Samsander

stalinx20

DnBrn
Well, if I've learned anything, it's don't pay for the stock overclock models.  Pretty much spot on, I assumed that a stock overclock would be stable and that they had rigorously tested it.  It is so odd though that it's not outright unstable, it's just in certain games.  Boosts to 1455mhz in 3d mark and doesn't crash, but crashes in Elite Dangerous.  Runs Wither 3 on Ultra with no crash, but crashes in The Repopulation.  This is why Nvidia themselves say it's not the card but when game developers then blame the card and EVGA wants to do an RMA, what are you supposed to think?  I don't know, I'm not a GPU tech or guru, so I'd expect EVGA to know and fix it.  Instead we just RMA RMA RMA until one finally works or pay more for a different card.  I'm sorry but I didn't pay all that money for a refurb within 3 months of purchase and I hate that I have to now step up for an additional cost.  Doesn't seem I have a choice though, I've really thought about it, and it seems like it's just the best course of action to upgrade in order to avoid further instability while I have the opportunity.

Never again.  I'll have to keep an eye on how future products go now, may have to switch brands if it continues to work like this.  I hope this is just a fluke and they made a mistake this time since I've been all about EVGA for years.

You're better off just buying reference models and overclocking yourselves. The good part about this is, if your overclock doesn't work, at least you can go back to reference clocks, and start over. You can't start over on an already overclocked card by the manufacturer, you're stuck.

Why not? You can downclock as easily as you can overclock.

Sure it sucks to see the -100 or -200 MHz but you can do it if you need to.

Then why spend the extra money when you can do it yourself...

Samsander

stalinx20

DnBrn
Well, if I've learned anything, it's don't pay for the stock overclock models.  Pretty much spot on, I assumed that a stock overclock would be stable and that they had rigorously tested it.  It is so odd though that it's not outright unstable, it's just in certain games.  Boosts to 1455mhz in 3d mark and doesn't crash, but crashes in Elite Dangerous.  Runs Wither 3 on Ultra with no crash, but crashes in The Repopulation.  This is why Nvidia themselves say it's not the card but when game developers then blame the card and EVGA wants to do an RMA, what are you supposed to think?  I don't know, I'm not a GPU tech or guru, so I'd expect EVGA to know and fix it.  Instead we just RMA RMA RMA until one finally works or pay more for a different card.  I'm sorry but I didn't pay all that money for a refurb within 3 months of purchase and I hate that I have to now step up for an additional cost.  Doesn't seem I have a choice though, I've really thought about it, and it seems like it's just the best course of action to upgrade in order to avoid further instability while I have the opportunity.

Never again.  I'll have to keep an eye on how future products go now, may have to switch brands if it continues to work like this.  I hope this is just a fluke and they made a mistake this time since I've been all about EVGA for years.

You're better off just buying reference models and overclocking yourselves. The good part about this is, if your overclock doesn't work, at least you can go back to reference clocks, and start over. You can't start over on an already overclocked card by the manufacturer, you're stuck.

Why not? You can downclock as easily as you can overclock.

Sure it sucks to see the -100 or -200 MHz but you can do it if you need to.

Because you pay a premium pretty much just for the overclock.  There's nothing extra that the FTW model has to justify the extra money.

Or I'm completely wrong and the card is supposed to have better components etc... Whatever.. I've said that I've been lucky, but there's a lot of people that haven't and that's a symptom of a bigger issue and not the first time I've had an issue with an over clocked card.  Doesn't seem worth it in hindsight.

Hello all, been hip deep in some other projects and this thread was brought to my attention today.  Previously, Customer Service had mentioned to myself and the QA/QC teams that there were complaints regarding this specific card. 
Separately I had read things from both Frontier and Blizzard stating issues with ALL overclocked cards, this leads me to believe we may be looking at 2 separate issues: 
1) Blizzard and Frontier software simply doesn't like OC'd cards (infact when the issue started with elite, it happened on NVidia AND AMD cards as well) seen plenty of posts indicating that.   Blizzard has said this off and on over the years and it is usually resolved by the combination of a Game patch and a Driver update.  Mainly why I believe there are 2 separate issues.
2)  Next, specific complaints about the 2986 (FTW) card.  This week we did receive a few of the FTW cards back that were RMA'd so we could test them and see if we can find a root cause of the problems.  Most of the cards we have received we cannot force a failure on (barring ridiculous overclocks/ turning off fans/nonsense like that) including the OP's card, we did however get a few with the problems described.  Currently we have several cards that we are sending to engineering, that we had received back from RMA citing this issue and had been tested here already, this is a mix of cards that both passed and failed our testing to provide as much info as possible for narrowing down the route cause.
 
