GTX 560 Point of View. Games crash with drivers newer than 347.88.

Hello there! Obviously, the problem is outdated and no longer relevant, but I would like to bring this matter to the end. As the title of the topic says, there is some kind of problem with the driver.

In 2015, I was faced with the fact that after updating to the new driver 350.05 at that time, absolutely all games started to crash after 1-2 minutes of gameplay. So I solved it by simply rolling back to the previous version of the 347.88 driver and stayed on it until buying a new GTX 770, and then GTX 970, which I remain with to this day. Since then, I forgot about the problem, just putting the graphics card in the closet. But now, in an era when every graphics card is a value, I would like to finally deal with this problem so that I have a fully working graphics card in reserve. Sort of like a sporting interest.

Now on to the point. After pulling the graphics card out of the closet, I replaced the thermal paste on it (only between the heatsink and the heat-distributing cover). The temperatures are like this: the idle temperature is 33 °C, in the most difficult scenes in games it reaches 75 °C, FurMark warms it up to 84 °C. Everything is within the normal range, I think.

Since I couldn't immediately remember which latest driver was stable before the broken one, I had to go through quite a few different drivers, including the latest one for my graphics card, so I made a table with testing data. Perhaps this will be useful.



As can be seen from the table, the last adequately working driver is 347.88. When I discovered this, I remembered that this driver was exactly the one that worked 7 years ago. It was released on 17/03/2015, exactly a month before the super popular title GTA 5, which may lead to conspiracy theories that NVIDIA specifically kills old cards with new drivers. I won't say if it's true or not, but in my case it seems to be true, haha.

At some point, I discovered that if one game works, then all the others work, and vice versa, if one does not work, then all the others do not work either. The exception is Battlefield 4, it does not work properly on any driver. It only works if you either manually lower the GPU frequencies from 810 MHz to 405 MHz, or the frequencies themselves will drop after the crash (the graphics card does not switch from power-saving mode to performance mode). I didn’t manually set the frequency, I just limited the frame rate to 10 frames per second so that the graphics card worked in power-saving mode, and the crash occurs if the frame rate is raised to 15 and the card goes into performance mode. I tried to manually lower the GPU frequency to 770 and 700 MHz, the video memory frequency and voltage, but it didn't help, apparently, there is some problem in the transition between power-saving mode and perfromance mode. Temperatures, of course, do not affect anything, so it's not about them clearly. I reflected this behavior for the 340.52 driver, but the behavior pattern for BF4 is the same on any other driver.

Battlefield 4 always crashes with this error and this is the only game that crashes with some error.



The FurMark test is passed on both stable and unstable drivers, but after the first crash of BF4 it does not start and crashes with this error.



Does anyone have any idea how to deal with this? Or do you need some other information? I could write somewhere else to NVIDIA, but I would not hope that "they will fix it in the next driver" because the final driver for GTX 560 was 391.35, which was released on 27/03/2018.

 

Normally if they say its final driver would mean its an EOL product that's no longer supported, might help looking at a more recent gpu if possible. Probably not what you want to hear only time I've seen a driver released for a EOL product would be if there was a major security issue that would warrant it. Only thing you could try resetting all over clocks and try a fresh install of windows.

 

I might even want to sell the graphics card so the card doesn't take up space, but I would like to ensure the card is fully stable so I can be sure I'm delivering a fully functional product. It seems unfair to me that after 347.88 driver drivers were released for another 3 years and for some reason they do not work.

The card has always been used at the stock frequency of 810 Mhz. For 7 years, I have already reinstalled Windows completely several times. Now I'm on Windows 10, when the problem happened I was on Windows 8.1. It definitely won't help.

 

Could you try underclocking your card's core and memory by -50Mhz or -100Mhz and then try? Perhaps the newer drivers are more sensitive towards the frequency and your old hardware might be worn out? I've seen such cases where the cards are no longer capable to operate at stock frequencies anymore and need a little underclocking.

 

hardware is bad.

 

Did you read his post? He stated he already tried that.

Probably.

 

In fact, his comment still makes sense, I thought about it and did more tests.

When I said that I lowered the frequency to 770 and 700 MHz and it had no effect, I was a little mistaken. At that time, I only tested Battlefield 4 and only on unstable drivers. On unstable 391.35 driver frequencies of 770 and 700 do not really have any effect, but lowering the frequency to 600 MHz, then even Battlefield 4 starts working. Other games run at a higher frequency of 700 MHz.

The phenomenon you describe most likely exists, but is my graphics card really so worn out that in terms of new drivers, I need to lower the core frequency by 210 Mhz for BF4 to work and 110 Mhz from the stock for all other games to work? 26% and 13.5% of the total performance is somehow too much and, in principle, all this is somehow too strange. I understand another 10, 20, 50 MHz, but 210 and 110 are somehow toooo much. So sensitive new driver, ha...

On the stable 347.88 driver, Battlefield 4 ran at 670 MHz, minus 140 MHz from stock. Also a lot, but this Battlefield is some kind of exceptional bullshit, it's too weird. It looks like EA is working closely with NVIDIA on a program to destroy old cards. The game was released in 2013, but has been actively updated over the years, even the minimum required driver for it is 340.52 from 29/07/2014, which says something. But this is all, of course, a joke (or not).

Maybe so, but on 347.88 everything else works fine, except for this damn Battlefield 4.

 

If specific game issues is something you observe then it could simply be an nvidia driver issue with that game, nothing can be done about that. We see such issues till date even with newer drivers and hardware.

About the hardware being worn out, test using a consistent and repeatable scenario on whatever drivers you're testing. I mean an application specifically used for stress testing like furmark, heaven. Timespy is also something which is sensitive to such hardware issues.

