Faulty GTX 680 or somthing else?

Page 2 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
Toggle sidebar Toggle sidebar
A 680 would run just fine on a 500w PSU. Have you tried the 480 in Furmark for a few hours? Furmark is a proper stress test, not a benchmark. Oh and P95 only tests the CPU.

I don't really know what to tell you other then asking gigabyte to replace it anyways since it doesn't seem to work for your config. I would explain that since they don't seem to find any fault with it then it should be no problem to re-sell that one and send you a different one.
 


I sincerely agree, but, I have been told that they HAVE to prove a fault to authorize a replacement, which HAS to come from gigabyte HQ in Asia if I go direct through GB rather than retailer.

The Retailer say they either HAVE to find a fault themselves or have it verified by Gigabyte. If they cannot prove it I will be charged a service fee and have the card returned

I have decided to do two things.

1. After starting a thread about the PSU, a couple of peeps here on Tom's have suggested it is more likely a bad PSU. Given that Gigabyte cannot replicate the fault I am starting to think that too. So i am going to buy a new PSU. Either an AX 860i or AX 1200 -same price, the 860i is platinum grade, but AX 1200 gold has 100 amp rail, which would come in handy if I want another 680 in a year to boost this pc.
2. if fault persists after this I will use my credit card company to issue a charge back against the retailer. I am speaking to them tomorrow to get details on starting the procedure.


 


Are you refering to this exact brand and model, or 600 series in genral? If you can point me to other poeple having issues it can only help to back up my case, most importantly if I have to get legal through redit card company.
 


This is what gigabyte asked for, but the nature of the crash, (an instant re-start and loss of video) and the fact that it is hugely intermittent makes this hard to do. Last time it ran on 3dmk11 for 2 hours beofre crashing. A screen shot would not prove much.

Could I use fraps? i wonder if it would save if the PC died during recording.
 


no video cam. problem with using phone is that video takes up alot of memory. it could be running for 2 hours+ and may not fit on phone sd card. But yes this would be the best solution, as i could then boot back up while filming and show them the error report live.

EDIT** Just ordered a new Corsair AX850W Gold £154.68 with shipping tomorrow. (company ships up to 11pm) 680 is due back tomorrow too so will update tomorrow night or morning after wheter the issue appears to be fxed or not.

 
Good luck man so many unnecessary hassles 🙁

its not like the 750hx isn't a shitty psu and also not like the 480gtx isn't anymore power efficient.

Hopefully in hindsight it was the psu and soon will fix all your problems,
 


OK, so the card came back today. I thought that befroe i opened the new (selaed) PSU and am unabl to return it I would just test the machine again.

first i put it through Fur mark 1920x1280 benmark on the 480. This put the 480 up to a whopping 93 degrees!, I then ran a pass of 3dmark 11 with no issue.

Put the 680 back in and ran same furmark test with success, but max temp was only 67.

Then ran 3dmark11 and and on the first pass, 4th graphic test, it crashed again. I am now running a loop on 3d mark 11, but it is not crashing... could take a while, but I do have it on video.

I am torn as to whether to open the £154.00 power supply. I do i am stuck with it and will end up with two PSU's... if I do not then I suppose I will never know. I am leaning toward opening it and leaving it on a loop run.


 
The 680 was crashing by simply opening jpeg files, is that still happening?

As its crashing more in 3dmark11 than furmark are you using the latest version of 3dmark11 1.0.3? http://community.futuremark.com/forum/showthread.php?159546-3DMark-11-patch-1-0-3

Do you have another spare hdd you could try reinstalling windows on? Iron out both operating system and hdd itself if possible, might not be a hdd fault but just in case the 680 is somehow making any faults more prevalent, software or hardware glitches.

Could try legacy drivers also, could be something nvidia have done. Would be good to know if gigabyte did use the same drivers as you or they know something you/we don't.

http://www.geforce.com/drivers/beta-legacy

Lastly would be no choice but to try the new psu imo.



 
if by chance the new psu did the trick. It might be an idea not to throw out the older psu as it might not be it's fault by fluke design incompatibility.

Seeing as many people are having unsolved issues with the 680, it just as well might be the card and how it interacts with the psu. Like how the motherboard communicates with the psu, sending a signal to keep power on after a successful start up.

http://en.wikipedia.org/wiki/Power_Good_Signal

It's a wild guess but gota wonder if its even remotely possible the 680 isn't receiving the correct operating wattage and amperage at any random given time. Hopefully whatever it is, problem is not present with the new psu.
 





OK, so thanks for the replies mate.

1. I contiued the testing and got it to crash again of 3dmark, but after an hour of looping tests. (on old PSU) - I am using the layest version, and it crashed on it on both this pc and the old one, so i m sure it it not a direct softwae issu like a corrupt instal
2. I cannot say if it would crash on furmark, but I only did 2 x15 min benches, and it may well have done if i left it fr longer.
3. As I have not really "used" my pc since building the new one I have not expereinced the crashes outside of bench tests, as that is all i have been doing.
4. No hdd, just SSD. OS was installed on a new 256gb SSD on old machne, and when building the new one I bought a second 128mb SSD and used that for the OS instead. So it has falted on two SSd's and two fesh, seperat instals of Windows.
5. annoyingly GB did not se the same drivers as me. I was using the whql from October 306.97. they chose to use the new 310.70 drivers that came out the day I sent it to them.
6. I did wonder about it being some random isse like you say, although I have little knowledge of such things. It couldexplain why one card works and another does not.
7. I was not aware of that many people having issues. Could yo pnt me in the direction of any other thread? - if things do go bad then any spport in terms of others having issues would be helpful.
8. New PSU is in and working. going to leave it running overnight and see wht happens.
 
