Hi all,

I've done some research these last days regarding DX11 crash with the NVIDIA GTX500 series cards. It all comes down to one simple thing: the GTX500 cards are underpowered - especially those that are factory overclocked. In my case I own the ASUS GeForce GTX 570 Direct CU II. This card comes with a tiny 10mhz core factory overclock. I could run Crysis 2 for about 10 seconds and then the game would black out, the NVIDIA driver (all versions of the driver) would crash and the fans on the card would increase in rpm until I killed the Crysis 2 app. This scenario is a classic OC issue. The solution is to add more voltage or you can slow down your core and shader frequencies. I decided to add more voltage. I used this guide to edit the BIOS for my card and turned up voltage from 1.0000 to 1.0250 volts. This has solved the issue and Crysis 2 is now 100% stable. I've read about other DX11 titles that crashes on GTX500 cards and the fix seems to be the same because the cards are underpowered from factory.
Simply put I used GPU-z to save the BIOS from my GTX570. Used Fermi BIOS Editor to edit voltage and save a new .rom file. Then booted from a USB key into clean DOS and typed: nvflash newrom.rom - and viola!

Another thing I can add is that you can get Crysis 2 to run great on a single GTX570 card with Vsync enabled. Download Rivatuner 2.24c from Guru3d and install. Then use the D3DOverrider app that comes with RivaTuner. D3DOverrider will enable triple buffering under DirectX - including DX11. You'll get stable fluent gameplay even at low FPS with VSync enabled. D3DOverrider still works great and it's a "must have" app.

Enjoy and remember - even though adding more voltage makes the 500-series cards more stable it's still something that you do on your own risk! To much voltage might fry your card. If you decide to change your BIOS and reflash like I have done - it's a warranty void!.

Omg dude wth are you on about...? Did you eat mushrooms or something? Did you actually go through all that **** just to increase voltage on your graphic card??
People simply put:
1. Open google and type "MSI afterburner"
2. Download it from their page
3. Install MSI afterburner
4. Go to the settings and click "unlock voltage control"
5. On main window you will see now voltage control enabled and you will see what is default voltage for your card(for example my card was at exacly 1000mV, for someone it can be 0990mV or 1020mV)
6. If you don't want to Overclock your graphic card then simply increase your voltage(by moving slider to the right) for aprox 20 or 30 mV which should look like if it's exacly on 1000, 1020 or 1030.
7. WARNING ! ! ! Increasing voltage too much and you might bust a cap on your VRM controler... Anything beyond 1150 is risky(atleast for my GPU) if we are talking about gtx 500 series. Also increasing voltage increases GPU temperature so don't up the volts too much or you might burn your card. I might sound cautious but it's really not that big of a deal. Just be careful and don't move slider too much if you're using your stock clocks. As i said move it for 20 or 30 mV.
-Peace out

I have a Gigabyte 580GTX SOC @ 855/1710/1025 and np for me...

UPDATE!

First - I know both EVGA Precision and MSI Afterburner very well. However - I prefer making changes to the BIOS instead. No mushrooms here.

Second - Not all owners of a GTX500 series card experience problems - however some do.

Status on my voltage tweak:
Gameplay time: 54 minutes (no crash what so ever)
GTX570 Core temp: 61 degress celsius
Room temp: 26 degress celsius

Mission accomplished. Must say - the SP in Crysis 2 is awesome! Lot's of action and the story is cool!. Graphics in DX11 with highres textures are the best graphics I've seen so far. Sound is perfect as well.

Finally Crysis 2 is what it should have been when first launched.

Yep... This is how crysis 2 should have been at the launch but oh well.
I got my gtx 560 ti OCed to 1GHZ with voltage at 1112 and now my GPU is running like a stock gtx 570

Talk about trying to cure the symptoms not the cause. First of all this is not an isolated Nvidia problem but an ATI one as well. Second of all the cause of this is not the lack of voltage on your GPU or too high of an overclock. I am assuming you are talking about the timeout which you can see as a display driver has stopped working and has recovered.

The major cause of this is a bad power supply or a borderline underpowered one and/or bad air circulation of the case which overheates your components or the said bad power supply. An inefficient PSU which is stressed to its limit might still be able to produce enough stable current to power your GPU till a point where its start to overheat because of the additional efficiency drop under full load and again once the heat builds up. So in order to prevent a BSOD a time out and purge command will be issued to the GPU.

