Computer Randomly Shuts Down
#1
Posted 04 August 2013 - 09:15 AM
What I have noticed is that when I am playing MWO (and only MWO), my computer will just shutoff as if it is overheating. Here is my system specs:
ASUS Sabertooth x79
Intel Core i7 3820
16g of DDR3 Gskill (4x4gigs)
XFX Radeon HD 7970
Thermaltake 850w PSU
Corsair 250g SSD
WD 250g HDD
Typical performance and settings of MWO:
1440x900
Shadows: Medium
Post Processing: Low
Everything Else: Very High
Vsync: On
User.cfg Edits -
cl_FPS = 55
e_Lods = 0
Typical FPS: 45 to 60
When I first saw this problem, I just through it was MWO just flaking out. And after I initially saw it, I gave it no thought because I did not see it again for a while.
But as time passed, I began to see it more often. So, I figured it was a heat issue. So I checks for dust (which there was some but not a whole lot), monitor values from various programs, Memtest86, ect.
This is what I found:
CPU Temp: 62C Maximum
GPU Temp: 72C Maximum
Motherboard Temp: 45C Maximum
As far as I can tell, those temperatures are perfectly fine. The motherboard temp might seem a bit high, 113F, but I think that is survivable.
So I reviewed my PSU for any issues because I can't monitor it. My original PSU, Corsair 850w 80%+ Professional Series, fan was not turning. So I figured I found the problem. I replaced it with the Thermaltake 850w and played and about 4 games in, it shut down. So that isn't the problem.
I checked voltages, which looked good, but I do not modify them. I run everything stock, mainly because for reasons like this, overheating being issues that I do not want to deal with.
I ran a Memtest86+ to see if maybe a my RAM went bad (its barely a year old), but everything checked out.
Ran some stress tests for specific units, like FurMark and IntelBurnTest, to see if a specific unit is causing the problem. Of course, no issues after running the tests for a while.
Maybe I have a virus that is attached to running MWO some how, which I am currently in the process of checking. I did find some infected files but they were malware stuff.
Interestingly, which makes me point this to heat problem even with all my data I have collected, is that I lowered the resolution to 1280x960 and turn Vsync on with cl_FPS = 50, I can get more games in (GPU is rendering less frames?). If I up the resolution to 1440x900, turn off Vsync, and remove cl_FPS = 50, I get around 110 FPS average but it has shut off 100% of the time in the first game. I can hear my GPU fan speed up really fast but the values never read over 75% Fan Speed and 72C.
Really, I am at a complete loss. I am unsure what else to other than just start replacing these expensive parts, which I can not do at the moment. I am wanting to say that MWO is causing the issues because it's the only game that it does it on but it's also the only game that pushes my rig in the way it does.
Anyone have any good ideas at this point?
#2
Posted 04 August 2013 - 09:30 AM
was working fine as long as it didnt use much of it, but once RAM usage reached more than 2gb it shutdown.
also, you should run prime95, will test both CPU and RAM.
if something is overlocked, try change the clock settings.
also, try out some benchmarks like the cryengine one and see if it has the same issues.
#3
Posted 04 August 2013 - 10:21 AM
had a buddy not to long ago who had 12GB of ram and one of his sticks died.. your issue is similar to what his was... I told him check his ram using this method and found out RAM was the culprit.. this after he bought a new powersupply and almost a new motherboard... but he was playing SWTOR and not MWO..
it could also be possible that one of the DIMM slots on the motherboard is going bad so i'd cycle through those aswell if you wanted to be thorough
Edited by Bloodshed Romance, 04 August 2013 - 10:24 AM.
#4
Posted 04 August 2013 - 10:32 AM
Also, I did some more testing with FurMark, I can consistently shutdown my computer through that benchmark now. It shuts down around 78C in the GPU. I run it at 1440x900 at 8x MSAA but everything else default.
I thought maybe my fans was pulling too much wattage but I turned the fan speeds down and it didn't change anything.
Could it be possible that my GPU is causing something else to overheat? If that is the case, what else could I do because this is the only location that my GPU can fit in.
