PC shuts off within minutes of loading a game

magoo

[H]F Junkie
Joined
Oct 21, 2004
Messages
14,554
I have what I think is a bad GPU?

Just a couple days ago I would notice my PC simply shut down during gaming.
It did this in Mad Max, Dying Light and Metro LastLight. It's like someone just flipped an off switch, no advanced warning.

I have a very stable X79 system at 4.5 GHz and a 1200 Watt Corsair AX PSU.
Currently running two GTX 980 Ti in SLI OC to run a game frequency of 1300.
The GPUs are watercooled, never go over 45C and are not overvolted.
Running the latest NVidia drivers.

CPU is AIDA 64 stable.
Memory is 100% running MemTest Pro.
I even used another PSU to power one GPU to rule out PSU problems.

I tested each GPU when I bought them and individually they were rock solid in Heaven.
I have even reduced the clocks to stock, same issue.
Currently the same issue is seen in Heaven about 60-90 seconds into the run.
My problem is how to identify the bad GPU with them watercooled.
My MB doesn't have jumpers to shut down one card.

I think I know the answer.....I'm going to have to tear the system down.....:(

Am I correct in thinking this is a GPU?
I can't isolate any other component and the shutdown only happens when I go to run a game or Heaven.
 
Can't you just disable SLI and test with the primary GPU? If that fails, you know that one is bad.
 
Last edited:
well it's a definitively a power issue,

1. are you using UPS, Surge Protector or Current Stabilizer?. if not continue to next, if yes how much VA? Surge protectors, stabilizers and UPS have a certain capacity if you surpass it they will shut down everything plugged. so check it if you are using one, with that system I would use at least at 2KVA Protector.

2. are you using dedicated PCI-E cable per connector? sometimes running those dual connectors cables tend to produce this kind of troubles with high-ends GPUs specially power hungry like the 980TI.


3.running out of power. 980TIs are power hungry cards and if we add Gigabyte, power is just even more due the high factory OC.. if you have a lot of Fans, HDD, SSD, Lights and so on I think you can run out of power specially with that OC'd 3930K which is another power hungry behemoth added to the formula.

4. Faulty PSU, yes you have a very high quality one, but how old is? and even new PSUs can just be faulty..

your cards are fine. your system it's fine. the issue is related to power.
 
Thanks for the suggestions.
Im going to try and isolate the power in, incase its my circuit protection.

Im not convinced its the gpus due to the abrupt cut off.

My psu is 5 years old.....maybe I need a new one?
 
test can be relatively easy, just connect your machine directly to the power outlet and do some gaming again. no shutdown? protection faulty.

In the other hand the PSU isn't that badly old but if has been heavily used Capacitor aging + Wear can't be a good formula mix. if you have a spare high wattage PSU I would go with that as second option to the above.
 
Since he has multiple GPUs, it should be easy to diagnose if it is the GPUs or the PSU.
 
Well, I used a different surge protector,outlet and placed my Kill-o-Watt in the line as well. nothing else on that circuit.

Idle watts @ 100 and in game 840 under full load.

While the computer performed flawlessly, it did shut off again, about 10 minutes or so into the testing.
I restarted and ran again, this time shut off sooner....maybe 5 minutes.
Restarted and the thing shut off at the loading screen of Mad Max.

Now my backup computer runs on the same basic circuit/surge protector and has no problems......

I think the circuits are OK.

I think this points to the PSU.....what do you all think?

I'm looking at RMA and in the meantime buying a Seasonic X-1250.
 
Since he has multiple GPUs, it should be easy to diagnose if it is the GPUs or the PSU.

Not really......my GPUs are watercooled and I have no motherboard jumpers to shut one GPU off and test the other one.

I might be able to test the master GPU by disabling SLi and running one monitor on the master.......but that doesn't say anything about the bottom card.:(
 
Seems like an issue with PSU.

Remove both cards and test them individually. If they're both OK, then it's time to RMA your PSU.
 
Well yeah that behavior seems to point to the PSU since it is the overload protection going off.

Surge protection is instantaneous and usually it is set to around 5+ times your rated max (well this depends on the type of machine but 5+ was what i remember of the top of my head). This is usually handled with a fuse.