At such high loads a bad power supply unit will produce a lot of noise and you could see a voltage drop on the 12V line. An increase in mV on the card might provide more stable power and an underclock will reduce the power draw. Most cheap supplies are "self" certified, rearly provide the designated power or efficiency levels, have a bad curve that is "tested" at 25°C instead of 50°C.

Keep in mind that this is generally the primary cause but not the only one and I would definitely not recommend increasing the mV on the card until you can find the source of the problem. You can read more about this and the other causes here:

http://forums.nvidia.com/index.php?showtopic=65161

lol i test my hd 6850 voltage is 1149 stock

thanks mann. i used msi after burner to do it but thanks. now i can play crysis again. muhaha

Talk about trying to cure the symptoms not the cause. First of all this is not an isolated Nvidia problem but an ATI one as well. Second of all the cause of this is not the lack of voltage on your GPU or too high of an overclock. I am assuming you are talking about the timeout which you can see as a display driver has stopped working and has recovered.

The major cause of this is a bad power supply or a borderline underpowered one and/or bad air circulation of the case which overheates your components or the said bad power supply. An inefficient PSU which is stressed to its limit might still be able to produce enough stable current to power your GPU till a point where its start to overheat because of the additional efficiency drop under full load and again once the heat builds up. So in order to prevent a BSOD a time out and purge command will be issued to the GPU.

At such high loads a bad power supply unit will produce a lot of noise and you could see a voltage drop on the 12V line. An increase in mV on the card might provide more stable power and an underclock will reduce the power draw. Most cheap supplies are "self" certified, rearly provide the designated power or efficiency levels, have a bad curve that is "tested" at 25°C instead of 50°C.

Keep in mind that this is generally the primary cause but not the only one and I would definitely not recommend increasing the mV on the card until you can find the source of the problem. You can read more about this and the other causes here:

http://forums.nvidia.com/index.php?showtopic=65161

That is nonsense. My PSU has more than enough power for my system. And....

1. Increasing voltage means more amps - which means that my PSU has to deliver more power. If your statement was true Crysis 2 would still crash - it doesn't. Actually the problem is fixed.

2. I have a 3 months old 750watt PSU that is 80+ certified. That means I am guaranteed at least 600watts when the PSU is running at 100%. According to Corsair my PSU will pump out 607,5 watts max. More than enough power to run my system when gaming. The 12volt rail delivers 60amps - more than enough for the GTX570.

3. I've seen people writing online that they have DX11 crash with their GTX570 cards and they own +1000watts PSU's.

So - stop posting nonsense and misleading people. Some might rush out and by a new PSU only to find out they just wasted their money. A new PSU will NOT solve the DX11 problem on the GTX500-series.

Well - not unless your trying to run a GTX500 card on a 200watt PSU.

thanks mann. i used msi after burner to do it but thanks. now i can play crysis again. muhaha

Happy gaming mate!

Yep... This is how crysis 2 should have been at the launch but oh well.
I got my gtx 560 ti OCed to 1GHZ with voltage at 1112 and now my GPU is running like a stock gtx 570

Don't tempt me.

I have a Gigabyte 580GTX SOC @ 855/1710/1025 and np for me...

Insane frequencies!

JCOnlineDK but a stock or factory overclocked card should not need any adjustments for it to be able to run any game. If it does then you should have it replaced because it can't run at spec. A user overclock is a different issue because not all cards (same model) can run at the exact same settings, some can handle more and some less. And again if a stock model (even with a stock OC) is causing this and you eliminate all the other causes then replace it because it is a bad model!

And take a look at the link I posted and you will notice that there are a bunch of other issues that can cause it.

Common issues that can cause a TDR:
- Bad memory
- Insufficient/problematic PSU
- Corrupt driver install
- Overheating
- Unstable overclocks (GPU or CPU)
- Incorrect MB voltages (generally NB/SB)
- Faulty graphics card
- Your asking too much of your graphics card. Not one that many people like to hear, but as the blurb from Microsoft states, if your game falls below a certain FPS and something graphically complex occurs, it could trigger a TDR.
- The issue can potentially be caused by a badly written driver or piece of software, but this is an unlikely cause in most cases.


