Sign in with
Sign up | Sign in
Your question

GTX 780 + 700W PSU = PC restarts during heavy games.

Last response: in Components
Share
July 18, 2014 3:09:42 AM

My PC restarts with a beep during gaming, no BSOD, no crash dump, no recovered from failure msg on windows startup. Restarts are inconsistent, means it can happen in first 5 minutes or after 50 minutes.

This issue is across the board and not specific to one game. However I tested it with Battlefield 3, Battlefield 4, Watch Dogs. Happens more in BF4 than BF3 and much more in Watch Dogs.

I was using a Gigabyte GTX 670 (150W TDP) before without any problems and not a single restart / BSOD during it's usage.

I upgraded my GPU to Gigabyte GTX 780 OC (250W TDP) which is Overclocked version of the 780.

I have checked my temperatures, I am using H80i and CPU never goes above 60c (Fairly warm ambient as its summer). GPU stays within 70-78c range depending on load. Cooler Master Trooper Case with very good airflow, rear as exhaust, front as intake. Motherboard varies between 40-50c.

The main difference between two cards are GTX 670 used one 6 pin and one 8 pin connector while GTX 780 I am using needs two 8 pin PCIe connectors. (GPU is fine in another system)

Prince95: Passed 5+ hours
Furmark: No issues
OCCT: 3+ hours Linepack, no issues
Heaven benchmarks: No issues
Voltage Logs: No issues, all within atx range
memtest: Pass

Drivers tried, all of 330+

Here are my system specs.

Processor: Core i5 3570K (Stock, 1.10 vcore)
RAM : Corsair Vengeance 1866 (1.5v)
Motherboard: ASRock Z77 OC formula (Latest BIOS)
Corsair Force SSDs x2
Seagate 2TB HDD x1
Creative PCIe X-Fi Titanium
Corsair H80i Cooler
Xigmatek 700W semi modular PSU (NRP-MC702)

120mm fans x4
200mm fans x1
Cooler Master Storm Trooper with built in Fan controller

USBs connected

1x Mouse SteelSeries Sensie
1x Corsair Keyboard with USB downlink
1x 1TB USB 3.0 portable HDD
1x Dell U2713HM USB 3.0 Hub connector (Makes no difference if connected or not)

I have lost track of what to do now. Any help would be appreciated.

More about : gtx 780 700w psu restarts heavy games

a b ) Power supply
July 18, 2014 3:37:24 AM

check the temps of cpu and gpu at load .....if its not a temp issue then i will say the psu is struggling in loads that's why u are getting the restarts...xigmatic psu i dont know how they are if they are 80+ certified or not so i will recommend u to get a seasonic psu they are the best brand for psu available and can handle loads very well...
July 18, 2014 3:43:26 AM

prit87 said:
check the temps of cpu and gpu at load .....if its not a temp issue then i will say the psu is struggling in loads that's why u are getting the restarts...xigmatic psu i dont know how they are if they are 80+ certified or not so i will recommend u to get a seasonic psu they are the best brand for psu available and can handle loads very well...


I am leaning towards RM1000 or AX760 but to invest into that, I need to make sure it's PSU issue. Issue started as soon as I upgraded to faster GPU and since power requirements increased so as chances of PSU failure, but why are +12V readings show no sign of degradation, so as +3.3V and +5V.

80+ Bronze is written on box, but I am not sure. I picked it up on a good deal about 2 years ago but it ran just fine.
Related resources
a b ) Power supply
July 18, 2014 3:50:45 AM

well you need 42 amps and 600watt of psu minimum to run the gtx 780........ corsair rm series is also good but 1000 watt is overkill even for sli but if u want it that way and had the extra cash then u can go or it..
July 18, 2014 3:59:10 AM

prit87 said:
well you need 42 amps and 600watt of psu minimum to run the gtx 780........ corsair rm series is also good but 1000 watt is overkill even for sli but if u want it that way and had the extra cash then u can go or it..


Correction: The Model is Xigmatek NRP-MC702 not PC702 which is rated for 56A on 12V Rail Jonnyguru gave it 9/10 rating, so not a bad PSU in itself but it may have become faulty or just aged badly. http://www.jonnyguru.com/modules.php?name=NDReviews&op=...

RM1000 is overkill, yes but it's also ideal since it will stay at 50-60% load giving maximum efficiency and less heat.
Second option is AX760 which is Platinum certified and of very high quality but has exactly the same price as of RM1000 in my area. So should I choose AX760 and leave out the Wattage madness ?
a b ) Power supply
July 18, 2014 4:09:40 AM

i will say for a single card go for the platinum one but if u want to sli in future then u can go or the 1000w rm ....your current psu has enough wattage and amps for your card but i dont know much about the brand so i will not comment on it ....
July 18, 2014 4:14:43 AM

prit87 said:
i will say for a single card go for the platinum one but if u want to sli in future then u can go or the 1000w rm ....your current psu has enough wattage and amps for your card but i dont know much about the brand so i will not comment on it ....


