Faulty GTX 680?

[dandare2012]

OK, so (very) long story short,

I bought a Gigabyte GTX 680 OC 4gb at the beginning of December. I installed it and on the second benchmark run in 3d Mark vantage, the screen goes black, pc reboots and there is no video signal, forcing a hard restart.

Due to a somewhat long story, I now own two PC's my old one - 1366 EX58-UD5, i7 920, corsair dominator ram, and my new one, 1155 G1 Sniper 3, i5 3570K, Samsung Green Ram. PSU's are Corsair HX750 and AX850

The exact same fault is consistent on both rigs, although highly intermittent. As soon as I put my old GTX 480 back in the problem goes away and the system runs fine.

BUT both Gigabyte and the retailer have tested the 680 card and say it is not faulty.

I have tried to brand new PSU's, multiple fresh OS installs, driver installs, different drivers, different wall sockets, surge protectors, internal cables and nothing makes a difference.

I have video evidence of the fault bu the retailer will still not acknowledge it.

If anyone has any idea It would be greatly appreciated.

 

[Jakus]

Turn off automatically restart in windows first !

Sounds like you have 2 unstable PC's , try prime 95 stress test and see what happens.
PC's crash totally when the CPU gives up not when a GFX card stops responding, seems a little strange you have 2 bad setups...you are consistent

Sure you should be ok after you have turned a few things down.

 

[dandare2012]

Turn off automatically restart in windows first !

Sounds like you have 2 unstable PC's , try prime 95 stress test and see what happens.
PC's crash totally when the CPU gives up not when a GFX card stops responding, seems a little strange you have 2 bad setups...you are consistent

Sure you should be ok after you have turned a few things down.

Thanks for the reply, I had thought of it being an OC issue, but the first thing I did was reset all back to defaults before continuing testing. Also worth noting that PC 1 worked fine for 3.4 years without issue at the same settings, and for 2 years with the GTX 480 in it.

Also the new pc is has been running fine for a month with the 480 in it. The issue only arises when the 680 is installed, that is what is so confusing.

Event manager lists it a an event 41 Kernel power, and sometimes as a video hardware error, but both only give generic codes that do not help to bug check.

 

[Gregster]

Sounds to me like the retailer is pulling a fast one. Apart from trying various drivers, I would be demanding it to be returned.

 

[Jakus]

Thanks for the reply, I had thought of it being an OC issue, but the first thing I did was reset all back to defaults before continuing testing. Also worth noting that PC 1 worked fine for 3.4 years without issue at the same settings, and for 2 years with the GTX 480 in it.

Also the new pc is has been running fine for a month with the 480 in it. The issue only arises when the 680 is installed, that is what is so confusing.

Event manager lists it a an event 41 Kernel power, and sometimes as a video hardware error, but both only give generic codes that do not help to bug check.

You have to look for something that has enough privileges and deep enough to mess things up randomly, like a GFX, Audio driver or Antivirus, Hardware incompatibility I would think would be a far more regular failure for you.

Obviously you have given this issue some time and effort but you must forget what has gone before and even if it's reset to defaults still make sure the system will run prime95 for 1hr.

Testing the GFX card is easier and quicker because you can overclock it 100/100 or more and accelerate the failure by reducing stability margins during testing.

Until you know the exact stability margins of everything I would not look to a software problem or anywhere else, you will just go in circles .

I have just spent 3 weeks on a cold boot bluescreen issue and not yet got to the bottom of it because the PC has to be stone cold and after 2 mins everything returns to massive stability margins....swap yer

 

[Kaapstad]

Sounds to me like the retailer is pulling a fast one. Apart from trying various drivers, I would be demanding it to be returned.

+1

If anything is going to make a PC unstable its a GTX 480 with its higher power usage and heat output.

As gregster says the retailer appears to be pulling a fast one.

 

[dandare2012]

You have to look for something that has enough privileges and deep enough to mess things up randomly, like a GFX, Audio driver or Antivirus, Hardware incompatibility I would think would be a far more regular failure for you.

