Faulty GTX 680 or somthing else?

dandare

Honorable
Dec 9, 2012
69
0
10,630
Hi,

Bought a new GTX 680 last week and since installing it I have been experiencing random black screen crashes. Screen goes black and pc re-boots, super slowly. does all the post checks on mobo digi display at ultra slow speed, then reaches "FF" which means successful boot on a gigabyte board. - Still no video signal. have to hard reset.

First time was on my second bench test in 3dmarks vantage, just after installing the card. Then it did not do it for 3 days. Sunday it crashed 4 times in a row when just simply opening small jpeg files.

Critical error listed in Windows event manager is: Source: Kernel Power, Event ID: 41, Task Category: 63

Power supply is a new Corsair HX750 which worked for 2-3 weeks powering my GTX 480 under heavy load. No issues.

I cannot get this error to repeat when I try to. I have run multiple bench marks and cannot get it to error.

Any suggestions as to what this could be would be appreciated.
 

dandare

Honorable
Dec 9, 2012
69
0
10,630
This is What I thought, but any ideas on how the company would test for this. Do you have a specific reason for thinking this... anything you know will help my case if I get in a fight with the retailer.

I bought it from Lambdatek in the UK and they have told me to return it, but were hugely rude and unhelpful about it. They say it could take 5 working days just for them to get round to testing it, and they may have to send it back to gigabyte. this would mean not getting it back before xmas...

They also said that if they cannot prove the fault I will be charged a service fee and have it returned to me. Based on the totally intermittent nature of the fault I am really S**ting it that they will say this and just send me back a dud card after 3-4 weeks.
 
Personally I would rather send it back to the manufacturer directly instead of going through the seller. Unless the seller is the UK version of Newegg then most times its just more hassle than its worth to send them a partially working product if they test.

Those retail stores mostly only test if a machine boots with it in the system and thats basically it. (if they test at all) (At least that was what my experience was with them in the US)

What are your temps? Is there any artifacting going on before these incidents occur?

 

dandare

Honorable
Dec 9, 2012
69
0
10,630


Thanks for the reply. Nothing unusual. runs the bench amrks fine. This card is otherwise awesome. runs at 25 degrees idle compared to 45-50 on the gtx 480. after running 3d mark 11 performance run (from free version) temps never exceed 50 degrees when monitored with GPU-Z...

the only clue I have is that kernel power report from the windows log, and to be honest I am not sure what it means, doing some research now.

I figured exactly what you said about the retailer. I do not see how they can properly test it. I have tried to talk to them by phone and email, but all I get is one liners about sending it back for testing and filling out an online RMA form. They will not communicate with me at all on a personal basis. I pretty much assumed that all they would do is plug it in and See if it runs, maybe a bench mark test at the most. what else can they do? they are not a card manufacture so they are hardly qualified to pull the thing to pieces... and that is kind of what worries me, they wont tell me how they will test it.

I already filled out an tech support request with gigabyte direct, and also sent them an email on another contact address. I really hope to hear form them tomorrow.
 

boju

Titan
Ambassador
have you tried turning gpu Phase Mode in the bios from auto (default) to off?

There could be problems regulating power from the motherboard with the new gpu. Well that's what i believe kernel power error 41 means.

Hope it helps.
 

dandare

Honorable
Dec 9, 2012
69
0
10,630


Can I ask what this would do? I have not altered anything in the bios since installing the card...

I found the following info which both sounds interesting...

one suggests that it is a power issue related to the socket in the house. Now there could be an element of truth here because we have been having intermittent problems with the power here, but I am unsure as it did not happen: http://www.tomshardware.com/forum/46250-63-random-restarts-critical-kernel-power-event-task

the other suggests that it is a conflicting audio driver issue, which seemed unlikely until I looked into my deceive manager and found I had two drivers listed exactly as this guy is saying... : http://www.techsupportforum.com/forums/f299/solved-kernel-power-41-63-error-434613.html
 

boju

Titan
Ambassador
G.P.U stands for Green Power Utility. It's supposed to manage power to the CPU so that it's not wasting electricity when the extra power isn't needed. Turning this off solves the kernel 41 error.

If you have that option in the bios, try turning it off.
 

dandare

Honorable
Dec 9, 2012
69
0
10,630
I cannot find that option... I have Gigabyte board, which is the Award Bios.

