Overclock.net › Forums › Graphics Cards › NVIDIA › GTX 970 G1 gaming Black Screen/Hard Lock. (SOLVED by changing faulty PSU)
New Posts  All Forums:Forum Nav:

GTX 970 G1 gaming Black Screen/Hard Lock. (SOLVED by changing faulty PSU)

post #1 of 9
Thread Starter 
Hello, I have my gtx 970 g1 gaming for less than 3 months.

At start everything was perfect. Tried some overclock with MSI getting like +100 core, +500 (or maybe even +750) looking stable (in furmark) and tried it on rocket league and Ark.
After some time for no reason (the only thing i remember is i was away for some days dunno if this have anything to do, so PC was shutdown for days) and then black screen/hard lock (caps lock not working) started showing up. Using furmark instantly black screen.

Started searching here and there, blamed mostly drivers so tried playing with versions. With changing drivers, playing with windows driver's update on/off etc made it fine for 1/2 days before black screen come again.

What i tried: What I tried: (Click to show)
-Different driver versions, from old up to the very latest one (for Win10 universary update), using DDU.
-Lowering CPU OC, even putting BIOS at default, also tried CMOS clean.
-Tried on my 2nd windows 10 boot in case my 1st got any problems.
-Re-seating GPU, changed where the PCI-E 6 & 8pin cables goes at the PSU (out of 4 or 6 slots i just picked 2 other in case there was faulty??)
-Underclocking GPU (75-85% TDP and -180/-150 core clock, but memory looks fine even at 0/+200 <- these settings look like were stable but didnt tried for more than day so cant say for sure)
-Putting my GPU on my Bro's PC (everything were find for 2/3 days except 1 hard lock before i uninstall AMD drivers and install NVidia and 1 hard lock on windows boot which he says he never got hard locked before, but games were running fine)
-Tried my bro-s AMD gpu 6870?? everything looked fine. (maybe hard locked once? not sure, i got some problems here and there before so i also give some chances for faulty motherboard??)
-Used MAX-Unleashed bios, in case the blackscreen was because of low voltage or delaying of giving enough voltage when boost clck was about to enable, maybe its more stable but still it gave me black screens :/
-More aggresive fan settings.

-TDR delay 2->10
-Disabling powermizer (propably its only in laptop but w/e)
-Nvidia settings for maximun power,..., even physX to CPU

I used to test with furmark, if it is about to crash it will do it 99% instantly, after that i mostly used Heaven, and just recently a bit the 3rd mark.


GPU-Z log doesnt show anything weird the moment before the black screen. No high temps (less than 68-70), usually if i m playing a game and its in boost clock, it just freeze with last log in boost clock, if i m at idle and i try to run furmark or so it will black screen before even boost clock is enabled. With the tweaked bios tdp doesnt go more than 80/81% in ARK before it crashes.

Things havent tried: Things I havent tried: (Click to show)
-with new bios doing clean install of drivers (maybe with a new method read here, manually uninstalling drivers and cleaning files/registry.) and maybe trying old drivers.
-Trying other cables (from psu->gpu) in case they are faulty??? the chance here is near 0 but when i open it up again i can't lose anything, plus re-seat GPU and be sure it's seating right.
-Baking GPU (it's in warranty so no thnx tongue.gif )

ANY suggestion is Welcome!!! Thnx a lot in advance for ur time!!
I still can't tell for sure if its faulty GPU (as it works on my brothers' PC <-not tested enought days though)
OR a faulty PSU
OR a faulty MOBO
I believe its propably a combination. Propably the GPU isn't perfect/more "sensible" with power. PSU isnt gold/platinum to be sure that its doing its job perfectly, same can go for motherboard, which is 6+ years old and maybe can't power propertly the GPU throught PCI-E 2??
I wanted to be as sure as possible if its GPU or something else, before RMA or anything. But maybe i ll contact my seller to see IN CASE i need to RMA it what i should do etc.