Obviously you have given this issue some time and effort but you must forget what has gone before and even if it's reset to defaults still make sure the system will run prime95 for 1hr.

Testing the GFX card is easier and quicker because you can overclock it 100/100 or more and accelerate the failure by reducing stability margins during testing.

Until you know the exact stability margins of everything I would not look to a software problem or anywhere else, you will just go in circles .

I have just spent 3 weeks on a cold boot bluescreen issue and not yet got to the bottom of it because the PC has to be stone cold and after 2 mins everything returns to massive stability margins....swap yer

+1

If anything is going to make a PC unstable its a GTX 480 with its higher power usage and heat output.

As gregster says the retailer appears to be pulling a fast one.

OK, so over the last month on the new rig I have run Prime 95 for multiple sessions 3 hours+ on all torture tests, + Intel Burn test on all options including Maximum stress 7000MB+ and also Super PI tests pass fine. That is at stock and with a 4.5Ghz OC. Max temps on GTX 680 do not exceed 70 degrees under furmark full hd test, so it is not a heat issue, as the 480 hits a whopping 94 degrees and stays stable.

I also tried reinstalling audio drivers and disabling audio devices too. Anti Virus is always off when testing, but you would expect to see errors with any card if it was going to cause them.

I am inclined to agree that the retailer is lying, but Gigabyte UK back them up, they have tested it separately and sent me test results showing that it ran for 15 hours+ without error. The retailer refuse to refund it as they say they cannot get their money back if they take it from me. I have argued that technically this is not my problem as part of the risk of being in retail is dealing with the occasional loss, and it should not be left to a consumer to protect them from that, but all they did was offer to sell the card as ex demo on their site and give me the proceeds.

The sole and only shared item between them now is the monitor, but as far as I know a monitor could not cause a pc to crash and reboot... it just receives signal and does not send it.

 

[Kaapstad]

Do you have access to any other PCs you could try the card in - Friends/Family/Work

 

[Jakus]

OK, so over the last month on the new rig I have run Prime 95 for multiple sessions 3 hours+ on all torture tests, + Intel Burn test on all options including Maximum stress 7000MB+ and also Super PI tests pass fine. That is at stock and with a 4.5Ghz OC. Max temps on GTX 680 do not exceed 70 degrees under furmark full hd test, so it is not a heat issue, as the 480 hits a whopping 94 degrees and stays stable.

I also tried reinstalling audio drivers and disabling audio devices too. Anti Virus is always off when testing, but you would expect to see errors with any card if it was going to cause them.

I am inclined to agree that the retailer is lying, but Gigabyte UK back them up, they have tested it separately and sent me test results showing that it ran for 15 hours+ without error. The retailer refuse to refund it as they say they cannot get their money back if they take it from me. I have argued that technically this is not my problem as part of the risk of being in retail is dealing with the occasional loss, and it should not be left to a consumer to protect them from that, but all they did was offer to sell the card as ex demo on their site and give me the proceeds.

The sole and only shared item between them now is the monitor, but as far as I know a monitor could not cause a pc to crash and reboot... it just receives signal and does not send it.

Why did you not say previously you had tested the basic PC's stability ?
More than a few of the factory OC cards have been on the edge with their std OC, normally that would not make the PC shut down/reboot though
Have you played with GFX card clocks, does it make things worse ? what's the max stable clocks etc ?

 

[Method0ne]

Sounds vaguely similar to an issue I have with my GTX690, in that if I have the drivers installed for the on-board graphics (hd4000) then windows will get to the point where it normally shows the desktop and will hard-lock the entire system.

Remove the on-board drivers, everything works fine, no problems whatsoever.

I don't know why this happens, but I can reproduce it at will.

 

[dandare2012]

Do you have access to any other PCs you could try the card in - Friends/Family/Work

No, sadly not. GF's PC is too pld to take this card, an ancient XP based pc. Work would never let me take a machine apart.

