When gaming, artifacting occurs followed soon by a complete crash

[Random_her0]

Hi all,

 

Hopefully someone can either confirm my suspicions or point me in another direction as to what could be the issue.

 

First, my specs:

CPU: Core i5- 4670K (not OCd)

RAM: 16GB Kingston HyperX @1600MHz

MoBo: MSi Z97M Gaming with latest BIOS

Graphics: 2x R9-290s in Crossfire with latest drivers (not OCd)

PSU: Corsair HX1000

Windows 7 Home Premium x64 with SP1

 

Basically, when using my PC for regular things (watching movies, browsing the net etc) everything runs swimmingly. However, when I start a game artifacting will begin to occur after 5-10 mins (depending on the game) and then the whole PC will crash within 5 seconds.

Some things to note:

• Disabling crossfire stops this from occurring 
• locking framerate either prevents the issue on some games OR delays it on others
• Have run memtest - no issues
• Have tried many different drivers and reinstalling Windows

What do I think is wrong with it? I think it's the PSU beginning to die, and when the power draw gets too high it croaks. However, Corsair makes incredibly reliable PSUs and at 5 years old, I figured it may still have a few good years left.

Any thoughts?

 

Thanks!

[Constantin]

From what i know artifacts are from GPU's.....

Maybe it is coming from the PCI ports... Have u OC bus speed?

[Random_her0]

From what i know artifacts are from GPU's.....

Maybe it is coming from the PCI ports... Have u OC bus speed?

 

Absolutely everything is currently running at stock.

And yes, I normally relate artifacting to GPU, but as a result of power issues (e.g being pushed too far)

[Constantin]

Absolutely everything is currently running at stock.

And yes, I normally relate artifacting to GPU, but as a result of power issues (e.g being pushed too far)

Your PSU is more than enough (1ooow), but maybe better check it

[Random_her0]

Your PSU is more than enough (1ooow), but maybe better check it

 

Yep, not concerned with the wattage, just the age. I've had it since 2010 in regular use. Will wait for some more feedback from others though

[Samsterstorm]

I would contact corsair about a RMA on that PSU, and if they wont then buy a new one and see if that has an effect! If that has no effect then try and see which card causes issues, or just return both of them. It has to be one of those two, or ur PCIE speeds! 

[Random_her0]

I would contact corsair about a RMA on that PSU, and if they wont then buy a new one and see if that has an effect! If that has no effect then try and see which card causes issues, or just return both of them. It has to be one of those two, or ur PCIE speeds! 

 

Pretty sure everything is out of warranty. 5 years on the PSU, both the graphics cards have custom water blocks on them (and worked perfectly in the previous build).

What do you mean about the PCI speeds? hasn't 8x/8x been proven to be more than sufficient?

[Dinodaur]

First I would try to uninstall the drivers and then reinstall, see if that leads to anything.

Then stress test both GPUs using software of your choice, have crossfire enabled and make sure you are testing both at once. You may not want to do this if you are concerned it really is the PSU and it can't handle both cards, it could perhaps ruin other parts, though I am not 100% sure. Especially so if the PSU has gone wrong after this many years.

If that doesn't work try using one card at once, i.e. completely removing the other card from the system and then run on just a single card. If it only occurs with one card then your problem is with the actually cards, if not then it may be the PSU.

If you can try a different PSU if you have one lying around, or can get one easily.

[Random_her0]

First I would try to uninstall the drivers and then reinstall, see if that leads to anything.

Then stress test both GPUs using software of your choice, have crossfire enabled and make sure you are testing both at once. You may not want to do this if you are concerned it really is the PSU and it can't handle both cards, it could perhaps ruin other parts, though I am not 100% sure. Especially so if the PSU has gone wrong after this many years.

If that doesn't work try using one card at once, i.e. completely removing the other card from the system and then run on just a single card. If it only occurs with one card then your problem is with the actually cards, if not then it may be the PSU.

If you can try a different PSU if you have one lying around, or can get one easily.

 

Thanks for the advice. Unfortunately, as I mentioned before the system is watercooled (with acrylic) so swapping cards in and out isn't the easiest endeavour. I've tried a variety of drivers and reinstalled Windows. Same issue. Also, nothing higher than a 750W lying around, and I don't think that's enough to really test out the system.

 

I think I may have made a diagnostic breakthrough of sorts. Three times in a row now the artifacting has begun when GPU 1 hits 60-degrees. I've set Overdrive to have a much higher thermal target than that, so that isn't causing it.

Any ideas? What would lead to such instability upon hitting only 60?

[Random_her0]

Update:

Some more testing was done. This time I limited the framerate to 55 using CCC and set all details to low (using Unigine Heaven). This resulted in temps not exceeding 52-degrees, and not crashing once. 
It definitely seems like there is some sort of thermal issue.

[Dinodaur]

Update:

Some more testing was done. This time I limited the framerate to 55 using CCC and set all details to low (using Unigine Heaven). This resulted in temps not exceeding 52-degrees, and not crashing once. 

It definitely seems like there is some sort of thermal issue.

Seems odd that at 60 the card crashes, instead of 90+. What do you use to get the temps? Could be a dodgy programme and really the temps are a lot higher (though you should be able to tell by being close to the PC). The temp sensors could also be broken. Any idea what the temps where like before you started having issues?

You could also try removing the power from GPU 1 and try with just GPU 2 but I don't know if your mobo will get angry because you have a card with no power. I think you could also disable it using windows device manager even when powered.

[Random_her0]

Seems odd that at 60 the card crashes, instead of 90+. What do you use to get the temps? Could be a dodgy programme and really the temps are a lot higher (though you should be able to tell by being close to the PC). The temp sensors could also be broken. Any idea what the temps where like before you started having issues?

You could also try removing the power from GPU 1 and try with just GPU 2 but I don't know if your mobo will get angry because you have a card with no power. I think you could also disable it using windows device manager even when powered.

 

I'm using MSi Afterburner for all monitoring, and I'm confident the temps are correct. I'll try unplugging one of the GPUs and testing again, but I suspect you're right and the MoBo will throw a hissy fit

[Dinodaur]

I'm using MSi Afterburner for all monitoring, and I'm confident the temps are correct. I'll try unplugging one of the GPUs and testing again, but I suspect you're right and the MoBo will throw a hissy fit

I would still try another piece of software just in case, speed fan springs to mind as something pretty good.

 

EDIT: Link to website for SpeedFan: http://www.almico.com/speedfan.php

[Random_her0]

I would still try another piece of software just in case, speed fan springs to mind as something pretty good.

EDIT: Link to website for SpeedFan: http://www.almico.com/speedfan.php

Thanks, will give that a go!

Testing the GPUs individually I think I've noticed something. It isn't the first GPU causing the crash at 60-degrees, I believe it's actually the second GPU at around 54-degrees which causes the crash.

I'll stress test the first GPU further but it seems, at least at this point, that there is an issue with the second GPU.

[Random_her0]

Ok, beginning to narrow it down.

When running in Crossfire, the system will crash when GPU one is at ~61-degrees and GPU two is at ~55-degrees

If only GPU one is enabled, there are no crashes, even when pushing the GPU to~63-degrees using FurMark

When only GPU two is enabled, the system crashes at ~55-degrees.

 

The issue? GPU two. Why? Still no clue. Why would it work perfectly up until that mid-50 point, when total system failure occurs? I'll try take everything apart tomorrow and see if I can see something. Perhaps an o-ring has slipped or something. The backplate feels pretty darned hot, so I suspect that the graphics memory temps are significantly higher than the seemingly fine GPU temp