Looks like its a common problem with 970, and maybe even more common with factory clocked cards. (based on many posts/complains in forums etc that i checked in internet)
Tried posting in nvidia forums, but no lack (lack of replies...),
In addition I done a ticket to gigabyte and they replied this:
[Gigabyte Reply:]Thank you for your kindly mail and inquiry. About the issue you mentioned, please try the following test procedure:
● Confirm there is sufficient power to the platform.
● Make sure there is no overheating problem with the system and graphics card.
● Update game patch to the latest official version.
● If possible, try to test your graphics card and the games in another system, or test your motherboard with different graphics card to see if there's a hardware problem. 
If the problem still occurs in another system, a further testing or examination to your graphics card might be required. We suggest you contact your supplier (where you bought the product) to arrange further examination and find out the root cause.
Regards,
GIGABYTE[/SPOILER]
PC Specs!!!! (Click to show)
CPU: AMD phenom x4 BE 955 (@3,8ghz)
Mobo: MSI 790FX-GD70 (did bios updates i think quite some time ago)
PSU: Corsair HX 750W 80silver+ (it got 7 years waranty, and its like 5years old so it's still in waranty!!)
GPU: Gigabyre 970 G1 gaming (Elpida Memory) (on monitor via DVI) (running in D_P bios)
RAM: Crucial BalisticX 8gb (2x4gb) 1600 CL8 1,5V
OS: Windows 10 64bit fully updated, running on SSD (got a HDD as well for back up/storage with widnows 10 as well)
post #2 of 9
What you describe sounds like a PSU issue all day... whenever you place a significant load on the GPU (POWER not utilization) you get the black screen, meaning your PSU cannot provide the power. Be sure you are not using "Splitters and/or adapters" for the PCI-e cables. They should be straight to the PSU. If your motherboard has a supplemental molex for the PCI-e slots, connect that also (usually needed for more than one gpu).

Not sure what size the other PC's PSU was... If it's not the PSU (insufficient or faulty) then it might be something power related on the GPU itself (extremely rare).

PS. Overclocking GPU MEMORY draws more power, you can keep the memory @ stock and see if that helps any.

Good luck smile.gif
post #3 of 9
I had a Gigabyte GTX 970 G1 Gaming that I oc past the factory OC with never an issue so I don't know where you heard it's common.. I still have my GTX 970 in the box, ran it for over a year with never a problem.
Anyway what I will say is NEVER use FurMark on a gpu! it alone can fry a perfectly good gpu.
I think it's a driver issue caused by something else on your system as the gpu worked good in a different system except you said it had 1 hard lock
in the other system, if you can try a different psu, or connect a digital multi meter to a pci-e power cable and monitor it when the system hard locks.
a psu problem would cause a reboot, as far as I know, but I could be wrong. never hurts to test the psu.
did you check windows event viewer for TDR's? (Time out detection & recovery) it shows up as nividia driver crash. a really bad driver crash can cause a black screen and system lock
event viewer sample not mine:


also shows up on task bar as nvidia driver has stopped responding and recovered, but since your screen goes black you can't see it.
Edited by jlhawn - 8/21/16 at 5:55pm
Old X58
(17 items)
 
  
CPUMotherboardGraphicsRAM
i7 970 6 core ASUS Sabertooth X58 MSI GTX 1070 Gaming X GSkill 24 gigs 
Hard DriveHard DriveHard DriveCooling
Crucial MX100 256 gig Crucial M500 120gig WD VelociRaptor 1TB Noctua 
OSMonitorMonitorKeyboard
Windows 7 Pro 64bit LG 29inch widescreen DP LG 29inch widescreen DP Corsair K65 
PowerCaseMouseAudio
EVGA SuperNova G2 1000w Corsair C70 Corsair M65 Sound Blaster Z 
Other
Custom sleeved psu cables 
  hide details  
Reply
Old X58
(17 items)
 
  
CPUMotherboardGraphicsRAM
i7 970 6 core ASUS Sabertooth X58 MSI GTX 1070 Gaming X GSkill 24 gigs 
Hard DriveHard DriveHard DriveCooling
Crucial MX100 256 gig Crucial M500 120gig WD VelociRaptor 1TB Noctua 
OSMonitorMonitorKeyboard
Windows 7 Pro 64bit LG 29inch widescreen DP LG 29inch widescreen DP Corsair K65 
PowerCaseMouseAudio
EVGA SuperNova G2 1000w Corsair C70 Corsair M65 Sound Blaster Z 
Other
Custom sleeved psu cables 
  hide details  
Reply
post #4 of 9
Thread Starter 
Thnx a lot for bot replies!!!

@Laithan
My brothers PSU was pretty new, 850W gold+ so yeah his PSU is good (corsair RM850w i think).

750W should be enough, w/ 62A at +12 and 25a @+5 but maybe its faulty (it also shutdowns when there is a little dusturbance in house's power like light flicker's once which my old PSU was ok, my brothers is ok as well, plus it supposed to got protection for these things, so maybe i ll check if i can change it either way.)

Can i test my PSU somehow? with the multimeter jlhawn suggested???

My questions are:
why it could play normally with my borther's cards which doesnt need much less W than mine.
why at 1st it was fine for almost a week and then it suddenly got this problem???
why sometimes i can play like 2/3 hours ark (with ur bios) and then just boom? i mean its already running at top voltage....

