is, but find one random program that finds and exploits errors until there is a crash. Many many people have experienced this, it is just the way it is.
Just because you have been fine in everything else you do, this doesn't rule out your overclock causing this crash. The only way to rule out your overclock is to put your chip back to 100% stock and see if your issue still exist.

But, I highly recommend you check your system ram really really good. I had some corsair ram go bad on me and it was the last freaking thing I expected to be my issue. This ram was a month old and all of a sudden I started having an issue.
Anyway, you may also want to stress test your GPU but I would check the system ram while overclocked first, if that checks out then return your chip to 100% stock, if you still have driver crashes, then check other things like stressing the GPU, etc.

I once had a bad hd that caused very odd things to happen, seagate tools found issues in the long and short test. I was having crashes and finally a blue screen code had a list of possible causes. Bad drive was way down on the list but It turned out to be the culprit

This; nicely put Ocre....

 

[Subyman]

Yup, Ocre is right. I've had systems that are perfectly stable in memtest and CPU stability tests but then still crash during normal gaming. Set every thing to stock and see if it still happens. You don't need any OC to run SWTOR maxed anyway.

 

[skeedo]

Welp after dialing back the GPU 50Mhz I played all day yesterday without a single crash so I guess this game just can't handle the default GPU clock. Wasn't worth attempting to going back to stock CPU speed, if Firestrike and 30 hours of FC4 gameplay can handle the OC then anything should be able to. Yes 4.5Ghz might see like a big OC for this chip, however need to remember that this series of Asus motherboards was specifically built for OCing. I could probably get to 5Ghz without a hiccup but waiting for summer to roll around before trying that. For shiggles I also did try a memtest overnight, 0 errors.

Anyway Swtor might as well be 14 years old the way it behaves. The Hero Engine was supposedly created in 2012 however it runs like it was created in 2001. Constantly getting all kinds of graphic anomalies, and furthermore have to deal with framerates in the 20s (which I guess is better than single digits on my old PC). Such a shame, but between the outdated engine and the way the game has been headed in general I think my days there are numbered. Might be time to try a more modern MMO like FFXIV.

Thanks everyone for your suggestions.

 

[Termie]

If you're getting game crashes at the shipping GPU clocks (i.e., not a manual overclock that you've set up), it might be worth RMA'ing. EVGA's support is very good, so if you decide it's not worth it for SWTOR, but then get problems in other games, you can just deal with it at that point.

 

[RadiclDreamer]

It could also be power, what power supply and what are your rails looking like?