Sign in with
Sign up | Sign in
Your question

I may have killed my 680...

Last response: in Graphics & Displays
Share
April 19, 2012 1:03:19 PM

Got a Galaxy 680. I have been playing around with Precision X and decided to try a 3dmark run at +125/+400. I had ran +120/400 earlier and that ran fine. Anyways, I load 3dmark, go to the bathroom and as I'm coming back the monitoring is turning off because it detects no video source. I'm certain it did not overheat because I don't think 3dmark had even loaded the first test yet. This happened within about 20 seconds of applying the overclock and launching 3dmark. Now, all startup attempts cannot get a video signal. The monitor searches but cannot find a source. The card is getting power because the fan is running but I have no display. Since when does a card completely crash upon a small overclock without a warning sign, and then be unrecoverable?

Any suggestions guys?
If not I guess I'll just have to RMA it.

More about : killed 680

April 19, 2012 1:48:17 PM

Unseat the card, insert your old card. Does your system boot with display now? If so, remove old card and re-seat the new card..if you still get no video signal you're probably hosed and time to start the RMA process.
m
0
l
a b U Graphics card
April 19, 2012 2:00:41 PM

yes probably bad overclock toasted something, safe way is RMA.
m
0
l
Related resources
a b U Graphics card
April 19, 2012 2:04:20 PM

Yea, or..... he could some how reset the overclock with the old card, heck even delete Precision X and then re-install it clean. Deleting the overclock. then his card should be back at stock...


Or... He could just RMA... that works to.. just trying to keep him from loosing his card for possibly long time
m
0
l
April 19, 2012 2:08:58 PM

That doesn't sound good... IMO never a good idea to leave your computer when you are testing OC settings. Just sit there with work to do or a laptop open or something to read, so that you can keep an eye on it.

I'm not sure exactly how GPU Boost and dynamic overclocking applies the voltages needed to supply extra power for the OC... my guess would be that something went wrong with the VRMs/voltages, not the temps.
m
0
l
April 19, 2012 2:16:35 PM

Yeah I realize that you should sit with the PC but like I stated, I had ran those clocks before without a problem so I guess I was just comfortable with it.

Anyways, I don't have another card to test with.

If I RMA to Newegg what exactly will happen considering they don't have any instock? Do they have like a back reserve they use for things like this or will I just have to wait?
m
0
l
April 19, 2012 2:19:35 PM

You'll just have to wait :( 

I hear ya about the small increase... I've suffered the same problem in multiple ways (test something for 6 hours no problems, go to bed and wake up with a dead PC, test something for an hour no problems and leave to do something for 5 minutes come back to a dead PC... etc).

My current philosophy is to never run OC tests without being present at the computer, no matter how stable I *think* it is.
m
0
l
a b U Graphics card
April 19, 2012 2:20:57 PM

jta98z said:
Yeah I realize that you should sit with the PC but like I stated, I had ran those clocks before without a problem so I guess I was just comfortable with it.

Anyways, I don't have another card to test with.

If I RMA to Newegg what exactly will happen considering they don't have any instock? Do they have like a back reserve they use for things like this or will I just have to wait?


I'm not sure how newegg works but probably they will give you the option, wait for the gpu to come in stock, or money back?

anyways yes, you should reset your clocks to default before returning it, the warranty is avoided by failure due to bad overclock, i'm not sure how newegg will react if they notice that the gpu toasted due to overclock...
m
0
l
April 19, 2012 2:25:44 PM

ricardois said:
I'm not sure how newegg works but probably they will give you the option, wait for the gpu to come in stock, or money back?

anyways yes, you should reset your clocks to default before returning it, the warranty is avoided by failure due to bad overclock, i'm not sure how newegg will react if they notice that the gpu toasted due to overclock...