Why did you not say previously you had tested the basic PC's stability ?
More than a few of the factory OC cards have been on the edge with their std OC, normally that would not make the PC shut down/reboot though
Have you played with GFX card clocks, does it make things worse ? what's the max stable clocks etc ?

Sorry... but yes I have extensively tested stability. I just ran more tests last night and this morning, Intel Burn test, super pi and Prime all pass with no errors. No WHEA errors in event manager either.

I have no experience OC's GPU's, and also felt that if I did they may try and blame me for damaging it. As the fault occurred within less than 15 mins of the thing being in my pc, I had always assumed that I would have no issue getting a refund etc as it was, i my opinion DOA.

Running GPU-Z during bench marks show that it was not even at max clock or power when it crashed.

Sounds vaguely similar to an issue I have with my GTX690, in that if I have the drivers installed for the on-board graphics (hd4000) then windows will get to the point where it normally shows the desktop and will hard-lock the entire system.

Remove the on-board drivers, everything works fine, no problems whatsoever.

I don't know why this happens, but I can reproduce it at will.

I would say it was worth giving it a go, but my old pc did not have on-board graphics as far as I know, + mine might function normally for like 2-3 days, then suddenly error a bunch of times. Web browsing, looking at photos, benchmarking have all caused it to crash. I would also add that Gigabyte tested much more than the retailer, and they also put it in both of the same boards I own to do so.

 

[Jakus]

You have to get serious if you want to fix this.

Ghost your os or find another HD and do a clean install with just Intel drivers GFX and sound.

Crysis 2 or Heaven bench and use EVGA precision X to monitor and OC/underclock gfx card and find its clock stability limits before you go further !

It may not seem relevant to you and there are a million possible reasons for your problem but you have got nowhere so far, have you ?

Fault finding is about Assuming nothing and finding out what it is not the problem, eventually you will be left with hopefully just the villain !

Intermittent faults are a real barsteward but hopefully along the way things can be provoked to a more unstable state that highlights these faults more consistently, which obviously is a big help.

 

[dandare2012]

You have to get serious if you want to fix this.

Ghost your os or find another HD and do a clean install with just Intel drivers GFX and sound.

Crysis 2 or Heaven bench and use EVGA precision X to monitor and OC/underclock gfx card and find its clock stability limits before you go further !

It may not seem relevant to you and there are a million possible reasons for your problem but you have got nowhere so far, have you ?

Fault finding is about Assuming nothing and finding out what it is not the problem, eventually you will be left with hopefully just the villain !

Intermittent faults are a real barsteward but hopefully along the way things can be provoked to a more unstable state that highlights these faults more consistently, which obviously is a big help.

Hi,

Thanks for the advice. I felt I had got pretty serious already tbh.

The long story is that my rig worked perfectly for 3.5 years all on one os install. It had 2 gpu's in that time, a 295 and a 480.

I bought a new psu to replace my 5 year old one, as it was getting hot, along with a new ssd to replace my old hdd. after two weeks I decided wtf! I'll buy a new gpu too! I want to play far cry 3 and metro 2033, and the 480 would just not cut it.

Within 15 mins of it being in the error occurred when running 3d mark. I thought it was an app crash tbh as it then ran fine. Then 3 days later, and with a massive dose of irony, while I was viewing the images of the 480 to put it on ebay the error came back... kind of funny if you think about it! then it errored 3-4 more times that day, a mix of web browsing and viewing jpegs.

The next day I phoned the retailer (3 working days after purchase) and reported it as faulty. The guy on the phone was massively rude and clearly did not give a ****, they would not replace it without testing and would charge me if no fault was found, he then hung up on me.

I got in an argument with them because their site did not specify what the "service fee" would be if no fault was found so I refused to accept the terms of their return, based on the fact it could cost me an undetermined amount of money.

They passed me to GBT UK directly, who suggested I update my BIOS to the latest BETA and reinstall windows. No Fix.

I also tied putting the 480 back in and the machine was running fine.