Enough wattage but single rail. CPU and GPU drawing power from single rail and one of the capacitors may be reaching it's end of life.

Update: I borrowed a 1000W spare PSU from friend and restarts stopped. Just checked. So PSU has reached the end of life after serving only 2+ years. Which only means low quality hardware used since it should've worked for single card.
a b ) Power supply
July 18, 2014 4:27:06 AM

so at least now u are sure about changing your psu ...... that's good..
July 20, 2014 6:18:23 AM

The issue is back even with new 800W PSU with 58A on 12V rail. Whats more frustracting is I am not able to find a perticular condition.

To me it looks like my GTX 780 is not stable at factory overclocks but I am inexperienced with GPU overclocking.
a b ) Power supply
July 20, 2014 6:35:13 AM

Yea I was going to say it may not be the PSU at all. Sometimes BSOD are resultant of unstable overclocks of CPU or GPU. In your case the problem started when you bought a new GPU. It's also running a little on the hot side.

I would call the warranty department for your 780. They will probably want you to RMA it. If you are unable to do so, load up your overclocking software for the GPU and underclock it. It shouldn't be that hard, just google the stock clocks for the GTX 780 and match them. You should only have to adjust the memory and core settings, not the voltage.
July 20, 2014 6:53:38 AM

Buying new PSU is not a problem since I can resale my old one and minimize the investment cost, at least my PC is future proof now for SLI/XFire.

I have been gaming like crazy on GTX670 and don't even remember when I saw BSOD last time. So the moment I installed GTX780OC (rev. 2.0) Gigabyte Windforce, I got my first restart 20 minutes into watch dogs. I tested with BF4 and random restarts were there. So I started playing BF3, it lasts much longer and restarts are much less. Which only helps trace the issue because it's not as GPU intensive as former two.

I am now tracing the issue systematically from a fresh windows 7 install. Noting down all driver version and conditions and time and temperature when restarts happen.

I have noticed a strange artifact though. Although this over-clocked 780 is rated for 954/3004 clocks and boost up to 1006. Though monitoring the GPU clock its reaching way beyond specs at 1100 Mhz when on load. I opened up MSI after burner and GPU Boost 2.0 settings are also inverted. Searched and found that this particular GPU doesn't down-clock when reaching 80C.

It used strange 8(6+2) + 8(6+2) power connectors instead of reference 6+8 pin configuration.

PS: I reassembled my whole build making sure everything is correctly connected. Amazingly it lasted for 1.5 hours into games, but then again the same behavior.
a b ) Power supply
July 20, 2014 12:51:45 PM

Artifacting is another tell tale that the GPU is the problem.

Like I said, either try to underclock the GPU, or better yet RMA and get it fixxed or a new one.
July 20, 2014 2:48:06 PM

Adroid said:
Artifacting is another tell tale that the GPU is the problem.

Like I said, either try to underclock the GPU, or better yet RMA and get it fixxed or a new one.


That's the beauty of this problem. There is not a single visible artifact while in games.

I tested two settings.

Factory settings is 954 and boost 1006

While in Games, the Boost doesn't stop at 1006, It goes up to 1084, which nvidia inspector also shows. System restarts are happening while in game and not in Furmark, Furmark doesn't provide DirectX stablity tracking.



So I downloaded the Gigabyte OCGuru utility and turned down the clocks to Reference specs

863 and boost to 915, So I applied the settings and playing BF4 for half an hour but experienced no crash.

So to confirm I ALT-Tab to GPU-Z and to my amazement, Boost wast hitting 992 Mhz instead of stopping at 915

So my theory is, since I bought this card from a friend (Still in warranty), He might have tempered with it's BIOS or someone misguided him, he is not very tech savy and can mess up things in the name of curiosity.

So I headed over to Gigabyte website and downloaded latest BIOS (Mine is older version). Hitting another wall, the VGA_BIOS utility tells that BIOS versions are mismatched. No idea why because it's the same card as of Gigabyte website.

Now I am trying to contact Gigabyte and explain this issue or if a third party BIOS is available for this card.

a b ) Power supply
July 20, 2014 7:31:46 PM

Well the good news is that Gigabyte has a manufacturers warranty of 3 years from manufacturing date. The bad news is if your BIOS was hacked, it may have voided the warranty.

It may actually be fine running the lower clock speeds, so if it was a hacked BIOS you might should just leave it at the lower settings and let it boost up to 992.. Hopefully that will solve the problem.

July 22, 2014 2:52:09 AM

OK. This is strange behavior. I have Splinter Cell Blacklist so I thought why not max it out and check, Guess what? No crash whatsoever.

This may be a confirmation that the issue is at driver level. May be try an older version it that makes a difference.
July 30, 2014 8:50:33 AM

Is there any way you can update us on whats happened? I'm having same issue, GTX 780, 17 3.5ghz 8gb ram and I can play witcher 2 maxed settings with ubersamping 60fps on vsync but I played Shogun 2 maxed settings and Rome 2 and these two games restart my computer in exact same way you do, I think I have 700wat psu, have you found a solotuion yet? I have latest drivers, are they the issue or psu?
a c 2353 ) Power supply
July 30, 2014 9:09:29 AM

What brand PSU?
July 30, 2014 9:11:13 AM

SR-71 Blackbird said:
What brand PSU?