Considering I can't boot the card, I'm not sure how I could reset the clocks. Besides, I'm fairly certain that software overclocks don't permanently change the timings on the card. If the software is missing, the card is running at its stock levels. I didn't have overclocks set to run on startup either. So if the card can be booted, it would still show the stock clocks.
m
0
l
a b U Graphics card
April 19, 2012 2:30:46 PM

jta98z said:
Considering I can't boot the card, I'm not sure how I could reset the clocks. Besides, I'm fairly certain that software overclocks don't permanently change the timings on the card. If the software is missing, the card is running at its stock levels. I didn't have overclocks set to run on startup either. So if the card can be booted, it would still show the stock clocks.

Well by resetting the clock i mean your software clocks to check if it will work, maybe it is not working because the clocks are being forced by software and it fails to work right away, just startup with another gpu on iboard if you have, and reset any software forced clocks...
and if the card is dead by overclocking there are signals that could be deducted by the the specific components that toasted, that is what i mean by they noticing that overclock caused it.
m
0
l
April 19, 2012 2:40:24 PM

the 680s run a bit warm... you prolly popped something in the VRM array.

m
0
l
April 20, 2012 3:24:06 AM

Well. Looks like I will be RMAing it. Tried clearing CMOS, reseating card, different PCI slot, different video port all to no avail. This kind of pisses me off. I realize I am the one who did it but I mean come on? When overclocks aren't stable you will simply get lockups and artifacts, and driver crashes. Since when does a 3rd overclock attempt completely brick a card? And it wasn't even a large OC. +125/400. Completely within what others are getting from these cards.
m
0
l
April 20, 2012 3:32:57 AM

And this is pretty much why I don't think the GTX 680 is a good card for people who like to overclock...

Also, overclocking on reference GPUs is always a bit risky. Sorry you gotta RMA... hope newegg gets some stock they can send you soon.
m
0
l
a b U Graphics card
April 20, 2012 11:33:33 AM

That is probably because the reference card is already at very high clocks for its cooling also new architecture. maybe this kepler is great and all but not so safe for overclocking? maybe you had a defective card that couldn't handle the overclock properly? maybe a software issue during the overclock caused some kind of overvoltage, and other possibilities.

No reference card is good for people who like to overclock, it is not safe unless you change the cooling solution.

This could not be the real problem, but it sure looks like it.
m
0
l
a b U Graphics card
April 20, 2012 12:16:14 PM

Well, if it doesn't even show the motherboard post screens (the stuff before windows loads) then it is not a software problem (as no software has loaded yet). If this is the case, some piece of hardware died (although the GPU is likely we can't prove it without another GPU to test that the rest of the system is ok). By the way, what are the rest of your system specs? If you have something like a 2500k, it has onboard graphics you can try to boot with.
m
0
l
April 20, 2012 12:50:47 PM

Currently Newegg have none of the GTX680 in stock. If you RMA your card, they'll charge you 15% restocking fee or you'll have to wait until 1 comes in stock. As those guys had suggested above, you should try to boot your PC with another card, any card and reconfigure your system. Make sure it's not anything else but your card that caused the problem before you RMA it.
m
0
l
April 20, 2012 2:52:46 PM

Yeah I guess I'll just wait. I'm not paying a 15% restock fee for a card that was a lemon.


Btw, I'm running a 2500k @ 4.2, 8gigs Corsair RAM, P8P67 Pro, and a Corsair HX850. The system had been running perfectly stable for 6 months with a 5970 so I highly doubt it is anything else. I highly doubt that something stable for 6 months would choose to go at the instant I load a graphics program with a card I just overclocked that has only been in my system 2 days.
m
0
l
April 20, 2012 3:46:39 PM

jta98z said:
Well. Looks like I will be RMAing it. Tried clearing CMOS, reseating card, different PCI slot, different video port all to no avail. This kind of pisses me off. I realize I am the one who did it but I mean come on? When overclocks aren't stable you will simply get lockups and artifacts, and driver crashes. Since when does a 3rd overclock attempt completely brick a card? And it wasn't even a large OC. +125/400. Completely within what others are getting from these cards.