Finally about the RAM clocks, im a bit convinced that its more about core clock, as if i underclock it to -200 or so with less TDP the memory clocks are fine at +0 or even overclocked to +200 but the vica versa having underclocked memory but normal core still gives errors, which i am afraid of the boost clock that also goes up and down when u change core clocks.

ps: Not trying to be a dumb ass or something with so many questions, i ll check the psu propably either way just to be sure, but except that i try to fix my error im trying to learn as well here ^_^


@jlhawn
I read about furmark many ppl saying it no good and i use i ll ditch it and use only heaven,3d mark and games, if it works there i m ok, but its weird not being able to run these things too...

To test my PSU i need to find/buy multimeter or some software like HWmonitor can help???

About event viewer when i check at Critical events it only shows the error that it takes from the hard shutdown i do (41 kernel-power). checked warnings as well and didnt found any of these (only one 13 days ago or something so its not the root cause for sure)





Shall i revert to factory bios? cause now i got the Laithan's unleashed bios, cause i though maybe gpu isnt getting enough voltage between idle and boost clocks.
Plus if i dont run furmark to instantly test stability, what shall i try? heaven, 3d mark??? i feel these can play quite a bit before they go down, and if i m lucky/unlucky will think its ok for no reason.
post #5 of 9
software to check your psu is not good, multimeter is the best.
have you considered at UPS?
if you think there might be bad power from the electrical circuit you should really get one.
I use one of these, but there are others that are good for less money, Cyberpower is good also.
http://www.newegg.com/Product/Product.aspx?Item=N82E16842101419&cm_re=ups_battery_back_up-_-42-101-419-_-Product
Old X58
(17 items)
 
  
CPUMotherboardGraphicsRAM
i7 970 6 core ASUS Sabertooth X58 MSI GTX 1070 Gaming X GSkill 24 gigs 
Hard DriveHard DriveHard DriveCooling
Crucial MX100 256 gig Crucial M500 120gig WD VelociRaptor 1TB Noctua 
OSMonitorMonitorKeyboard
Windows 7 Pro 64bit LG 29inch widescreen DP LG 29inch widescreen DP Corsair K65 
PowerCaseMouseAudio
EVGA SuperNova G2 1000w Corsair C70 Corsair M65 Sound Blaster Z 
Other
Custom sleeved psu cables 
  hide details  
Reply
Old X58
(17 items)
 
  
CPUMotherboardGraphicsRAM
i7 970 6 core ASUS Sabertooth X58 MSI GTX 1070 Gaming X GSkill 24 gigs 
Hard DriveHard DriveHard DriveCooling
Crucial MX100 256 gig Crucial M500 120gig WD VelociRaptor 1TB Noctua 
OSMonitorMonitorKeyboard
Windows 7 Pro 64bit LG 29inch widescreen DP LG 29inch widescreen DP Corsair K65 
PowerCaseMouseAudio
EVGA SuperNova G2 1000w Corsair C70 Corsair M65 Sound Blaster Z 
Other
Custom sleeved psu cables 
  hide details  
Reply
post #6 of 9
Thread Starter 
Well most PSU got some small protection. I don't got full power downs usually, neither got fried electrical devices so didn't thought its a must. And for now its out of question/budget. Maybe in the future if really bad power can short by a lot the lifetime of the PSU.


UPDATE about PSU!!!
I called my supplier and talked to the technical stuff. Told him about my PSU that supposed to got protections but pc shutdowns/freezes even on slight irregulation, but other PCs doesn't (my older PSU didnt either) and also that i bought a gtx 970 and got black screen freezes, he said its propably the GPU, or maybe the PSU doesnt got enough power (750W im pretty sure is enough, checked many sites with calculators as well and agree) so i guess this is just a random BS he said. He didnt say its a fact just a possibility.

So i got 2 options: Take whole system there and pay 20 euros to check it for me in hours, i guess if i need to change parts etc will take up to days.
Just take the PSU there to check it and cause its in waranty its free, but they ll need about 2 days or so.

Im thinking of propably unplagging the PSU and bring it there for a check up tongue.gif
And meanwhile put back my old Nexus whatever 1000W PSU (long story short, a technician said once that its faulty and its hangign around at the house till then.)

It will propably kill 2 birds with one stone, trying one other PSU bigger, maybe faulty as well (so 1 bird maybe flee...) but at least i ll have my pc running and will get the other getting checked.

It's propably what i ll do today or tomorrow (got a friend who maybe can check the system with multi meter so if thats the case i could consider doign that 1st.)

EDIT/UPDATE:
Sent PSU to the shop where i bought it, they ll check it but he said it's maybe difficult to find if its faulty. The thing is that i got only Corsair warranty and will need if its faulty to send it back to them but i think there isnt one in greece and i need to send it propably to Italy?? And i guess 10-15days will be needed at least, so so many days without PC will suck... If they find it faulty i guess if the fees to send it are normal / cheap will send it . Now, if they don't find it faulty, which doesn't mean it isn't but at least it got tested on other components and it works fine , then i ll try to check if something else is at fault and won't send it to corsair till its the final option.

Asked my GPU seller what i ll need to do if its faulty. Got 2 options
-Either i send it to them and they ll do the rest, propably will need 1 month or so.
-Or I need to send it myself, which i think there isn't a representative here in Greece and well still it ll cost something but maybe its faster than 1st option.
Made a ticket to gigabyte to be sure if there is one in greece cause propably there isnt.

My old PSU 99% is faulty so technician propably was right, booted once, but it got black screen and hard lock, after that couldn't boot, it opened but full black screen, and couln't even hard power off it from the button, needed to close the PSU from behind, so im without normal PC, only just a ancient laptop.
Edited by 1stratos1 - 8/23/16 at 3:47am
post #7 of 9
Thread Starter 
The shop that i bought it found the same problem as me, when stressed they got shut down or black screen/freeze.

So yeah its faulty, im doing a ticket right now at corsair, the SN number on the PSU though is ripped in half, still on the box i can find it, so i hope they will accept it.


Question, my CPU/Mobo even if old the wont create propably any issue with the 970 right??

I were thinking maybe updating to skylake i5 6600k sooner or later, so in addition while waiting maybe i ll do it as well.


Thanks to both of you for the replies!!!
it's not that the problem is solved, but if the PSU is faulty, then the chances this was the cause are near 100%.

I still keep my card at my brother's PC for a few more days just in case. Anyway for now i don't got pc, propably i ll put it back at my PC till i got a reply from corsair, hope tomorrow, and then i ll try to find a replacement or something if its for more than a week total which wil propably be,

Thanks again, i ll update it as soon i got something new.
post #8 of 9
Quote:
Originally Posted by 1stratos1 View Post

The shop that i bought it found the same problem as me, when stressed they got shut down or black screen/freeze.

So yeah its faulty, im doing a ticket right now at corsair, the SN number on the PSU though is ripped in half, still on the box i can find it, so i hope they will accept it.


Question, my CPU/Mobo even if old the wont create propably any issue with the 970 right??

I were thinking maybe updating to skylake i5 6600k sooner or later, so in addition while waiting maybe i ll do it as well.


Thanks to both of you for the replies!!!
it's not that the problem is solved, but if the PSU is faulty, then the chances this was the cause are near 100%.

I still keep my card at my brother's PC for a few more days just in case. Anyway for now i don't got pc, propably i ll put it back at my PC till i got a reply from corsair, hope tomorrow, and then i ll try to find a replacement or something if its for more than a week total which wil propably be,

Thanks again, i ll update it as soon i got something new.


thumb.gif

Sorry for the delay, have been away for a few days. Glad you've been able to confirm it. It's bad news but good news that you'll soon have resolved the problem.

Keep us posted
post #9 of 9
Thread Starter 
Corsair said they couldn't accept it cause of the ripped serial number... After i thought mabe i could change the rippen serial number of the box with the one in PSU without telling them (which i already did and send photos so propably doesnt matter anymore)) and propably it would be ok? but still could risk like 30-35euro for sending it there...

So i bought Corsair RMx 850W for like 150euro, i ll get it in a few hours and i ll update this post on how it goes.
I read that they are solid with 7years warranty (i know about evga supernova being a bit better at about same price but they arent popular here :/ )

UPDATE 1:
PSU installed, it looks fine for now, im even trying OCing it, by trial and error i got artifacts, and benchmarks stop responding, BUT no shutdown!!!

Propably it's fixed, but i ll update once more in 5/7 days while i can test it more with benchmarks and in gaming.

Thnx a lot again guys!!! Faulty PSU by having problem only when stressed was more at the bottom of my list so i really got saved so much time that i could "waste" looking for million possible solutions!

UPDATE 4-Sept-2016

The PC seems to be ok now, no black screen and freezes etc etc.
I had once or twice a freeze but it was because of drivers, so its something entirely different and propably was cause of high overclock (in cpu?) or weird settings in the game.

Thnx once again guys!!!
Have a nice day!
Edited by 1stratos1 - 9/4/16 at 5:46am
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: NVIDIA
Overclock.net › Forums › Graphics Cards › NVIDIA › GTX 970 G1 gaming Black Screen/Hard Lock. (SOLVED by changing faulty PSU)