All components were bought new, except for one of memory modules, which was working fine in my older system.
I have fully updated Windows. I think I have installed pretty much all necessary drivers for the motherboard.
...and the system seemed stable and ready to go!...Until I ran my first game, the Nvidia driver crashed within 1 minute of executing (error code 7 - TDR timeout). Then I tested a few other games, and all seemed to hang up or crash as soon as you enter in some sort of 3d processing. I tested OpenGL and DirectX games and all were affected.
I thought it was the GPU, so I download Furmark and ran the burn in test expecting it to crash also. To my surprise it finishes a 15min test, with the load temperature located on the high 70s Celsius.
I have already physically reinstalled the card. I also did a clean installation of the Nvidia driver, but to no avail, the problem is always the same.
I have also removed and used the different memory sticks individually.
My next step will be installing my older GPU to try to exclude a motherboard/CPU/memmory problem.
But my main question for now is if it would be common to have a faulty GPU that makes the driver crash but stands an artifical stability test like Furmark?
I read somewhere Furmark doesn't test every part of the GPU, so maybe it makes some sense that a game will use much more diverse components in the graphic card and therefore may still crash due to bad hardware. Though I would like to know the comments of someone with more experience on this.
Or if since Furmark test is good, I should turn my attention to the motherboard instead.
Thanks in advance for any input into this question which is really irritating to me.
The GPU is a full computer in its own right, so if there is a minor flaw in a ram chip or somethin like that the gpu may seem to run fine until that bit of the system is needed. It is not uncommon for a defective video card to work fine on the desktop and then artifact or glitch out as soon as you put load on it. Falures are not always temp related.
I think your on the right track and you should test the system with your old GPU. If the system works fine with the old gpu I think its safe to say the problem is the new one.
"I think I have installed pretty much all necessary drivers for the motherboard. "
have you installed the newest drivers from nvidia's webside
You mean the drivers for the graphic card? If so, yes I installed them directly from nvidia website.
Did I have to install something for the motherboard too?
I thought I had only to install the ones supplied by Asrock for the intel chipset.
At least that's what I have been doing in my previous computers.
I installed my old GPU (ATI 5770) (previously removing the nvidia drivers) and it ran everything with no crashes.
As an additional test I installed the new card, the GTX 570, on my old setup, and it takes a minute or two longer but it crashes with same error code. So I guess there is some problem with this graphic card and I will have to return it.
I just hope they (whoever decides the warranty) really try to find the problem and don't assume the card is working with no flaws and return it back to me in the same state.
Could it possibly be from mismatched ram? Since you said you had one "used" module from your old rig in that new rig of yours. It could be a slightly different speed or voltage than the other 3 and could cause issues.