Any clue as to where it is located, I did look under everysetting and googled it and cannot find the answer
 

dandare

Honorable
Dec 9, 2012
69
0
10,630


Gigabyte GA-EX58-UD5 (rev 1.0)

I am not at home right now so cannot check the bios for anther 5hours or so.
 

boju

Titan
Ambassador
further investigating, seems gigabyte's des is software based by the looks of it.

http://www.youtube.com/watch?v=AIPqdbjAjzw

Reading on it, enabling C1E and CPU EIST in the bios is required to run des and perform it's power managing function.

Could try disabling those two and test for a bit or uninstalling des it if it was ever installed, seems to cause more problems then it solves be reading some of the comments from people around the net.


If its not a power issue then it could be sound chip related conflict with duplicated sound drivers. See if there are multiple sets of the same sound drivers in device manager left behind from the 480. Uninstalling all sound drivers and beginning again may help.
 

dandare

Honorable
Dec 9, 2012
69
0
10,630


Thanks for spending the time on this, really apreciate it!

The des software has not been insatlled (never heard of it) - this is a relatively fresh install of windows in a new SSD.

I do oddly have two high definition audio drivers listed in device manager... this board has on board Realtek sound. Device manger lists the two drivers as loction (0) nd locatio (2), both s onboard. One is listed as just "High Definiton Audio" anf the other as Realtek High Dfeinition Audio.

I never install Nvidia audio drivers, but I am now thinking that maybe I did not un-check for the nvidia high def aduio the box when installing the gtx 680 drivers... I have no other idea where theycould have come from.

When I get home I will unintall all drivers and see if the entry goes...

EDIT** I uninstalled the Realtek Audio drivers and still two entries remain under device manger

1. Location 0 (Internal High Definition Audio Bus)
2. Location 2 (Internal High Definition Audio Bus)

Reinstalling the Realtek driver only overwrites entry two.

I also found the following log in the windows reliability monitor:

A problem with your video hardware caused Windows to stop working correctly.

Problem signature
Problem Event Name: LiveKernelEvent
OS Version: 6.1.7601.2.1.0.768.3
Locale ID: 2057

Files that help describe the problem
WD-20121205-2353.dmp
sysdata.xml
WERInternalMetadata.xml

View a temporary copy of these files
Warning: If a virus or other security threat caused the problem, opening a copy of the files could harm your computer.

Extra information about the problem
BCCode: 117
BCP1: FFFFFA800A127010
BCP2: FFFFF8800F1A7A88
BCP3: 0000000000000000
BCP4: 0000000000000000
OS Version: 6_1_7601
Service Pack: 1_0
Product: 768_1



 

boju

Titan
Ambassador
Running out of ideas to try avoid you rma'ing and risking the chance it happening again if its something else, but at this point its leaning towards a gpu fault.

Only other thing i could suggest which you may have done already is seeing if the computer remains stable when reinserting back your 480 if you still have it and make sure nothing was bumped during the replacement with the 680.

If all is well then try uninstall the gpu drivers using driver sweeper, reboot then afterwards shutdown. Install the 680 on a clean slate to iron out any in depth driver conflicts.

And lastly have windows set for performance power and same with in nvidia control panel power management mode.

I'm leaning towards a faulty card myself at this point.

 

dandare

Honorable
Dec 9, 2012
69
0
10,630
Well it just did it again... aaaaah... first time since Sunday. This time all I did was open a dxdiag and it crashed.

Tonight I uninstalled the video drivers and reinstalled from fresh and it obviously made no odds.

I have not yet tried the 480 reinstall because I am waiting for gigabyte to come back to me and I am assuming they will ask me to run some test with the 680 installed. however it is the only way to be sure it is not the rest of the computer... so it may have to be done. Pain in the behind though as it has taken two days for it to randomly happen again, last time it was three.

Thanks for your other suggestions, sound advice that I shall probably follow.

I did speak to the retailer again today, and they basically said that if i send it to them they will stick it in a pc and run a continuous bench on it for a solid day to try and get it to fault. If it does not then I will be charged £15.00 GBP + shipping to have the card returned to me as is. If I wish they could also send it to Gigabyte, but that will cost me an additional £10.00 + shipping both ways.

