Computer being restarted during graphically intense games

user8831

Commendable
Sep 22, 2016
5
0
1,510
Hey everyone,

First, let me post my specs:

OS: Windows 7 Home Premium 64-bit
Mobo: ASRock Z68 Extreme4 Gen3
Processor: Intel Core i7-2600k (Sandy Bridge)
RAM: Corsair Vengeance 16GB 1600MHz DDR3
GPU: NVIDIA GeForce GTX 950 (EVGA FTW model with backplate)
PSU: Corsair HX850
Primary (OS partition) SSD: Crucial M4 128GB SATA III SSD
Secondary (storage only) hard drive: WD Caviar Black 1TB HDD

Here's what's going on:

I built this computer back in 2012. Originally, I had an NVIDIA GeForce GTX 560 Ti graphics card installed. The computer ran beautifully with no issues until mid-September of 2016. All of a sudden, my computer would randomly restart (no BSOD, it was like the power went out and came back on a couple seconds later) during graphically intense games. Games like Just Cause 3, Mass Effect 3, Saints Row IV, and others. The computer would never reboot unless I was doing something graphically intensive.

After a lot of various troubleshooting with Prime95, MemTest86, and FurMark, I narrowed the issue down to what I believed to be the graphics card, as FurMark wouldn't make it more than 10 seconds before my computer would restart, and Prime95/MemTest86 tested fine. It was not an overheating problem that I could see, given that the card never got warmer than 76C under load, and usually ran around 40C idle.

Since the card wasn't overheating, I wavered between blaming the graphics card and the PSU. To test the PSU, I removed the graphics card from the computer and hooked up my monitor to the onboard graphics port. I ran FurMark again and it ran for over 20 minutes with no issues. Games would play with no issues, albeit with terrible graphics. So I decided on replacing the graphics card. I believed that there was an issue with it that was beyond overheating - something that wasn't right with the internal workings of the card that caused the computer to just restart on high load. Maybe some sort of power issue with the card itself.

At the end of September, I purchased an NVIDIA GeForce GTX 950, which I received in early October. Once I installed this card, my computer ran perfectly again like it used to. No random reboots during gaming whatsoever. Until... a couple of days ago. The exact same problem came back. My computer started rebooting during graphically intense games again, and FurMark refused to run more than a minute.

I checked the GPU temps again and they were even cooler than the previous card. I was getting about 30C idle and 72C under load, so I still do not believe this is an overheating problem. Here are the troubleshooting steps I have tried so far, and each step had several testing measures done between them to see if they worked:

- Updated my NVIDIA graphics drivers to the most recent version.
- Updated my mobo BIOS to the most recent version.
- Moved the GTX 950 from the top-most PCI Express slot to the middle PCI Express slot.
- Changed the way my monitors connect to the GTX 950 (for example, going from DisplayPort to DVI).
- Examined the graphics card for any physical damage (I did not open the card up, I just looked in through the sides and looked at the back plate) - no damage was found.

Right now I am at a complete loss as to what could be going on. My next thought is that maybe it was the PSU all this time after all, but if that were the case, why would my brand new 950 be shutting down a month after I installed it? Shouldn't it have been doing it when I first got it, too? Did the PSU destroy my 950, or did my 950 just become faulty in the exact same way my 560 Ti did?

Does anyone here have any other ideas on what else I can try, or what else I can look at?

Thanks in advance!
 
Solution
Hello.

Most probably your PSU is faulty. You cannot test the PSU at full load using the onboard graphics, so it is not a viable test.

The 560Ti is a much more power-hungry card than the 950. It may just be that your limit wasn't reached till now with the newer GTX 950.

I highly suggest changing your PSU at this point, or testing another if available.

Natsukage

Estimable
Oct 28, 2016
1,264
0
2,960
Hello.

Most probably your PSU is faulty. You cannot test the PSU at full load using the onboard graphics, so it is not a viable test.

The 560Ti is a much more power-hungry card than the 950. It may just be that your limit wasn't reached till now with the newer GTX 950.

I highly suggest changing your PSU at this point, or testing another if available.
 
Solution

user8831

Commendable
Sep 22, 2016
5
0
1,510


Thanks for your feedback, Natsukage. Do you have any other idea on what it could be if the PSU isn't at fault? I just want to see if there's anything else I can try before I drop $120-150 on a new good PSU and discover that wasn't the issue. I know they come in cheaper varieties, but the decent ones are around that price. Just feeling kind of defeated after already spending $150 on the new graphics card, and spending another $120-150 on a PSU needlessly would really be upsetting.