This community has largely functioned as well as it has because (with a few feisty exceptions) this community REALLY tries to help each other.  So on that note I will make a request.  Any of you who do an RMA for the 980FTW having issues EVERYWHERE (NOT if it is ONLY Elite, Hero's of the Storm, or The Repopulation (not familiar with Repopulation before this thread) and it works great at stock clocks or even overclocked EVERYWHERE else but these couple games, then this will be a game and/or driver issue) if the cards are NOT downclocked, please follow up with me, drop me a PM.  Let me know the RMA number and a FULL system spec, hardware and OS.  I frankly have no interest in pointing fingers at any company, it is not my nature, BUT, with an issue like this that we are having a hard time reproducing here the more information we can get, the better.  If you want to put your spec here, that is fine, but if the info can be PM'd to me, or even emailed that would be a big help.  The more information we can get, the better.

It's good to see some action taking place on the FTW. Nice. I hope you all get it resolved soon.

Never expected this but i am very happy about it. I will test a bunch of games before i will sent it in for step up and sent a PM with system spec's, games tested, duration, settings, temps and everything else i find noteworthy. Hopefully You guys can somehow fix it. Good for anyone who isn't able to step up and only has the RMA option. 
 
I am probably one of the feisty one's i bet.  
 
Regards 
 
Almost forgot:

Samsander
3dmark skydiver is a perfect example of this. With power draw jumping all over it will crash the card in minutes.

 
I second this. It crashes everytime for me. 

I'll verify that some of the benchmarks wouldn't crash my card; SkyDiver would do it every time, as would pretty much any game that was remotely graphically demanding.

djmorgan

Jshiver90
I'll verify that some of the benchmarks wouldn't crash my card; SkyDiver would do it every time, as would pretty much any game that was remotely graphically demanding.

EVGA GTX 980 FTW

Yes, you're right. This thread and the feedback asked from the EVGATECH is totally unwarranted because your card works fine. How silly of everyone.

EVGATech_DaveB
Hello all, been hip deep in some other projects and this thread was brought to my attention today.  Previously, Customer Service had mentioned to myself and the QA/QC teams that there were complaints regarding this specific card. 
Separately I had read things from both Frontier and Blizzard stating issues with ALL overclocked cards, this leads me to believe we may be looking at 2 separate issues: 
1) Blizzard and Frontier software simply doesn't like OC'd cards (infact when the issue started with elite, it happened on NVidia AND AMD cards as well) seen plenty of posts indicating that.   Blizzard has said this off and on over the years and it is usually resolved by the combination of a Game patch and a Driver update.  Mainly why I believe there are 2 separate issues.
2)  Next, specific complaints about the 2986 (FTW) card.  This week we did receive a few of the FTW cards back that were RMA'd so we could test them and see if we can find a root cause of the problems.  Most of the cards we have received we cannot force a failure on (barring ridiculous overclocks/ turning off fans/nonsense like that) including the OP's card, we did however get a few with the problems described.  Currently we have several cards that we are sending to engineering, that we had received back from RMA citing this issue and had been tested here already, this is a mix of cards that both passed and failed our testing to provide as much info as possible for narrowing down the route cause.
 
This community has largely functioned as well as it has because (with a few feisty exceptions) this community REALLY tries to help each other.  So on that note I will make a request.  Any of you who do an RMA for the 980FTW having issues EVERYWHERE (NOT if it is ONLY Elite, Hero's of the Storm, or The Repopulation (not familiar with Repopulation before this thread) and it works great at stock clocks or even overclocked EVERYWHERE else but these couple games, then this will be a game and/or driver issue) if the cards are NOT downclocked, please follow up with me, drop me a PM.  Let me know the RMA number and a FULL system spec, hardware and OS.  I frankly have no interest in pointing fingers at any company, it is not my nature, BUT, with an issue like this that we are having a hard time reproducing here the more information we can get, the better.  If you want to put your spec here, that is fine, but if the info can be PM'd to me, or even emailed that would be a big help.  The more information we can get, the better.

Below is my hardware.  I also PM'd you with this info and the RMA #s.
 
I RMA'd 2 cards.  The 1st FTW card had problems out the box - constant and immediate crashes.  I spoke with tech after reading in the Nvida forum that it was a BIOS issue (I had the bad BIOS) and EVGA sent me a BIOS, but that did not fix the problem and was told to RMA the card.  I reinstalled the 2x 680 and it hummed along without a hiccup.  When I spoke with Tech about the 1st card I was told that the card did not past the test.  I recently RMA'd the 2nd card as it started down the same road as the 1st card.  I had to down clock the card by 100mhz to play my games - BF3, BF4 and FC4.  I ran everything at stock settings to eliminate any OC issues.  I hope the company shares its findings.
 