Not just the PSU even if it is the most common one. In your case you have a very good PSU but you would be surprised how many people see "1000W" on a 50€ no name PSU and buy it.

I had the same problem with my Gigabyte 560 SoC. Underclocking corrected the issue but the cause was an borderline PSU which was unstable due to heat and efficiency loss at high load levels. The symptoms where TDR's at high load scenarios 30min+ into a game. It looked like a heat issue and I was able to run it on factory levels with a open case again. My guess was that some other components on the GPU where overheating but again this was not the real cause.

The thing is some people experience this problem and it can be hard to find out what is causing it and in my case it was masking itself as a bad GPU and overheating. But at the end of the day no single solution can be the correct one and you have to think and eliminate to solve it. If someone is dumb enough to run to the store for a PSU because of what I wrote and it does not correct it then so be it since the link contained so many bullet points about this and they did not read or think about it.

You always have an option to ask a good friend with similar gaming gear to borrow the equipment you need to test and eliminate.

Let's see if I can answer your long post.

First off - I agree with you on the fact that NVIDIA and ASUS have failed by creating a product that cannot run stock settings on stock voltage and yes - I ought to replace it. But I'm not going too. For me it's much easier to "nerd" the card and make it run perfectly. The extra 0.025 volts is not going to kill the card - worst case is that it's lifespan might a little bit shorter but that's not an issue for me. I'll have the card replaced by an upgrade before it dies.

Regarding common issues:

- Bad memory: not the case here. Then I would have had problems with Bad Company 2 and that game runs fine in DX11 even on stock voltage.
- Insufficient/problematic PSU: We can rule that one out.
- Corrupt driver install: Not the case here either. I would have symptoms in Bad Company 2 also - which I dont.
- Overheating: 61 degrees celsius it not even close to being overheating.
- Unstable overclocks (GPU or CPU): My entire system runs stock settings - including my GPU. My GPU is only factory overclocked with 10mhz - not even worth mentioning. OC is not directly the problem here.
- Incorrect MB voltages (generally NB/SB): All settings are at stock - this one can be ruled out.
- Faulty graphics card: Not the case since I would have problems with Bad Company 2 also - which I dont.
- Your asking too much of your graphics card: If a GTX570 is not enough to run Crysis 2 then god help us all.
- The issue can potentially be caused by a badly written driver or piece of software, but this is an unlikely cause in most cases: Not the case here.

I only see two options regarding DX11 crash on GTX500 series graphics card. Either it's the games themselves. DX11 crash in Crysis 2, Dirt 3 and Fear from what I've been reading online. It can also be a driver problem but since these DX11 crash happens no matter which driver version I use - I rule out driver problem.
The last option is hardware. Crysis 2 makes much more use of DX11 than Bad Company 2 does. This means that my GTX570 has to work harder when playing Crysis 2. Adding a tiny bit of core voltage has solved the issue because the extra voltage gives the GTX570 the power it really needs.

If you ask me NVIDIA and Crytek should really investigate this issue since it happens to a lot of people. If it turns out that many GTX500 series cards are actually sold with undervoltage they need to do some serious testing and release a BIOS update if the cards can run at a higher voltage without problems and premature aging.

I would have never thought this because my voltage on my 2 MSI GTX580 Twin Frozr's (in SLI) were set through 3DMark 11 benchmarks. So I am even more bewildered to see DX11 in Crysis 2 require more voltage than 3DMark 11 with all maxed settings to maintain stability. I always interpreted the crashing was due to nVida drivers.

I will try this out and see if it also balances out my temps on the two cards also.

I would have never thought this because my voltage on my 2 MSI GTX580 Twin Frozr's (in SLI) were set through 3DMark 11 benchmarks. So I am even more bewildered to see DX11 in Crysis 2 require more voltage than 3DMark 11 with all maxed settings to maintain stability. I always interpreted the crashing was due to nVida drivers.

I will try this out and see if it also balances out my temps on the two cards also.

What was the result?

thanks and thanks again for this post, solve my problem, i own giga 580GTX , and the crash with black screen is over.