I run mine at +175/475 :o  maybe i should stop that! Mine is an EVGA card though... this post has got me scared now. But with a three year warranty on the card right, should I be :heink:  :sweat: 
m
0
l
April 20, 2012 3:50:08 PM

Sonny73N said:
Currently Newegg have none of the GTX680 in stock. If you RMA your card, they'll charge you 15% restocking fee or you'll have to wait until 1 comes in stock. As those guys had suggested above, you should try to boot your PC with another card, any card and reconfigure your system. Make sure it's not anything else but your card that caused the problem before you RMA it.



No you do NOT have to pay a 15% restocking fee if the card is not working, that is only if you decided you dont want it. You have to do it within 30 days, I have done RMAs with newegg for three GPUs in about a two to three month period. Newegg has good customer service
m
0
l
April 20, 2012 3:51:40 PM

jta98z said:
Yeah I guess I'll just wait. I'm not paying a 15% restock fee for a card that was a lemon.


Btw, I'm running a 2500k @ 4.2, 8gigs Corsair RAM, P8P67 Pro, and a Corsair HX850. The system had been running perfectly stable for 6 months with a 5970 so I highly doubt it is anything else. I highly doubt that something stable for 6 months would choose to go at the instant I load a graphics program with a card I just overclocked that has only been in my system 2 days.


Call newegg you do not have to pay a 15% restovking fee, at least I did not have to.
m
0
l
April 20, 2012 3:56:18 PM

IMO everyone running dynamic overclocks on reference PCBs are asking for trouble... I wouldn't be surprised if bricked 680s start showing up all over from people who OC'd reference cards in the next year or so.

I'm just pretty pessimistic about the technology... I don't understand the point of it and think that it has more potential to screw things up than anything else. GPUs have scaled back their clocks for a long time to save power (my 7970 is idling at 300/150 as I write this...) and I don't see dynamic OC'ing actually increasing performance over regular OC'ing. I see it as a technology that's going to blow a lot of VRMs, make a good marketing slogan, and not do much else.

I love Nvidia but I dunno if I can ever see myself buying a card that can't be clocked and OC'd normally.

Also I agree about the restocking fee - I don't think you have to pay that if the card is dead.
m
0
l
April 20, 2012 4:10:33 PM

I just don't see why you'd need to overclock the 680. In fact I've never thought that overclocking video cards was a good idea. The % increase is way too small for the risk...
m
0
l
April 20, 2012 4:12:09 PM

^^ I have a 30% increase on my 7970 vs a reference card... I'd say that's definitely worth the risk!

(Although since I now have a Lightning model, there's much less risk involved as well)

It really just depends on the card. Some GPUs are fine with overclocking and some aren't. My opinion is that the 680 falls into the latter category but like I said I'm probably a bit pessimistic about it compared to most.
m
0
l
a b U Graphics card
April 20, 2012 4:14:02 PM

yes the point about those 680 is the GPU boost they use, they AUTO OVERCLOCK it and Downclock it, depending on the game, etc, maybe that overclock + this gpu boost caused instability on overclock.

who don't know what it is:
GPU Boost intelligently monitors graphics work load and increases the clock speed whenever possible.
more? http://www.geforce.com/whats-new/articles/introducing-t...

GPU Boost doesn't take away from overclocking. In fact, with GPU Boost, you now have more than one way to overclock your GPU. You can still increase the base clock just like before and the boost clock will increase correspondingly. Alternatively, you can increase the power target. This is most useful for games that are consuming near 100% of this power target.
m
0
l
April 20, 2012 4:23:01 PM

I called EVGA and asked if bumping up the specs in the precision x control panel was OCing, they said no because the card was meant to operate within the +132 power range. So doing this does not affect your warranty and is not really overclocking. I mean no one has bumped the voltage yet, you can't, when you can on non reference cards with more VRMs and unlocked voltage controls it will be amazing!
m
0
l
April 20, 2012 4:26:10 PM