At this time I had been considering buying a new mobo and cpu anyway, so I took the plunge and shelled out for a new mobo, cpu, ram, second SSD, case and cooler. Put the rig together, and as soon as I had the card and drivers in, bang, the error was back again. I then put the 480 back in and ran multiple tests with no errors. I also tried two different bios for the new board which seemed to make no difference.

I then sent the card to GBT UK for testing. They said no fault found after 15+ hours in different bench tests + a QC check on the card.

When card first came back for Gigabyte UK I did a fresh OS install with just intel chipset drivers, graphics and sound + bench and testing software. It still crashed, so I replaced the PSU, which at that time was only 3 weeks old, with another brand new one at a cost of £154.00. It seemed to run fine for a while (completed 8 hours solid in 3d mark 11) + 3 runs of furmark full hd. Then 3 days later when just browsing the web, it came back. I think it error ed 7 time on xmas eve during a variety of tasks.

I then got fed up, took it out, boxed it up and put back the 480, which has been running since xmas day with no problems, just like it had for the last two years+

I cannot currently test the 680 because it is with the retailer.

In reality I was simply wondering if any one had any clue as to what the issue might be, but as most said above it is more likely that the retailer is pulling a fast one.

I mean I have used it in two completely separate pc's, and the bug still happens. I refuse to believe that to pc's bought 3.5 years apart could both contain an identical bug in their hardware that would cause the card to crahs, but not another card, I mean it is almost craxy to suggest it.

The card is from the same manufacturer as the mobo too, so there should not be any comparability issues there.

I think so far I am up to 54 mails to the retailer (sends and replies) + around 30 to the manufacturer. £154.00 for a new PSU and £15.00 for shipping the card to them. + a shed load of hours spent emailing, bug testing, posting on various forums etc.

It is a real tricky one, I mean to me it is cut and dry - it errors with it in, and does not when it not in, on two pc's. They argue other wise.

 

[Kaapstad]

The next day I phoned the retailer (3 working days after purchase) and reported it as faulty. The guy on the phone was massively rude and clearly did not give a ****, they would not replace it without testing and would charge me if no fault was found, he then hung up on me.

Surely DSR means you can just ask for a refund.

 

[Jakus]

Hi,

Thanks for the advice. I felt I had got pretty serious already tbh.

Snip..

It is a real tricky one, I mean to me it is cut and dry - it errors with it in, and does not when it not in, on two pc's. They argue other wise.

You do have to be careful where you buy from, I can totally see the retailers frustration as so much fully working kit does get returned, but from what you say they have not been very professional at all !

The only people you should let talk to the retailer is your credit card company You don't come across as someone who has not made a reasonable effort to overcome issues, It has not worked out, let your credit card company get a refund.

If you choose to pursue this I would be dealing only directly with Gigabyte !

The more things unfold the more I smell BS, Please post that report from Gigabyte.

 

[dandare2012]

Surely DSR means you can just ask for a refund.

DSR laws are really hazy.... they are actually open to exploitation from either party;

They specify in their T&C that the item must not be opened or installed, which is kind of legal and kind of not. Technically if you send them a written notification of cancellation under DSR's (within 7 working days, from the day after the day you got the goods); by the letter of the law They MUST refund you within 30 days, including any postage paid, and cannot insist on the goods being returned before hand, or on them being unopened. They can however legally attempt to recover funds if the goods turn out to be damaged after they receiver them.

However, DSR's only cover you to be able to "examine" the good the same as you would if you were in a shop. If you were in a shop they would not let you open the box of a new product, or break a seal, as they would generally have "demo" unit to show you - so they use this as the basis for their argument.

We get this where i work all the time, a customer opens an item, ruins the packaging, or marks the item and sends it back under DSR's - if we want to fight them then it means going to court, which is simply not worth it, and added to that it means they leave bad reviews. This is why most companies let it slide if you send opened stuff back.