EZCOOL AMBIANCE 80+ 700w 4+4 PIN
a c 2353 ) Power supply
July 30, 2014 9:12:01 AM

That poor power supply could be an issue.
July 30, 2014 9:19:22 AM

Iv'e just tried rolling back driver, since Iv'e heard some people with the 780 and the same issue was fixed by rolling back.
August 4, 2014 12:20:18 AM

DTLeonard said:
Is there any way you can update us on whats happened? I'm having same issue, GTX 780, 17 3.5ghz 8gb ram and I can play witcher 2 maxed settings with ubersamping 60fps on vsync but I played Shogun 2 maxed settings and Rome 2 and these two games restart my computer in exact same way you do, I think I have 700wat psu, have you found a solotuion yet? I have latest drivers, are they the issue or psu?


Sorry for late reply but here is the story so far.

Replaced my PSU with a 850W 80+ Bronze Xigmatek Centauro (9/10 rating in reviews). To my surprise the issue remains at it is.

- The typical behavior would be; you launch a game, it loads fine, plays fine with no stutter and excellent performance. How long you can play is your luck. Suddenly it will look like you have lost power to PC. After 2-3 seconds the PC will boot itself from cold state, means not a typical restart but something like power failure. Funny thing is I have very powerful UPS connected to PC including a surge protector. But if surge protector is tripping, monitor and modem should restart too. THIS IS WITH BRAND NEW PSU.

- Second behavior is BF4 (Lots of restarts), BF3 (Takes longer to trigger a restart), Hitman Absolution (No restart even after 4 hours), SC Blacklist (Stable, no restart).

So to end my frustration I stopped playing BF4 and there was no restart in games for last one week. Then I bought recently released Divinity: Original Sin. Gorgeous game and excellent performance. I played once for like 1 hour to check the game and went to bed. Second day I loaded the game and there after just 20 minutes PC lost power and restarted.

Amazing thing is the PC was stable when I borrowed a 1000W PSU from a friend, in all games. But this is unbelievable since it's a single card and cannot Max out even a 700W or 850W PSU.

- I will got to local electronic store and buy an electronic meter reader. If power draw from wall socked is less than PSU rating while in game, than these are rouge cards. I will sale it and cross Gigabyte off my future GPU manufactures.
August 4, 2014 12:39:18 AM

SR-71 Blackbird said:
That poor power supply could be an issue.


No necessarily, TH forums are littered with same issue however typical rate of failing PSUs is not that high, specially when manufacturers happily give 3-5 year warranty on PSUs.

In my last one month of frustration, I am strongly inclined on believing these are badly manufactured cards. They perform great WHEN they work. However This problem is more popular with Gigabyte Kepler Cards.

My last solution would be to move to ASUS or EVGA.

I should have paid attention to why the guy was willing to sale it at just $400, without mentioning the problem (if there is).
a b ) Power supply
August 4, 2014 5:14:20 AM

This round of Gigabyte cards got very high reviews, and Gigabyte has a 3 year warranty and a pretty good reputation.

I'm not convinced it's Gigabyte's fault here - especially when you are running a hacked BIOS. Running a hacked BIOS ruins your warranty for a reason, and blaming the manufacturer when they actually have a pretty decent RMA program, and make a good quality product, is questionable.

I would try contacting Gigabyte anyway. They may work with you. I would just tell them straight up some guy before you installed a hacked BIOS. It couldn't hurt to try.
August 4, 2014 8:22:19 AM

Nops, using default shipped BIOS, I usually never temper with VBIOS, I counter checked the version with shipped BIOS.
a b ) Power supply
August 4, 2014 10:05:03 AM

Ok so they have a good 3 year warranty, just RMA it.

Even EVGA and Asus cards will have a lemon once in a while. Just because you got a single bad card doesn't mean the manufacturer is horrible. I have read all good reviews. And no I don't work for Gigabyte - if you want to go spend another 500$ on a card from a different manufacturer go for it.

If I was you I would just RMA and get the problem fixed for free, and move on with life...
August 5, 2014 3:51:02 AM

We have all been wrong all this time. It was never PSU or GPU problem. It was Bad RAM chip (Corsair Vengeance) causing a restart in games.

This can also explain why some games were working and some were restarting. x32 games cannot address more than 4GB and x64 Games can use all the RAM.

Since Battlefield 4 is x64 Game, it was utilizing both RAM sticks at maximum capacity. Even if GPU has 3GB VRAM, data has to pass through main memory and sent to VGA on CPU command.

I borrowed a single 8GB chip from a friend and since then, system is stable, rock solid. No restarts whatsoever in BF4.

Guess I now need to test memory to identify the bad Chip and replace it. Only regret is waste of money on new PSU.
!