And if you don't mind, could you please clarify what you mean about not reaching my limit with the current graphics card until now? I was still playing all the same games on the 950 when I first got it up until now, so shouldn't my card have shut itself down within the first few days of using it instead of lasting an entire month without any issues?

 

Natsukage

Estimable
Oct 28, 2016
1,264
0
2,960
An intermittent PSU shutdown can be due to many reasons. Usually, it is either because the PSU detects incorrectly an overload and shuts down.
Or, there is a cold solder joint, which only has issues when a certain amount of current is reached. I've seen this multiple times in many different power supplies. Either the load or the temperature is changed, thus leading to failure.
Did the ambient temperature change since last month? Have you moved the PC around? Any amount of vibrations or temperature change can make a cold solder join go bad.

Since you cannot exactly monitor the PSU temps, there's no way to know if the PSU temps are higher or lower. You could try blowing cold air into the PSU and check if the error comes back?

As for buying another PSU, you usually are entitled to return it if you don't want it after all or are unsatisfied with 30 days? Check this out before buying another, and if so in your area, buy another, test your pc again, and if it's not that, simply return it.

Your power supply is way overpowered for your computer as well. Any reputed power supply with 500W or more will work perfectly fine.
 

user8831

Commendable
Sep 22, 2016
5
0
1,510
That makes sense, thanks for taking the time to explain. The ambient temperature has changed only a little between the last two months (going from summer to fall weather) but if anything, it has gotten cooler in the room and not warmer. As for the PC being moved, it was gently moved from one desk to another about 2 weeks before it started having issues with the 560 Ti in September, so I suppose that could have caused an issue as well, even though I was super careful when moving it.

I'm not in the best position to get a new PSU from a local retailer right now, since the closest one is over 30 miles away and I'm without a car for the next two weeks. I would have to buy one online, then pay return/restock fees if I wanted to return it. I think I will contact Corsair's customer support and try to RMA it if they agree that it is the issue, since my PSU has a 5 year warranty on it. I'll do that now and post back here when I have a better idea of what's going on.

As for my current PSU being overpowered, yeah, I definitely agree. I originally bought it in the event that I decided to SLI two cards together in the future, but that idea fell through and I never ended up doing it, so I now sit with a beefed up PSU for no real reason. Haha. If Corsair denies me the RMA, I will be getting a lower wattage PSU replacement instead.

Thanks again, Natsukage. I really appreciate your help and expertise on this whole situation!
 

user8831

Commendable
Sep 22, 2016
5
0
1,510
Natsukage, I just wanted to update you (and whoever else may be watching this thread, now or in the future) with the progress so far. This post is lengthy, but I want to detail the process for others in case they come across this thread looking for a solution. If you want to skip the story, then yes, it was the power supply all along and the solution was to replace it. Read below for more info if you'd like.

Alright, so I contacted Corsair customer service using their customer service portal. I had to register and sign up to it, then I was able to create a support ticket. They had various options to choose from on the issue description, one which matched mine almost perfectly: "My computer restarts intermittently." I chose this option, filled out the other information required about my power supply, then in the comments, I just explained briefly that my computer was only restarting during graphically intense games like Skyrim and Just Cause 3, and even though that sounds like it may be a graphics card issue, it doesn't appear to be, because it's doing it to two different cards.

Within 48 hours, I received this reply from Corsair: "What's the system temperature while playing games? Please check the voltage rails for your PSU by going to your BIOS and ensuring +3.3V, +5V, and +12V are within 5% of the values listed. And test your system with another PSU and GPU if possible. We'll be more than happy to replace it for you if it comes back with an error. Thank you."

My reply back to them was: "I will attach a few pictures that were taken from my BIOS and from HWMonitor. The HWMonitor picture shows a screen capture right before the system restarted a few days ago. The BIOS pictures show the +12V, +5V, and +3.3V values from this morning with no graphics card plugged in (I removed it temporarily). There are three pictures of the BIOS because the +12V and +5V were flipping between 3 different values rapidly. Again, these pictures are without my graphics card plugged in. If you think having my graphics card plugged back in would produce different BIOS results, I will upload new pictures of the values once the card is back in. As for testing with another GPU, I have tested with two cards and they still have the same results. An NVIDIA GTX 560 Ti and an NVIDIA GTX 950. The issue first started happening with the 560 Ti, so I bought the 950 to replace it, but the issue still persists. The issue is always the same; I play a graphically intense game for a few minutes and my whole system restarts with no warning, no BSOD, and no event in the Event Viewer except a "system has shut down unexpectedly" error. I do not have another PSU I can test the system with unless I go out and buy a new one. Is there anything else I can do to troubleshoot this one? Thank you."

