Hoping for advice, computer freezing, think it's AMD GPU related

[TheBigCheese]

Hi guys

I'm having a lot of trouble with my computer at the moment and I'm not sure how to accurately troubleshoot it.

This is my setup:
i7 3770K
Corsair 32GB DDR3
Gigabyte Z77X-D3H
2 8GB AMD R9 290 Sapphire Tri-X in crossfire
EVGA SuperNova G2 1000W PSU
Windows 10 64bit

Radeon Software Version - 16.6
Radeon Software Edition - Crimson

Everything is running at stock in a large Coolermaster case with good ventilation.

I'm using it for 4K gaming and it runs really well, I've been playing Dying Light lately with everything at max (AA switched off) and it runs at a constant 60FPS (v-sync on).

HOWEVER...

Every now and then the computer will hang and will require a press of the reset button.

There doesn't seem to be a regular pattern, expect for the fact it's when I'm gaming.

I was gaming earlier for a good hour or so without issue, and then it froze.

When I've just loaded it up just now (which has spurred me to write this post) it didn't just freeze, it black screened, made a slightly strange noise, was black for a second or two and then restarted by itself which is a different crash behaviour.

There doesn't seem to be a pattern to it. When it crashed earlier I had multiple programs open in the background including Chrome, Photoshop etc., so there was a lot going on in the background but it worked for an hour without crashing.

Just now it had nothing running (having recovered from teh crash earlier) and it crashed instantly I went into the game (as in, into the map, the loading screens were OK).

I have been reading about AMD cards and there seems to be an issue with them not drawing power properly when they have eco-mode switched on. Some people recommend switching this off using MSI afterburner. I don't really like to do anything that's outside the standard recommended parameters so I thought I'd get some advice before doing that.


What do you think OCUK, any ideas what could be causing the issues?


Thanks very much!


Hope this post all makes sense, I'm rather tired while typing it and wanted to describe the issue while it's fresh in my mind.

 

[hobib_ali]

hey man did you check your temps? also why dont you take one card out and run in single set up for a few hours and see what happens? if it does then try your other card and see if its the same issue, if it freezes on both cards it may be something else,,,, try the above suggestion to make it easier to pin point the issue..

 

[TheRealDeal]

On the graphics front as above try one card at a time. If it happens on only one card you know that card is most likely your problem. If it happens with both or does not happen with both your cards are probably fine.

You could try updating to the newer 16.10.3 drivers.

As the tri-x cards are pre overclocked you could try backing off the memory clocks to stock 1250 from 1375. What you are describing happened to me when I pushed my memory a little to far. One could be on the edge even at stock Tri-x clocks.

A RAM stick may have went faulty so you could run a memory test.

If you find no faults with your hardware and drivers then I would reinstall windows. Finding niggly faults can take longer and be much more frustrating than doing a fresh install.

 

[Brizzles]

2 x 290s' will be drawing a lot of power, but your psu is more than capable of handling that.

Turning off eco mode* would be a good start, it will not damage your gpus'. ( * it will draw more power/or the power needed - it does not over volt them / void warranty )

As posted above, try 1 card for a time and see if the problem persists.

 

[Griffildur]

The fact that it only happens while gaming makes me want to check temperatures.

Check your CPU as well not just the GPUs. Maybe your cooler is not working as well as it should.

 

[r7slayer]

Have a look in event viewer see if you can see a driver crash. I would have asked if your CPU is overclocked but i noticed you said everything is stock because i would get hangs with a slightly high overclock and needed to dial it down a notch to make things stable. But stock should almost certainly not have this issue. Id check all temps like others have suggested as a starting point and event viewer. Then try one card at a time in the top slot.

 

[CAT-THE-FIFTH]

Try dropping the clocks on the GPU slightly - it might the overclock on them needs a bit more voltage now??

 

[tommybhoy]

try one card at a time in the top slot.

No need to remove and swap cards to test, just disable CrossFire and hook up the display cable exclusively to the card you want to test.

Testing the pci-e slots if gpus pass individually needs to be considered too though.

 

[TheBigCheese]

Hi guys

Thanks very much for all the thoughts, sorry it's taken me a while to reply, very hectic day, I'll try and test what you've said tomorrow.

hobib_ali - yeah, CPU doesn't go above 70ish, GPUs get to about 84 max. I actually moved the computer from under my desk to give it more air circulation so it didn't get hot air circulating under the desk, it's not really in a position to move now...not easily any way. Thankfully tommybhoy has a cunning plan for that!

TheRealDeal - good plan with the underclocking, I got the cards a while ago and hadn't considered that they could be clocked to the edge of stability. What should I use to underclock them, the AMD software or MSI Afterburner? Thanks for the tip on the RAM as well. I wouldn't think it would be that, I do a lot of Photoshop/Premiere work and frequently virtually fill up the RAM with video stuff and it doesn't fall over then, so I would guess the RAM is working correctly, not sure though!

Brizzles, thanks for the confirmation! Again, is that best done through MSI Afterburner?

Griffildur - I'm pretty sure they are in the OK range. On one of the times it crashed, I rebooted, loaded back into the game and as soon as it loaded the game (as into the playable game, not the menu) it froze again, it shouldn't have heated up in that time, but I can run CPU-Z and GPU-Z to monitor the temperatures.

r7slayer - I have looked in event viewer but I can't see anything relating to a crash, just that it says it's recovered from a crash, nothing to indicate what caused it unfortunately (which is actually in line with what others have said about AMD GPU related crashes).

CAT-THE-FIFTH - soon as I figure out how to do that! lol

tommybhoy - this is gold information! As I mentioned above, moving the computer at the moment would be a real issue so if I can test one card at a time without having to drag it out will be superb!


I'll try testing it tomorrow and let you all know how it goes!

 

[TheRealDeal]

For Under clocking the memory i would just use the AMD Crimson app but if you want to try increasing the Voltage a little to test then i would go with Afterburner. You can tweak a lot more from afterburner along with turning off power saving settings.