#5
Posted 04 August 2013 - 10:40 AM
instant shutdowns is usually caused by a broken psu (or doing something stupid like connecting the gpu with molex->6pin pci-e on the same 12V line as the cpu.)
If you can borrow a psu from another system that's an easy way to test it.
#6
Posted 04 August 2013 - 10:42 AM
Flapdrol, on 04 August 2013 - 10:40 AM, said:
instant shutdowns is usually caused by a broken psu (or doing something stupid like connecting the gpu with molex->6pin pci-e on the same 12V line as the cpu.)
If you can borrow a psu from another system that's an easy way to test it.
Might want to re-read my original post. I had issues with my old PSU (Corsair 850w Gold) and replaced it with a Thermaltake 850w, had the exact same issues.
#7
Posted 04 August 2013 - 10:49 AM
maybe something causes a short circuit if it heats up or something?
driver issue or bitcoin mining virus seems unlikely but could test for it anyway.
you do have the gpu connected directly to the psu cable and not with the cables that came with the card? (just checking)
#8
Posted 04 August 2013 - 11:02 AM
also you are running very hot. my i5-3570K doesnt even crack 45 degrees, and my GPU temp never goes over 70.
could be bad RAM or motherboard.
also you are running very hot. my i5-3570K doesnt even crack 45 degrees, and my GPU temp never goes over 70.
#9
Posted 04 August 2013 - 03:33 PM
Colonel Pada Vinson, on 04 August 2013 - 11:02 AM, said:
also you are running very hot. my i5-3570K doesnt even crack 45 degrees, and my GPU temp never goes over 70.
could be bad RAM or motherboard.
also you are running very hot. my i5-3570K doesnt even crack 45 degrees, and my GPU temp never goes over 70.
From what I can tell, a GPU below 80C is fine as they are rated up to 100C.
I am not sure how much more I could cool the CPU at 62C.
And I don't think it's RAM as I just did Memtest86 and it came up with no errors.
Flapdrol, on 04 August 2013 - 10:49 AM, said:
maybe something causes a short circuit if it heats up or something?
driver issue or bitcoin mining virus seems unlikely but could test for it anyway.
you do have the gpu connected directly to the psu cable and not with the cables that came with the card? (just checking)
Yes, I have the GPU connected directly to the PSU. Its a 8 Pin and 6 Pin PCI-E power cables plugged from the GPU to the PSU and is the only thing those cables are plugged up to.
But how can it be the PSU if I already swapped it out with a new one?
Edited by Zyllos, 04 August 2013 - 03:37 PM.
#10
Posted 04 August 2013 - 04:29 PM
The 4890 runs at around 72C but does not shutoff doing the same FurMark tests as the 7970.
What exactly does that tell me? When doing the tests, the 7970 seems to just keep heating and heating up (to around 78C is usually when it shuts down). Is there any way to pull back the voltage and underclock the videocard so that it runs cooler?
Edited by Zyllos, 04 August 2013 - 04:34 PM.
#11
Posted 04 August 2013 - 06:11 PM
#12
Posted 04 August 2013 - 06:19 PM
also make sure your gpu fans are running and speed up with the heat rising. if not get a program like msi afterbuner (for the GPU)and enable the automatic user defined fan controll. me i usualy have it set to 30%@30C and 100%@65C. since i did that my temps have never gone over 65C since. and im running everything on max settings @1080 and geting 60-70 fps with my asus 660ti. even when its around 80F in my room(no airconditioning)
also how well dose your case ventelate air? i got rid of the stock fans in mine and swapped em out for 2 140 85CFM fans 1 in front and 1 on top. and 5 120MM 75CFM fans 1 in front 2 on side and 2 on my liquid cooler in push pull. my temps droped over 10C from were they were with the stock case fans.
Edited by ICUBurn, 04 August 2013 - 06:23 PM.