Gigabyte tech support have still not picked up my ticket after 26 hours... there RMA email team replied to me in 4 hours, but they say they are not meant for public use, I am meant to go through my retailer. They only offer a testing and repair service and not a replacement or refund...


 

dandare

Honorable
Dec 9, 2012
69
0
10,630


I have not no...

what did happen last night is it did it twice again, same error. Interestingly, after it crashed when trying to run a dxdiag I rananother and it said that last time it crashed hen trying to access direct 3d... now I do not think it is a directx error... but maybe? orit could just be that thae fact it was trying to sent a signal to the card tocheck the hardware and th faulted...

I have bought a paid license for 3dmark11 (always used the free basic versions in the past) and am going to leave it on a 10x loop tonight wth GPU-Z running in the background... hoping that it willl do it and I can then record soem data from the time it crahes. I am also going to start leaving GPU-Z running a constant log int he bckground whnever the pc is on. that way I should get some data when it dies again.

Once that is done (should take an hour or two) and provided that it crashed at least once, I will put the 480 back in and do the same to see if I can get it to fault again.
 

dandare

Honorable
Dec 9, 2012
69
0
10,630


Kept doing it mate..

So I bought and built a new PC... :whistle: i was planning on upgrading everything anyway soon, so i thought WTF, at least i will know i is not the rest of the hardware.

Only parts tht remain the same are the Samsung SSD, which is now a Games drive; I bought second new one to be the system drive, and the PSU, which was new 3-4 weeks ago anyway.

So the story goes that I spent all weekend building this thing and then put in the new card, installed the drivers and on the second run of 3dMk11 - boom - does it again, same error.

So I run a few more tests and get it to error several more times, and it started artefacting as well.

Put in my GTX 480, and system ran fine.

STILL the retailer will not believe me... they have insisted that it go back to Gigabyte directly for testing.

I shipped it today. Will update with news...

Oh, what I will say is that so far Gigabyte have been really good. the only positive thing the retailer has done was give me the direct number for gigabyte UK. I got through to the same guy who picked up my online support ticket and we have spoken several times. He has offered more help and advice than the people I bought it from and he does not even have to, as i do not think this number is usually used by the public.
 
Yeahhh, do us all a favor and never buy from that retailer again. Even if you have to pay a customs fee. They don't seem all that nice. Usually gigabyte is good about customer service. Hence why people buy their stuff and not a no name brand like CrystalMeth - Power Supplies and co.

hah, well anywho I wish you the best of luck!
 

dandare

Honorable
Dec 9, 2012
69
0
10,630


Thanks mate...

The retailers name is Lambdatek. They operate in Europe only I think, so no worry for you. I bought from them rahter than my usual supplier as they stocked the 4Gb version of the card. Not many places did, and they seemed to have a great ruputation for customer service. However after spending 60 mins on hold to get through this morning just to talk to them, and listening to the repeated message over and over, telling me how great their customer service is, I was ready to punch someone :pt1cable:


 

dandare

Honorable
Dec 9, 2012
69
0
10,630


As another update, Gigabyte UK have deemed the card not faulty. :pfff:

They claim to have put it through over 18 hours of stress testing on both the EX58-UD5 and the G1 Sniper 3 boards and cannot replicate the fault.

I will commend their turn around time of just 48 hours, but i am now mighty annoyed and starting to get a bit desperate. I mean c'mon.. I built a whole new system and the damn thing still faults.. and yet they cannot replicate so say it is not faulty.

The ONLY thing that is in the new build that was in the old one is the PSU, and that may make you think, ahah... there's your culprit.. but then explain how my GTX480 works absolutely perfectly with this rig and the old one. That PSU is only 4 weeks old, so not exactly old. 5 hours solid of 3dmk11 and the 480 does not fault, 3 hours of prime 95 torture test and the rig does not crash.

I am at a loss as to what to do here. I am faced with being left with a £450.00 paper weight, and I have not even owned it for 2 weeks.

My current PSU is a Corsair HX 770 Gold - so pretty good. I see no reason why this could not power 1x GTX680, unless it is faulty.

My only option is to buy a new PSU, a bigger one. And therefore see if it that or not. It is the final remaining common factor between my two rigs, and it would leave nothing else. Problem is it will cost £175-200 for a decent 1000W PSU. And it may turn out not to be that.....