If you pass such tests, which you should if the hardware is ok then proceed with testing games. If not, then hardware is bad.

 

I only tested FurMark, perhaps I need to test the ones you mentioned as well, I think it will be useful. FurMark I tested up to a maximum of 15 minutes, because it hurt me to see 84 °C and I turned it off. And I didn’t see the point to test more time, because all the games crashed before 5 minutes of gameplay. Actually it really surprised me that at 391.35 BF4 crashes in 1-2 minutes, and FurMark, in turn, burns my card for 15 minutes in a row, as if nothing is happening. That's why I still think that the problem is in the driver, and not in the hardware.

 

Try TimeSpy instead. That benchmark will be more sensitive to instability than Furmark. It will be close to BattleField. Do this with the latest available driver and find your stable clocks. But I guess your card is pretty much a paperweight at this point.

 

test 392.68

 

I used to have an Asus GTX 560 Ti. It stopped accepting new driver versions at one point, though technically driver support was supposed to continue long after that arbitrary point. I kept trying new drivers occasionally for a while, but eventually gave up and stuck to the last working driver until I bought a new card. If there was a firmware update for the card, I never checked, though.

I guess things like that just happen.

 

Sorry for missing. I have tried Time Spy, Fire Strike and Heaven. On 347.88 Fire Strike and Heaven run absolutely fine, but Time Spy didn't start because the driver is outdated. At 391.35 Fire Strike is running fine, Time Spy has already started and also completed without incident, but Heaven shows instability. At 810 MHz, at some points in the test, the image starts to turn off for 2-5 seconds, but then it turns on and the test continues, reaching the end. Sometimes, at the moments of such shutdowns, after turning on the image, the frequency is reset to 405 MHz. As the core frequency decreases, overall stability increases. At 650 MHz, the image briefly blinked only once, but in general the test was passed to the end. Probably, at 600 MHz everything will be without incident at all. But it's still not what I'm looking for, working on 600-650 MHz on 391.35 is not at all what I need.

After 391.35 comes immediately 397.31. 392.68 is a driver for the Quadro series, I don't know how you managed to get it for the GTX 570.

I thought about updating the BIOS, but according to techpowerup, there is only one BIOS version 70.24.21.00.02 available for the GTX 560 POV. Although GPU-Z says I have BIOS version 70.24.25.00.02, which seems to be an even newer version.

Spoiler

https://www.techpowerup.com/vgabios...l=GTX+560&interface=&memType=&memSize=&since=

 

If gradually decreasing your clock speeds below the defaults reproducibly and significantly increases stability then your hardware is obviously faulty. You can either sell is that: faulty. Or you can keep it as a backup card and use it at low clocks (the highest working ones you iterate out by trial-and-error testing), however low those clocks might be.

 

All this is true, of course, but generally applicable only for the latest drivers. Of course, it would be possible to continue using the 347.88 driver, but I'm worried about non-guaranteed support for games that came out after this driver for the next three years. I would like to be able to play them, although not at the highest settings and with low performance. This upsets me, and I thought that this situation could be somehow fixed, since it seems to me that the problem is software in the driver side. I also allow the possibility of faulty hardware, but to a lesser extent, since the 347.88 driver shows that the graphics card can work stably.

 

I had an EVGA GTX 770 SC and it had the same error on BF4. I could only play this game if I underclocked the GPU.

 

disable "security boot" in bios UEFI, if you using
restart

run command in dos prompt with administrative privilegy

bcdedit -set loadoptions DISABLE_INTEGRITY_CHECKS

bcdedit -set TESTSIGNING ON



restart pc

show test mode in right screen



if show perfect


download 392.68 for windows 10 with inf modded all geforces are supported gtx 460 to 1080ti
https://www.mediafire.com/file/f4t5...ebook-win10-64bit-international-whql.rar/file

extract and install run setup.exe


result your old gpu with new quadro driver work perfectly



support directx 12

 

there is no need for any of this,

you can just create a new catalog file and sign it, as long as the migration reg key is set you can self sign drivers so long as your self created certificate is installed into the root authorities.

 

I have no idea how to create the original installation file, so I don't need the test mode, if you can do a step-by-step tutorial for us, we'd appreciate it

 

Thanks, I completed the installation following your instructions, but unfortunately the 392.68 driver behaves the same way as 391.35 in terms of stability.

I said that I tried to reduce the core frequency and this led to some success, but for the normal operation of the graphics card this was unacceptable due to the frequencies being too low to 700-600 MHz. Now, on the contrary, I tried to increase the voltage from the standard 1.000v to 1.037v. As enthusiast29 said, since the driver after 347.88, the drivers have become much more sensitive towards to voltage and frequency, which began to lead to instability on my graphics card.

On the 391.35 driver, other games worked at 1.025v and a stock 810 MHz frequency, and the capricious Battlefield 4 worked fine already at 1.037v and the card even overclocked to 860 MHz.

347.88, as always, showed much more stability and surprising overclocking results. At this driver all games already worked at stock 1.000v and 810 MHz, but to my surprise, BF4 started to show off, although 7 years ago I thought that this driver was stable without exceptions. But even this crap worked already at a reduced voltage of 1.025v compared to 391.35. And the card achieved amazing overclocking results: at 1.025v, the graphics card reached 890 MHz!

I have reflected all this in this slightly complicated chart:


Apparently, with this thread I wanted to force the 391.35 driver not to be so unreasonably sensitive to voltage and frequency compared to 347.88, but it seems that we will not succeed. But I still made at least some progress, which is not bad. Probably, plus three degrees to the chip temperature at 1.037v 810 MHz mode compared to 1.000v 810 MHz is not so bad, I hope in this mode, the card will not experience big problems, although lower temperatures would be desirable, but this can only be achieved by scalping the chip, which is probably not worth the effort.