#13
Posted 05 August 2013 - 06:33 AM
StainlessSR, on 04 August 2013 - 06:11 PM, said:
As far as I know, it's still under warranty so I did do an RMA. But the issue is that I have had this thing working fine for almost a year before it happened. That is what I don't understand.
ICUBurn, on 04 August 2013 - 06:19 PM, said:
I think it ventilates fine, but it's all fans/air cooling. I have some really big fans on the top and sides and two smaller but faster fans in the front and back. The CPU has an aftermarket fan while the GPU is the XFX DD 7970 stock cooler. XFX has even said the card is suppose to run at approximately 70 to 75C. They said the card should not have issues until it reaches 95C.
The thing is that the card, when it shuts down, is still heating up pretty fast (based on the curve seen no the graph in FurMark) when it shuts down. The fans are at ~80% when it hits ~70C but I could bump it up to 100% at 70C to see if it helps at all.
Only thing I know what to do is go ahead with the RMA and see what they find out.
#15
Posted 05 August 2013 - 07:07 AM
Bloodshed Romance, on 05 August 2013 - 06:56 AM, said:
yes, if one is indeed bad
Yes, that having all four sticks in would catch it, or yes, you have to do each stick individually to find the error?
It seems like a flawed program if there a different outcome just dependent if you have sticks in there or not. Either way, I am almost certain this is a GPU problem instead of a RAM problem because of swapping out the GPU for an old one and it does not shutdown.
#16
Posted 05 August 2013 - 07:36 AM
#17
Posted 05 August 2013 - 07:54 AM
MrPils, on 05 August 2013 - 07:36 AM, said:
Ok, if that is the case, could I not just open the side of the case and blast a desk fan into it?
I wish I took a picture but that 7970 GPU sits right on top of my Southbridge heatsinks and the XFX 7970 has a metal sleeve, which does get hot along with the card. So I was thinking that maybe the thing is transfering heat between the GPU to the Southbridge but how can I tell?
If those numbers seem fine for the GPU, then maybe it's the motherboard. But the two PCI-E 16x slots leaves the GPU sitting right on top of the Southbridge heatsink/fan so I am unsure what else do.
But if I knew the motherboard was the cause of the problem, not the GPU, that would save me $100's in buying a replacement.
#18
Posted 05 August 2013 - 08:04 AM
#19
Posted 05 August 2013 - 08:05 AM
and about mem test... its a long explanation but to put it simple.. it has a better chance of marking a Ram Stick as a fail if you do them one at a time.. I had to go through a machine that kept dying because of ram that had 8 sticks.. one of my engineer friends said do them one at a time.. I was pissed because its a grueling process but sure enough.. it got one bad doing them one at a time.. but if you've narrowed it down to GPU then great, because XFX support has been kind to me thus far.. only had to contact them twice... my 6770 fan went out and they send me a fan off of a dead card... and my 7970 GHZ which went pop...
Edited by Bloodshed Romance, 05 August 2013 - 08:08 AM.
#20
Posted 05 August 2013 - 08:17 AM
Bloodshed Romance, on 05 August 2013 - 08:05 AM, said:
and about mem test... its a long explanation but to put it simple.. it has a better chance of marking a Ram Stick as a fail if you do them one at a time.. I had to go through a machine that kept dying because of ram that had 8 sticks.. one of my engineer friends said do them one at a time.. I was pissed because its a grueling process but sure enough.. it got one bad doing them one at a time.. but if you've narrowed it down to GPU then great, because XFX support has been kind to me thus far.. only had to contact them twice... my 6770 fan went out and they send me a fan off of a dead card... and my 7970 GHZ which went pop...
Well, I have narrowed it down to "If I remove the 7970 and put a 4980 in it's place, doing the same test, it does not fail". This is not enough information to make an informed decision as to either the 7970 or X79 motherboard is the issue.
But if you replace the GPU with a new one, and do the same test and it does not fail, that makes me not point to the RAM, but maybe I am wrong on that subject.
Edited by Zyllos, 05 August 2013 - 08:18 AM.
1 user(s) are reading this topic
0 members, 1 guests, 0 anonymous users