Not directly to your problem as you or i would have found search results to your unique problem already, but just saying the 600 series of the searches i've done with the issues i've read tends to be more about the troubleshooting questions rather than solving answers. Searching of phrases like 'gtx680 directx 11 crashing'.

Most responses either point towards either faulty cards or faulty/insufficient psu/power and suggesting the 2 6pin connectors are connected properly.

Rest of those responses are just repeats orientating motherboard/cpu and ram which your gtx480 had no trouble with and then there's software, os and drivers.







 


good morning...

Thanks for the info above. So, got up this morning and it was still running, so I take that as a good sign. That is the longest it has run for wth that card in without crashing.

I will not say the issue is solved yet... but I am hopeful. Let's see what happens over the next few days. fingers crossed I might be able to play some games this weekend... I will run benchmarks again tonight and see.

 



This is good news! Glad to see now there is a change in events and hope it stays this way.

Very curious to know what actually happened since the corsair 750hx is a popular psu, guess we'll never know exactly :\

th



 


Thanks boju.. picture cheered me up. 😀

I am of two minds what to do with the PSU. Considering what you say above, and the fact it works with one card and not the other, it could simply be one big weird ass incompatability issue.. gremlins :)

I could RMA it, only to be told it is not actualy faulty and have to pay shipping both ways, adding to the expense. This card cost me £429.00 inc shipping. + £15.00 to send it to GB for testing + £154.68 for the new PSU. For that money I could hav had two GTX 670's (almost) so a little depressing really

Or I could sell it on ebay, but I may feel a bit bad as I cannot be sure it is woking 100%

Or I could just bin it and move on.

Mind you, I may find that over the weekend the problem comes back!
 
Hmmm, I would wait on doing anything until we know for sure its solved but uh, I would probably sell it on ebay. Honestly I have no clue what the issue could be, if its fixed with the new power supply good. If not then well... we just go cry in a corner and blame wizards.
 


Well I have been sat in a corner hating the wizards...

After 4 days of no erros it came back on Xmas eve... like a perfectly timed smack in the face to spoil christmas. :pfff:

I have now tried everything. Nothing left to test, it is the sole remaining component tht were shared by the two computers and the falt persists. i have tried different drivers, different sockets and surge protectors and nothing works.

Looks like I will now have a new year fight on my hands to get refunded. Sour news considering I shelled out on that new PSU.
 
bummer what a slap in the face, meh.

So, the problem still lingers. Its a weird suggestion but could it be the connectors, possible faulty power cable leading to the card or the connection plugs on the card itself. I would assume you would have used the new cables that came with the new psu, and even then the 480gtx had no problem so there could be a slight power short with one of 680's 2x 6pin ports.

Either Gigabyte got miserably lucky and failed to find or they lied and didn't test at all.

 


Indeed.

I had also though of the power connectors. One of the ealrier things I tried was swapping the cables over even on the old PSU; and yes, I did use all new cables with the new PSU too.

I think GB did test... I Can't imagine they would lie, I mean if they did find a fault then for a massive international compnay like them, what is replacing one card? They did send me test results, and it shows the card running after 15 hours solid. however it ran on my system after putting in the new PSU for 8 hours solid, and then did not fault for 4 days, then 7 times in one day, under no particular load. rest of the erroor (Kernel Power 41) remains the same no matter what.

I wrote a long emil to the retailer today, asking again, that given the circumstance they refund me. If they refuse then I shall be persuing a claim with my credit card provider and ensuring that I smear their name on every forum I can find +post bad reviews wherever I can. I would then have to persue GBT directly for a replacement.

Will kepp you posted with what happens.

I know it is not a software issue as it happens at different times doing different tasks. Weirdly one bug code that is displayed only occaisionally after the crash, in windows reliability monitor; tells me that it was a blue screen BC code 117, which is a failed driver code. However it does not do this every time, and has done it with both 306 and 310 drivers.

EDIT*** The retailer replied to me, and have agreed to collect the card and return it to their HQ. Dowside is they still insist on testing the card to prove a fault. I have politley argued that I have done more than enough to prove it already, but they have ceased responding to my emails.
 
OK, so, the retailer have had the card back for two weeks, and guess what? they cannot find a fault... they are refusing to refund me because they say they would be unable to get their money back if they did.

They even cheekily offered to sell the card second hand on their site and give me the money for it! another reason to steer clear of Lambdatek Ltd, as they seem prepared to sell faulty goods as used on their site.

I am open to suggestions as to what the fault could be! I mean both Gigabyte and the retailer say that they have tested the card and it does not fault, yet when it is in my two separate machines, one a 1366 socket with i7 920, and the other a 1155 i5 3570k it produces an identical fault on a regular basis.

I have tried different wall sockets, different surge protectors, two brand new PSU's, new leads, multiple installs of the OS and different drivers and it always faults for me - yet they both say the card is fine....
 
Hmm, I'm not entirely sure what to tell you. If its indeed faulty and it apparently is yet the retailer and manufacturer are refusing to take it back... I just don't know what to say. Other than physically taking it apart and seeing if you can find a fault that way (unlikely and just DON"T do it as doing this WILL void any warrantees you have)

I'll keep thinking on it but this is pretty much the limit of my expertise. Pursuing this any further would likely require legal advice sadly.

You could try going into GPU-Z and using that to report temperatures for your vram/mosfet/etc while running a stress test to see if anything is overheating.
 
You must log in or register to reply here.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom