Nvidia 980 unstable TDR

[EnjoyTroll]

I have had a 980 for a little over a year now and it has never been such an ass of a job for the past year I have had to be dealing with TDR issues on every single game I throw at it normal computer use it is fine just when gaming.
The TDR issue is random sometimes it happens seconds into a game loading and sometimes I can play for over an hour without a TDR....Now this is where it gets weird I can run the Unigine Heaven maxed out for hours no issue temps in limit's as it's water cooled never goes past 60c.

As you can see the card is more than happy to keep pushing with boost 2.0 and with more increased power target it does keep boosting more I believe Voltage caps will be the limit for this card smile.gif This is the same story for when gaming everything stays within checks no spikes or anything.

My card is an Inno3D iChill GeForce GTX 980 stock clocks for this are shown in screenshot

As you can see I have tried already to turn down the default memory clocks to the stock Nvidia 980 to see if this was an unstable memory clock however on doing this it makes no difference I was still getting TDR problems this did not change when they happened either.

Things I have tried so far:

- DDU the driver and try again with clean install

- Clean install of windows 10

- TDR manipulator or something which changes the TDR values for the GPU

- Running the card at stock 980 speeds both on GPU and Memory :core:1126:

- Setting up Nvidia control panel to "prefer max performance" on all the
games I try and play

- Clean install of drivers only installing PhysX and the driver nothing else
- Tried all the drivers from 358.70 and up so far and some either crash faster or some the same time but none fixed it....Really no joke I have tried every driver going including the new ones just come out "362.00" is the most stable I have found so far with "359.06" in second. I have done so much Google action on which drivers are finding better for the TDR issues.
- Removed RAM sticks you name it
- Tried different PSU cables
- Tried different PSU and RAM
^^Tried my old GTX 570 with the same drivers on windows 10 and that was fine form what I tested and my 570 is one old hot cookie ^^
- Put my CPU back to stock clocks no difference
- Moved the GPU to a different PCI slot on the mobo no difference
- Asked a few friends I know they also have this issue but not as often as I am getting it
- Even underclocked the GPU both on the core and memory...This just made the games run like crap and still TDR
- Only used one screen to see if there was an issue with that
[/LIST]

I have done so much looking into this subject of the TDR issue and I know it's been an issue for years I don't believe it to be a windows 10 issue as I am still seeing windows 7 and 8.1 getting this issue with the drivers.

EDIT: Quick note I have had a 980 before which was the 980 HOF I had this on day of release as I loved the white card I however had to sell this due to personal reasons at that point I was running windows 7 and what ever drivers were out back then and I never had an issue back then. It was a while later before I could afford to buy a 980 again so much the drivers have gone from early 34* drivers to 35*+ by this point.

I have seen people RMA cards for this issue and it has clearly not fixed the issue so I have not bothered with this stage.

I am however losing the will to live with Nvidia and their drivers and I have always had Nvidia

I brought this GPU on sale from Overclockers with them recommending it and assuring me they was good cards and not cheeping out on parts I am now starting to wonder if the card is just simply a dud and I have been spending all this time trying to fix it for nothing. As I said I had a 980 HOF before this and it never had these issues however that card ran very hot and coil whined really bad.

Thanks for reading and any ideas on this matter I know it's a lot to read

 

[MrMD]

Do you have GPU-Z open when gaming? If so do you notice core voltage and clock drop down just before the crashes occur?

 

[LegendaryDread]

Tried updating your motherboard's BIOS? Both my HD5870 and GTX580 hated my AM3/AM3+ boards - I believe they were ASUS, can't remember the models though. Very similar behaviour to what you're describing.

 

[EnjoyTroll]

Do you have GPU-Z open when gaming? If so do you notice core voltage and clock drop down just before the crashes occur?

Hi thanks for getting back I don't get a chance to check through GPU-Z when the TDR happens it stays on the black screen for a good 10 seconds and then it makes msi afterburner and GPU-Z reset and the readings get glitches to numbers in the millions Intill I reset the gpu in device manager.

Tried updating your motherboard's BIOS? Both my HD5870 and GTX580 hated my AM3/AM3+ boards - I believe they were ASUS, can't remember the models though. Very similar behaviour to what you're describing.

I never thought about updating the bios for my Asus crosshair tbh sins moving to Windows 10, from when I last looked they hade only got one new one from when I last updated it but didn't say if it helped support windows 10 or anything as Asus class the 990fx chips etc dead now

 

[kola]

My card is a 780 Inno3D iChill , and TDR's allot , I have to reduce the memory clock & core clock buy 40Mhz . Not buying another Inno card again . should have returned it way back :/

 

[MrMD]

Hi thanks for getting back I don't get a chance to check through GPU-Z when the TDR happens it stays on the black screen for a good 10 seconds and then it makes msi afterburner and GPU-Z reset and the readings get glitches to numbers in the millions Intill I reset the gpu in device manager.

This sounds similar to an issue i and many others have had wth 900 series cards related to how maxwell deals with power states and dynamic clocks,its usually more apparent on OC'ed cards but it has affected quite a few people at stock clocks too.

The easiest way to check if this is your issue is to download EVGA percsion and enable 'K-boost'.This will lock your card at 3d clocks at voltage.Do this and try some games and see if your card still TDR's.

 

[EnjoyTroll]

This sounds similar to an issue i and many others have had wth 900 series cards related to how maxwell deals with power states and dynamic clocks,its usually more apparent on OC'ed cards but it has affected quite a few people at stock clocks too.

The easiest way to check if this is your issue is to download EVGA percsion and enable 'K-boost'.This will lock your card at 3d clocks at voltage.Do this and try some games and see if your card still TDR's.

I use MSI afterburner is there a option to do this through there? if not then I will download that and have a look as I am losing the will to live with this card so unpredictable haha

My card is a 780 Inno3D iChill , and TDR's allot , I have to reduce the memory clock & core clock buy 40Mhz . Not buying another Inno card again . should have returned it way back :/

I have set this card back to stock 980 clocks and it made very little difference on the TDR issue however when I turned down the memory clock it improved loads hence why I have now got the memory at stock speeds however it stills TDR's

 

[MrMD]

I use MSI afterburner is there a option to do this through there? if not then I will download that and have a look as I am losing the will to live with this card so unpredictable haha

Theres is an option in afterburner to do it,but for whatever it doesnt work at all on nvidia cards as far as i can tell,even MSI ones

 

[EnjoyTroll]

Theres is an option in afterburner to do it,but for whatever it doesnt work at all on nvidia cards as far as i can tell,even MSI ones

Oh okay then maybe that's worth a shot thanks

Tbqh I just want to RMA it I think with them but on the website they don't stock them anymore as far as I see it if the cards not stable at nvidia stock clocks something can't be right every driver I have tried TDR's even drivers that never did with my old 980 HOF although that card had its own issues this generation of card has been nothing but hassle

 

[StrongSad]

Hi I had known issues with mine.

What I did as a short term fix was to use Prescison X, enable K Boost to fix the card to its boost clock speed to prevent it dropping its clocks. In all my fault finding (reducing clocks, format reinstall, blah blah blah) I found that the voltage dropping below 1 volt and then reboosting caused my drivers to crash D3D games due to the voltage dropping. Synthetic benches were more than fine and card is stable in games (with K boost) stable at 1499.

However this was not just my inno 3d card. There are many maxwell users (see elite n GTA forums who have had this issue across all Board partners. Mostly on factory clocked cards... K boost or voltage Bios modding seems to fix. However far from ideal that is..

 

[Bluntwrapped]

You could always edit the cards BIOS to remove the voltage cap. Have you tried maxing out the voltage slider on AB?

Out of interest what is the displayed ASIC quality of the card?

Edit: If you can upload a copy of your cards BIOS I will be happy to take a look at it for you.

Taking a quick look it appears you may be limited to 1.21v but would need a copy of your BIOS to confirm.

 

[Rroff]

My card is a 780 Inno3D iChill , and TDR's allot , I have to reduce the memory clock & core clock buy 40Mhz . Not buying another Inno card again . should have returned it way back :/

Had a problem with some nVidia drivers awhile back (somewhere around 358-360 IIRC) where my 780 would TDR in a good few older games unless I enabled K-Boost which is a bit of a pain as you don't really want it running K-Boost 24x7 but it hasn't been happening with the last few drivers.

(Was also seeing similar behaviour in other Kepler based cards at the time so wasn't my 780/system specifically).

 

[EnjoyTroll]

You could always edit the cards BIOS to remove the voltage cap. Have you tried maxing out the voltage slider on AB?

Out of interest what is the displayed ASIC quality of the card?

Edit: If you can upload a copy of your cards BIOS I will be happy to take a look at it for you.

Taking a quick look it appears you may be limited to 1.21v but would need a copy of your BIOS to confirm.

Hi I can send a copy of the bios for you no problem but yes I am sure the cap is around that mark for the volts whowever if I use the evga tool or the MSI one I can push the volts much higher +87mv however I tried this tactic by smashing the volts up and it did not help all this did was make my card boost more and run hotter.
My ASIC is 78.5% which is higher than what I had on my old 980 HOF and that come shipped with same clocks on the GPU however that one had Samsung memory and this has hynix thats why I thought to turn down the memory clock to stock but didn't really help https://we.tl/DwGkaGfOwR

I am going to give the K boost option a go and see how that works but this is a pain just to get a card to run without TDR issues at any clock oh man....

Edit again I have been playing a game now for 10 min so far which is better than what I get normally without using K-boost however I can before run for hours no issue before it did it so will keep that part posted here is a screenshot of the card when underload with K-boost

 

[EnjoyTroll]

Update to the thread I have tried using the K-boost through the app with the memory clocks to stock and nothing has changed for the card at all still TDR random times at the moment it was while testing less than five minutes into a game it would do this and I tried many different games as well I am lost at what to do now.

 

[MrMD]

Update to the thread I have tried using the K-boost through the app with the memory clocks to stock and nothing has changed for the card at all still TDR random times at the moment it was while testing less than five minutes into a game it would do this and I tried many different games as well I am lost at what to do now.

hmmm

Your card is WC,by an AIO i assume? Have you got any airflow over the VRMS?

Other than that i cant think of anything else to try,might be time for an rma

 

[EnjoyTroll]

hmmm

Your card is WC,by an AIO i assume? Have you got any airflow over the VRMS?

Other than that i cant think of anything else to try,might be time for an rma

Ye it has a blower style cooler fan right over the VRM's and the there is a rather large heat sink on them as well

 

[StrongSad]

Mmmmm, is there any reason your memory clock is running at 1744?

Also try running it at 1390 boost

I would again run K Boost though to see if it makes a difference.

I wonder if you are running and unstable OC...

 

[EnjoyTroll]

Mmmmm, is there any reason your memory clock is running at 1744?

Also try running it at 1390 boost

I would again run K Boost though to see if it makes a difference.

I wonder if you are running and unstable OC...

It's running that speed because the stock speeds on the memory are really unstable it tdrs seconds into a game.

I can't see it being anything to do with an unstable overclockers as I have even underclocked the gpu to 1000mhz gpu and stock 980 memory clocks and it's still crashed the same amount

 

[StrongSad]

It's running that speed because the stock speeds on the memory are really unstable it tdrs seconds into a game.

I can't see it being anything to do with an unstable overclockers as I have even underclocked the gpu to 1000mhz gpu and stock 980 memory clocks and it's still crashed the same amount

So it is you who has underclocked the memory? That is where your issue will be. You absolutely should not have to drop your memory clock below reference or more importantly out of the box clocks. Do you any Firestrike scores as heaven is not good for this fault finding...

I spent a long time with random TDR on Elite and GTA V. K boost did sort it for me as it was/is a voltage issue when it drops it clocks to idle. I would RMA with as much info as you can. Test with Firestrike, that for me was the best test for stability as it drops between tests....