Computer crashing after a few minutes of gaming

[ninjames]

I bought a new build a few weeks ago and things have been working great up until today. I have an i7-7700k and a 1080TI. This is my build: https://pcpartpicker.com/list/kBQY7h

I've been gaming at 4K/60 since I put it together, but today, I started getting crashes after a few minutes of gaming. The crashes are constant and when I check the temperatures prior to crashing, it's in the 65C range, which with a 1080TI, that seems normal. Those are the temperatures for the GPU btw. It seems like it shuts off completely at or near 70C every time like clockwork, and works fine up until that point. I tried this with Dark Souls III and Forza Horizon 3 and both times led to crashes (this was after I played around 2 hours of Dark Souls III at 4K this morning without issue). I've crashed around 10 times now.

I'm running Windows 10 64-bit. I updated my graphics driver (the last update was more than a week ago), Windows is updated, I updated my SSD firmware and flashed the latest BIOS. Still having the issue.

How can I determine what is actually causing the crash? I have no idea if it's temperature related or not, and by the time the computer boots back up the temps have dropped significantly so I can't tell precisely where it shut off.

 

[Yamitime]

Hey James . Do ya fancy going into your bios and checking what the voltage is set for that memory please. It should be set to 1.35v .I know your using xmp profile but its worth a look.
I'm so sorry nobody has found a solution yet . Must be so frustrating .I would have thrown it across the room already .

 

[ninjames]

Hey James . Do ya fancy going into your bios and checking what the voltage is set for that memory please. It should be set to 1.35v .I know your using xmp profile but its worth a look.
I'm so sorry nobody has found a solution yet . Must be so frustrating .I would have thrown it across the room already .

It says Profile1: DDR4 3000MHz 15-17-17-5 1.350V. Then up top it says DRAM Voltage 1.376V -- is that in practice wrong or is the 1.35V an approximation? Here is a picture I took of the BIOS: http://i.imgur.com/ZpAZAyJ.jpg

EDIT: And honestly, thanks for all the help, even if I haven't figured out what's wrong yet, I appreciate the time you and scout_03 have put into helping me.

 

[Yamitime]

Memory is ok. After trawling the web trying to find similar problems with your components I did find an interesting thread on your Graphics Card.

There is a problem with the cards reaching 70 degrees .This is some of the text from a guy who found a workaround but the whole thread contains clues that point to the Card being the culprit.


Also, I have noticed that when I set a custom OC profile in the Aorus Engine the GPU boost clock does not match GPU Z. HOWEVER, when I apply the default OC Mode that comes with the card and THEN apply my custom OC profile, clock speeds exactly match GPU Z. In addition, I have a higher boost. I max at 1987 mhz when applying a custom OC profile at max power (no voltage tweak), but when I first apply OC mode and then afterward apply my custom OC profile, my boost stabilizes at 2025 mhz at load in games (BF1, Overwatch, etc.). My default OC profile right now is only +112 on the power slider (of 125, F3 Bios) with a modest OC of +53 mhz (1773 Boost clock per GPUZ and Aorus software).

So, my problems with the card were entirely resolved by a) making sure that the voltage curve was not set to custom and b) always applying OC Mode before applying my custom OC profile. Now, I never have any issues, the card generally runs at 60% to 80% TDP while gaming, and temps with a moderately aggressive fan curve hover between 60 to 65 C in game. I'm satisfied with the results. I would like to push my card higher in the future, but I'm just extremely relieved that I didn't have to RMA the card. I initially thought I got a dud, but I suspect there are many other Aorus users who actually have decent if not great cards but are suffering instability entirely due to similar software bugs.


The full thread is http://www.overclock.net/t/1627238/gigabyte-aorus-geforce-gtx-1080-ti-owners-thread/550
I know its not much but it does point to the same sort of things happening to your system.

 

[Yamitime]

If your running G Sync could you try turning it off and testing

 

[ninjames]

If your running G Sync could you try turning it off and testing

So, a couple things -- with the post linked above, I'm not sure how to do that stuff. I don't OC, but I just downloaded the Aorus software and adjusted the Boost Clock, and it immediately switched in GPU-Z when I did that. But now that I see there are so many weird quirks with this card, I'm tempted to just return it and try a different card. Unless you think trying to replicate the circumstances in that first post are worth doing.

And I'm not running G-Sync as far as I know. I don't have a G-Sync monitor/TV.

 

[Yamitime]

You do what you feel is best . Its pretty likely the fault is the Graphics.

 

[ninjames]

You do what you feel is best . Its pretty likely the fault is the Graphics.

Well, the GPU is officially back in the mail on its way to Newegg. I convinced them to give me a refund for credit so now I have to determine which GPU I'll be buying as a replacement. Looking at the EVGA FTW3 ICX 1080Ti.

 

[scout_03]

the evga draw 280 watts so chek if your 650 watts psu could handle it

 

[ninjames]

the evga draw 280 watts so chek if your 650 watts psu could handle it

PCPartPicker shows an estimated 460W with the FTW3 in my build

 

[Yamitime]

The psu has 20a per pcie cable . Theres 4 port for pcie on the psu but 2 ports share same power .So he would have to plug pcie into vga 1 and vga4 to get full 20amp from each . 20amp + 20amp = 40amp and multiplied by 12 v would be 480w available .If the evga uses 280 that leads 200w free on the 12v rails.
I think if the psu is working ok it should be ok for the card.

 

[ninjames]

The psu has 20a per pcie cable . Theres 4 port for pcie on the psu but 2 ports share same power .So he would have to plug pcie into vga 1 and vga4 to get full 20amp from each . 20amp + 20amp = 40amp and multiplied by 12 v would be 480w available .If the evga uses 280 that leads 200w free on the 12v rails.
I think if the psu is working ok it should be ok for the card.

I'm not actually sure what most of that means. This is a picture of the back of my power supply:

https://images10.newegg.com/NeweggImage/ProductImageCompressAll1280/17-438-054-14.jpg

I see VGA1 and VGA2, which is what I plugged my two cables in for my previous card. Is that not what I'll be doing for the new one? I don't see a VGA4 or anything like that.

EDIT: Should I try and return this and buy the 750W or 850W which have VGA1, VGA2, VGA3 and VGA4?

 

[scout_03]

since ther are 8 pins one make shure they are well plug in sometimes the little 2 pins parts it not fully inserted in the socket on gpu or psu .

 

[Yamitime]

The psu has 20a per pcie cable . Theres 4 port for pcie on the psu but 2 ports share same power .So he would have to plug pcie into vga 1 and vga4 to get full 20amp from each . 20amp + 20amp = 40amp and multiplied by 12 v would be 480w available .If the evga uses 280 that leads 200w free on the 12v rails.
I think if the psu is working ok it should be ok for the card.

I'm not actually sure what most of that means. This is a picture of the back of my power supply:

https://images10.newegg.com/NeweggImage/ProductImageCompressAll1280/17-438-054-14.jpg

I see VGA1 and VGA2, which is what I plugged my two cables in for my previous card. Is that not what I'll be doing for the new one? I don't see a VGA4 or anything like that.

EDIT: Should I try and return this and buy the 750W or 850W which have VGA1, VGA2, VGA3 and VGA4?

That picture is a EVGA supanova g2 . The one in your partpicker list is a supernova nex 650 .

The PSU you have (supernova g2) is actually pretty good 54amp on dedicated 12 volt rail which is plenty for the card.