My issue is that I was honest and reported it as faulty; DSR's cease to apply if you state the item is faulty, it then moves over to Sale of Goods act 1979, under which I am entitled to reject the goods on the basis of quality. The burden of proof legally lies with me in this case, and the retailer argues that the videos I have of the crash occurring only constitute proof that the two systems are faulty, not the card. They refuse to accept my proof, and refuse to refund.

This leaves me only able to either pursue it through my credit card company under the consumer credit act of 1974 or take the retailer to court. My credit card company have said that I would need concrete proof and also I have been advised by citizens advice that I would need concrete proof for the courts, and whether my proof was enough would not be decided until it had already gone to court and cost me at least £85.00 to open the case, which may also take months.

The lesson here is do not buy any goods from Lambdatek Ltd! they are fraudsters. Also when you buy pc components, open them carefully and test them a lot in the first 7 days. If they turn out to be faulty, don't tell the retailer and send them back under DSR's! - problem is that this is a sh**** thing to do as it means some other poor schmuck will end up with the goods, but if retailers are going to treat consumers in this way then I see no other option.

Thus far being honest has cost me a great chunk of six weeks of my free time, £160.00 extra that I did not need to spend, and I am still without the £430.00 card I paid for to begin with.

You do have to be careful where you buy from, I can totally see the retailers frustration as so much fully working kit does get returned, but from what you say they have not been very professional at all !

The only people you should let talk to the retailer is your credit card company You don't come across as someone who has not made a reasonable effort to overcome issues, It has not worked out, let your credit card company get a refund.

If you choose to pursue this I would be dealing only directly with Gigabyte !

The more things unfold the more I smell BS, Please post that report from Gigabyte.

The report was just a series of images showing the benchmarks.

The only "conclusive" proof is a screen grab of MSI combustor showing a run time of 15 hours, and two completed bench scores from 3d Marks 11 on two different mobo's

Thing is I feel I should be able to trust GBT UK, they have been very helpful and polite, and I arranged all the testing with them myself outside from the retailers influence. I mean, I tested the card for over 8 hours without getting an error, so I guess it was just (bad)luck? - thing is they have no legal requirement to do anything. My contract is with the retailer, not them...

The retailers report was even worse though. They did not even send any screen grabs to show that the card was running!, all I got was a Metro 2033 benchmark report and a gpu z log file that showed "a card" running for one hour. There are no details to show it was my card, I mean these could belong to anyone, and are not really proof at all.

 

[eyetrip]

I recently had a run in with them too. But purchased monitor with faulty pixels centre screen.

They would not accept as fault and didn't really want to let me return under dsr!

Eventually got it sorted but lesson learned. Certainly never buy from those sharks, and always, always stick to my 4 usual staplers of which ocuk in top two.

 

[dandare2012]

I recently had a run in with them too. But purchased monitor with faulty pixels centre screen.

They would not accept as fault and didn't really want to let me return under dsr!

Eventually got it sorted but lesson learned. Certainly never buy from those sharks, and always, always stick to my 4 usual staplers of which ocuk in top two.

Can I ask what you did to get it sorted?

I too usually use OCUK or others, but I wanted the 4gb version of the card, and only a few places do it. For my sins Lamdatek were cheaper by £30.00, which I am now regretting heavily.

 

[modez]

Delete.

 

[eyetrip]

Can I ask what you did to get it sorted?

I too usually use OCUK or others, but I wanted the 4gb version of the card, and only a few places do it. For my sins Lamdatek were cheaper by £30.00, which I am now regretting heavily.

I was just forceful, and wouldn't take no for an answer.

They try their best to worm out of DSR's, and the number times they replied stating item would be tested to confirm fault and returned if none found was annoying. Constantly had to keep stating was DSR.

And then, with most places I have dealt with, as long as you get in touch with regards to DSR, they typically give you longer to actually return the item. These guys wanted the item back within the 7 days.

They were just unfriendly and horrible throughout the experience, and I will never, ever buy from them again, even if they were the only people to stock something I wanted.