About 6-10 hours after I replied and attached screenshots of my system temps/voltage readings, the tech went ahead and approved my RMA with no reason given. I didn't question it, but I do wonder what made them decide it really was the power supply going bad. Not sure if it was the fact that I tested on two different graphics cards, or if my screenshots gave something away that I just wasn't seeing. Anyway, the RMA instructions stated that I would need to send my PSU back, wait for them to process it, then they would send out a new one if they approved it. Knowing this process might have taken a couple of weeks, and I didn't want to be without my computer that long, I decided to do the express RMA.

The express RMA is basically the quickest way to get your replacement part. They put a hold on your debit/credit card for the amount of the new PSU, then they send the replacement PSU to you, then you send your old PSU back and they process it. If it's approved, the hold will be taken off your debit/credit card. If it's denied for any reason (like you don't send in your old PSU, or the PSU wasn't actually under warranty, etc.) they will charge your debit/credit card for the new PSU. I had to call in to Corsair, since they don't offer the express RMA option on their website and they give you a number specifically to call to do it. I called in and the rep told me they no longer make the HX850, so they would be replacing my unit with an HX850i instead. She said the total came to $199.99.

I took two things into consideration before I approved the transaction:
1. My current PSU was overpowered for what I had in my build.
2. That $200 could be coming out of my bank account if my RMA is rejected for any reason, which is always a risk.

I asked the rep if there was a lower wattage model that I could get as a replacement instead. She offered up the RM650i. Yeah, it's a lower-quality PSU because it's not the HXi series, but the lowest wattage you can get on an HXi is 750W. The rep compared the specs for me and said the RMi series is almost identical to my old HX series - obviously the HXi series is better, but in terms of being a replacement, it specs out almost the same. Anyway, I went ahead and requested one of those. The total went from $199.99 to $139.99, so it was a lot nicer in case I actually do get charged that amount in the event my RMA is denied for any reason. I don't anticipate that it will be, but it's always good to be prepared just in case.

It took a week and a half for the new PSU to arrive, but once it did, I immediately removed my old PSU and installed the new one so I could start testing it out. Let's just say I was able to play Just Cause 3 for 4 hours with no restarts - truly a beautiful thing. I didn't want to leave it at just one testing session, though, so over the next several days, I continued playing Just Cause 3 for several hours and even other games in my Steam library. The new PSU handled everything like a champ with the GTX 950.

As of right this moment, I'm still in the RMA process with Corsair. They technically only give you 10 days from the shipping date of the new PSU to send your old PSU back, but you can request an extension if you can't make the 10 day mark. They extended mine to December 9th and I just took my PSU to the post office this evening. I shopped around and the cheapest way I found to send it back was using a USPS priority mail large flat rate box, which came to $18.75 with the current postage rates. Tracking comes with it, and you can buy insurance for it, so I stuck $140 insurance on it (the fee was $3.35 extra) just to cover any incidents that could happen with it. $22.10 was the final price, which was considerably cheaper than UPS and FedEx, and even regular USPS ground service.

The expected delivery date is 12/1/16, and Corsair asks for at least 48 hours to process it once they receive it, so ideally, I should hear back from them by end-of-day Monday, if not sooner. I don't see any reason for them to deny the RMA unless they claim my PSU was mishandled by me or something silly. I've taken super good care of that thing, from the day I unpacked it out of its box, until today when I packed it back up nicely in a new box to send it back. It was never dropped, shaken, exposed to any extreme cold/heat, etc. so I really hope Corsair cuts me a break and accepts the RMA so I can keep my $139.99.

Anyhow, that's my story so far. Thank you again, Natsukage, for your expertise in this matter and giving me the confidence to reach out to Corsair for an RMA request on my PSU. I had a feeling in the back of my mind that it might have been the PSU this whole time, but I feel like it's one of those situations where it could either be the graphics card or the PSU, and you don't really know until you replace one or both. Trial and error.

I hope this thread and response will help anyone else having issues with their system. Good luck and don't give up hope!