Case: HAF 932 Advance
CPU: i7-2600K (stock)
Memory: Gskill Trident (1866 @ 32gb)
MB: Asus P8Z68-V Pro
GPU: EVGA GTX 980 FTW
PSU: Corsair AX750 Gold 80+
Monitors: U3011, U2412

The only reason why the "4k method" makes sense is it because it's making the card use all the power it can to render the images necessary to obtain a certain FPS rating, and it also causes the card to push the power, and locking it at that power mode, preventing the power required to operate from fluctuating up and down, as soon as you get a power spike (a drop in power), it causes the card to hang. (think of it as Vdroop in the bios from Intel) as soon as you drop to 1080p or so, I'm assuming that the card is no longer requiring alot of power and that is when the power spikes start to occure, because there are only certain areas on 1080p that require certain power, especially with an FTW, obviously. THen (another assumption) before the GPU can underclock itself, the power already decreased "too much", causing the GPU to hang due to the driver no longer responding, giving you a 116 error code. What happens if you try to enable "maximum power" in the 3d settings?
 
So, again, it's a power issue for the FTW, and apparently a few guys figured out that the power is fluctuating up and down. That makes sense.
 
Obviously that looked like a blob of text, but i hope it makes a little sense to you guys. If there was an "Nvidia Vdroop" it would most definately help the FTW.
 
Would this help the FTW at all?? http://www.evga.com/Products/Product.aspx?pn=100-MB-PB01-BR

Jshiver90

stalinx20
The only reason why the "4k method" makes sense is it because it's making the card use all the power it can to render the images necessary to obtain a certain FPS rating, and it also causes the card to push the power, and locking it at that power mode, preventing the power required to operate from fluctuating up and down, as soon as you get a power spike (a drop in power), it causes the card to hang. (think of it as Vdroop in the bios from Intel) as soon as you drop to 1080p or so, I'm assuming that the card is no longer requiring alot of power and that is when the power spikes start to occure, because there are only certain areas on 1080p that require certain power, especially with an FTW, obviously. THen (another assumption) before the GPU can underclock itself, the power already decreased "too much", causing the GPU to hang due to the driver no longer responding, giving you a 116 error code. What happens if you try to enable "maximum power" in the 3d settings?
 
So, again, it's a power issue for the FTW, and apparently a few guys figured out that the power is fluctuating up and down. That makes sense.
 
Obviously that looked like a blob of text, but i hope it makes a little sense to you guys. If there was an "Nvidia Vdroop" it would most definately help the FTW.
 
Would this help the FTW at all?? http://www.evga.com/Products/Product.aspx?pn=100-MB-PB01-BR

I've tried the "maximum power" setting, and it just resulted in more frequent black screens; usually immediately after starting an application that utilized the GPU.

Yes, like I said in this post http://forums.evga.com/FindPost/2351842 the problem is not the core clock it's the fluctuating power draw.
Enabling "Maximum Performance" or "K-Boost" will lock your core clock while you are in 3D or in Boost mode but the card will still be unstable depending on the power draw that the game requires. If it fluctuates to much to fast everything crashes.
 
Now I wonder where is EVGATech_DaveB and if there are any good news with a possible solution coming from engineering that he sent those cards to.
Or is this the last we will hear from EVGA and there is nothing to be done here?

I'm wondering if that pcie stabilizer that I posted would do anything?

The GTX 980 FTW cards are from what I've seen the least reliable GTX 980 variant eVGA makes.

Anybody that has these crashes, they won't occur without warning and especially with windows 7 you will find an 'item' in the event viewer explaining why the issue came about.
 
David

I wish I hadn't early adopted these 980FTW. My step up window expired right before the 980Ti came out. Now I'm stuck with 3 of these 980 FTWs and I get crashes anytime I try to use NVIDIA Surround on 5760x1080 for more than 10 minutes in any game. Frankly that was the whole point on buying these cards. I would give anything to be able to Step-Up and be rid of these cards. First product from EVGA that has ever given me Buyer's Remorse. :(

VGAMASTER198
The GTX 980 FTW cards are from what I've seen the least reliable GTX 980 variant eVGA makes.

the 980 Hydro Copper is around the same league sitting on Refrence PCB (i guess) clocked up as high as the classified cards but on 2x6pin as a water cooled card, have RMAed 2 cards already sitting on my third and have driver crash galore on drivers 350.xx and up sitting on 347.88 atm and for now it seems very stable no crashes yet
 
Intel I7-4790k
Asus Maximus Hero VII
Kingston FuryX 16GB 1600MHz
Corsair HX750W
EVGA GTX980 Hydro Copper
 
been reinstalling my comp a few times driver changes back and forth, even though EVGA support says the 2 cards i sent in has passed their testing it fails my gaming since those cards where stable in 3dmark but not in real gaming atleast for me :P