Overload protection is usually set at whatever it was rated +10% and is a thermic based protection, usually a metal sheet that will expand after some time at the overload current, the fact that in successive attempts the shutdown happened faster is because the initial temperature of the metal sheet was higher each time. Since this is a mechanical effect it tends to give out after some time.

edit to add:
Forgot to finish the explanation, the metal sheet pushes the connector away once it expands enough, kinda like a switch.
 
Thanks guys.
Corsair RMA in progress.

To fill the gap in time ordered a Corsair HXi 1200 new from Amazon.

I'll either keep the new one or sell it if I get a 1200 AXi on RMA from Corsair.

I appreciate all the suggestions. The symptoms seem like the old AX 1200 just finally gave up.
 
I think you can unplug the power to one card at a time and test them that way. You will have to switch your monitor cable to the bottom card when you test it, but it should work that way.

I saw a similar issue happen once to a computer. The issue with it was caused by a heatsink on the mobo that was loose and not making good contact. Browsing the net the pc worked fine, but once he fired up anything that put some stress on the board it would cut off within a few minutes. Then it would cut out again faster each time. So that's just something you may want to check also.
 
I think you can unplug the power to one card at a time and test them that way. You will have to switch your monitor cable to the bottom card when you test it, but it should work that way.

I saw a similar issue happen once to a computer. The issue with it was caused by a heatsink on the mobo that was loose and not making good contact. Browsing the net the pc worked fine, but once he fired up anything that put some stress on the board it would cut off within a few minutes. Then it would cut out again faster each time. So that's just something you may want to check also.
I've had to deal with that issue before. The heatsink on my motherboard over the VRMs came imperceptibly loose, but it was enough to shoot the temperature sensor near them to near 90C. Once the heatsink was reset the temperatures fell back into the high 50C range and I stopped getting weird behavior. With an overclocked CPU I would definitely check the motherboard temperatures with HWiNFO and make sure there isn't an issue there.

It also wouldn't hurt to check the temperature of the PSU while you're at it. That is a high-quality PSU that should be good through hot temperatures, but efficiency is going to drop the hotter it gets and you have a pretty demanding system. If it is bottom-mounted with the intake fan pointing down, make sure the case has clearance from where it is sitting and that you're using a clean dust filter.
 
Thanks guys.
Corsair RMA in progress.

To fill the gap in time ordered a Corsair HXi 1200 new from Amazon.

I'll either keep the new one or sell it if I get a 1200 AXi on RMA from Corsair.

I appreciate all the suggestions. The symptoms seem like the old AX 1200 just finally gave up.

I agree with this because I had the same issue last month when I went from a GTX 760 to a GTX 980. The PSU at the time was an Enermax EIN720AWT that was 5-6 years old and when I switched to a brand new EVGA G2 850W, everything went back to normal.
 
Looks immediately like a faulty PSU, so hopefully that RMA is justified. I personally don't like corsair anymore, since all their PSU's for the most part are made by Channel Well Tech now. Not that channel well can't make a nice unit, those AX and HX units are nice, but they're known for not having consistant quality control especially in terms of hardware assembly. If you happen have any issues I highly recommend either a Be Quiet! Dark Power unit or a Seasonic unit. However this sort of thing can happen to any unit from any manufacturer, they are not all perfect. Good luck!
 
I read the title and immediately thought "PSU is bad." You might be able to turn off a PCIe slot from your BIOS to disable a card, but I'm not sure that would help you fully diagnose this. You might not have the same power problem without the draw of *both* cards simultaneously.
 
I have had this issue a few times in my builds. The first 2 times it was due to the power supply not providing enough Amps on the rail. The next time it was a faulty power connection on the GPU. Now I have the problem every now and then with certain games, and I found out apparently I am drawing too many Amps on my circuit and my building can't figure out why it's cutting out. Needless to say, if you are not blowing the circuit, it is likely the PSU as you are testing now with the RMA.
 
I thought OP siad he tested another PSU already. I won't be surprised if the problem persists after psu RMA. Maybe something on the motherboard (vrm or even cpu) that is overheating.
 