So I don't know why people seem to think the 7970's OC better than a 680. The 680s cannot be over clocked!!! wait until the 6+8 pins or dual 8 pins with more VRMs and unlocked voltage, then compare
m
0
l
April 20, 2012 4:31:41 PM

People think the 7970s OC better than the 680 because ~30% overclocks are standard on the 7970 (including reference models) but only ~15% OCs are standard on the 680...
m
0
l
April 20, 2012 4:40:19 PM

YOU CANNOT OVERCLOCK A 680, its locked you cannot up the voltage beyond what Nvidia has done yet. If you check out EVGAs in house OC guru he added more VRMs and unlocked the voltage control (works for EVGA so EVGA unlocked 4 him) and guess what creamed a 7970.
m
0
l
April 20, 2012 4:44:51 PM

Sounds to me like somebody just likes green labels on the GPU...

Here's a fact:
On average, a reference 7970 overclocks MORE than a reference GTX 680, which by the way does overclock. Overclocking =/= overvolting, those are two separate things.

You wondered why people say a 7970 overclocks better, well that is why. Sorry you seem to think that's an invalid reason.
m
0
l
April 20, 2012 4:49:17 PM

You are limited right now by software, so many things are locked on these reference 680s. People usually bump up there voltages on 7970s to get higher clocks that is why it is not really overclocking. Apples to pomegranates.
m
0
l
April 20, 2012 4:53:36 PM

You realize that this topic is WHY things are limited, right? Because reference cards will blow up if not. Reference 7970s hit 1125-1150 core clock (22%-24% OC) with no overvolting. So, even then, the reference 7970 overclocks better than the reference 680.

To be clear, just because a card doesn't overvolt, does NOT mean that it doesn't overclock.

If you later on want to compare custom 680s to custom 7970s, that's fine, but it's a different discussion and it's one that isn't possible right now since we're still awaiting custom PCB 680s (and many of the custom PCB 7970s). It's also a discussion that has no bearing at all on which reference card overclocks better.
m
0
l
a c 125 U Graphics card
April 20, 2012 5:05:29 PM

This is why I suggest running GPUZ in the sensors tab. It shows a lot more than just core temp - people often don't realize just how hot VRM and VRAM can get, especially if they don't have any sort of heat sink on them.
m
0
l
a c 125 U Graphics card
April 20, 2012 5:08:30 PM

MANOFKRYPTONAK said:
People usually bump up there voltages on 7970s to get higher clocks that is why it is not really overclocking. Apples to pomegranates.


What? Increasing the clocks = overclocking. That is literally the definition. The voltage increase can of course increase the overclocking potential, but if 680s die from OCing without a voltage bump, how do you expect to gain a higher clock with more voltage?

Did you ever see the videos of GTX 590's literally exploding with overclocks? Their VRMs go POP and sparks and smoke explodes out.
m
0
l
a b U Graphics card
April 20, 2012 5:17:34 PM

yes, probably nvidia don't allow overvoltage because the cards are already on the maximum safe voltages. unlike the 7970 that aren't, i think that they should have improved 7970 to the same level as 680, so they would have better stock performance, but then just like the 680 would not be able to overvoltage with safety.
m
0
l
May 23, 2012 5:23:02 PM

This thread is probably long dead but all you have to do is boot into safemode and you should be able to go in and uninstall the overclocking tool then reboot. That is if you are getting a video signal into at least bios. If not your card is fried. Good luck trying to RMA through newegg. GPU RMA's are usually done through the manufacturer.
m
0
l
May 23, 2012 5:47:20 PM

After what happened to my gtx 570 I wouldn't try overclocking anything from nvidia for awhile. At least something with a reference design. My 570 died @ 1.05v on gpu. I hadn't even got into the game yet and VRM fet was fried. I replaced it with a 7950, which comes with 6 phases, runs at lower power and can overclock quite a lot without even messing with gpu voltage.
m
0
l
!