I thought OP siad he tested another PSU already.

Nope, he tested the circuits meaning the place were the PC is connected, since circuits use the same type of safety devices with the thermo-magnetic breakers providing both surge and overload protection, but he noticed that the PC turned itself off while everything else on that same circuit kept on working.
 
Does it still happen with cpu at stock ?

I had game (MWO) which reset my few hours long prime95 stable overclock haswell until I dropped oc by 200 Mhz.
 
I thought OP siad he tested another PSU already. I won't be surprised if the problem persists after psu RMA. Maybe something on the motherboard (vrm or even cpu) that is overheating.

I have run the PC at stock and it's usual OC (has been running same OC for several years) and it has been rock solid under AIDA 64.
I tested the RAM, same.
The GPUs seem fine, but I have no good way to test them due to the watercooling set-up.

Does it still happen with cpu at stock

I had game (MWO) which reset my few hours long prime95 stable overclock haswell until I dropped oc by 200 Mhz.

No the CPU is solid, as is the MB.

Tell us what was the issue when you get a new psu.

I am expecting a new PSU today. Hope to have this up and running by Friday.
 
Tell us what was the issue when you get a new psu.

Looks like it was a PSU issue.
I finished the new install of the Corsair HXi 1200 this evening.

Ive played well into MadMax for over 45 minutes at a GPU overclock and I have not had any issues so far.

I still need to give LastLight a whirl, but it appears that the old AX 1200 just couldn't take the load anymore.
 
if you have more than 1 hard drive connected. Try disconnecting all the ones you don't boot on.
 
I had very similar issue, where the system would just turn off, the same as if the power cord had been pulled. sometimes it would last 5 minutes, other times 12 hours. it would always boot right back up. then one of my 6850's died, and with it the random power offs.

not sure if it was the card triggering it, or the PSU not being able to keep up with the load, but with a single card, it has been rock solid stable.
 
I had very similar issue, where the system would just turn off, the same as if the power cord had been pulled. sometimes it would last 5 minutes, other times 12 hours. it would always boot right back up. then one of my 6850's died, and with it the random power offs.

not sure if it was the card triggering it, or the PSU not being able to keep up with the load, but with a single card, it has been rock solid stable.

My GPUs are fine.

It was the PSU.
 
well all pointed to PSU, glad to see you solved the issues.
 
well all pointed to PSU, glad to see you solved the issues.

Yes, I'm glad it was that simple.

The system was used off/on all weekend both for work and gaming and there were no problems at all. :D
 
Well, even after replacing the PSU to a brand new Corsair HXi 1200 the symptom returned about one month later.

So this time I did a little more work.

I was able to isolate the GPUs singly, and the system was rock solid running one GPU.
Both GPUs are fine in testing.
I tore down all the cables and tested them with a multimeter, no shorting or bad connectors.
The PSU WAS set by default to multi-rail by Corsair. Otherwise was spot on on the multimeter.
I benched the MB and CPU alone and found no issues. The MB was advanced to the most recent BIOS version.
I repeated RAM testing and found no errors.

So I put it all back together (gave me a chance to clean up all the watercooling and put in a new reservoir)
I installed the Corsair link software and changed the PSU output to single rail.

I haven't had an issue since. All I did was change the PSU output from multirail to single rail.
 
FWIW, I had these exact symptoms about a week before my 670 shit the bed. Thinking it was the PSU, I threw away my CP850 after swapping to an AX650 gave me what appeared to be a stable system. Then it literally went up in smoke.
 
FWIW, I had these exact symptoms about a week before my 670 shit the bed. Thinking it was the PSU, I threw away my CP850 after swapping to an AX650 gave me what appeared to be a stable system. Then it literally went up in smoke.

I was convinced it was a GPU given it only happened when a game was run.

I tested both my GPUs under full load looping Heaven Benchmark for > 12 hours each and had no issues.

I've been now 2 + weeks using the computer and I've seen no issues.
My schedule doesn't let me game for hours on end, but I've gone 2 hours under full load in real world gaming with no issues.

Crazy shit, this.:D
 
Back
Top