Random restarts

[vwj]

Lately, my computer has been randomly restarting (not shutting off) while I'm playing games. I have yet to have it restart on me while I'm searching the net, going through files in my computer, etc or anything that doesn't involve graphic intensity. Sometimes it will restart when I'm just a few minutes into the game, or even when the game is still loading, or it can restart when I'm 30 min into the game. It's seems completely random.

It doesn't matter if my cpu and gpu is overclocked or not, and it doesn't seem to be a heat issue. I've been monitoring my temperatures and everything seems to be fine. My gpu, EVGA 8800gt, seems to run a little hot, but I think that's just how it is (~65 C idle to anywhere from 80 C to 90 C).

The only thing I could think that could be causing this is the PSU, but I've had my system for a few years now and everything's been fine up to this point. Another possibility is my RAM or maybe even my hard drive that's running my OS (Win 7)...I'm at a loss for sure. Any input would be appreciated. Thanks.

 

[rolli59]

It could be power issue since it only happens during load. PSU's age and that could very well be your problem. Post your full system specs!

 

[vwj]

I currently don't have any stress programs that I could test each component, mainly the CPU, GPU, and RAM, but I have been running games that have any sort of benchmarks as an alternative to stress testing. It seems like it doesn't really matter if I OC my CPU and GPU or not. If I'm anywhere near my computer, I'll try to loop the benchmark over and over. Usually I don't have any restarts when the benchmark is running, but a few restarts do happen.

On the occasions that I leave the room when the benchmark is running (I'm going to make the assumption that the benchmark has finished and it's sitting idle in the results screen), I almost always come back to my startup screen where I have to insert my password, which means that my PC restarted.

Here are my system specs:

MB- GIGABYTE GA-P35-DS3L
CPU- E6750 @2.66
RAM- 2GB (2 x 1GB) G.SKILL F2-6400CL5D-2GBNQ
PSU- hec ZEPHYR 650
GPU- EVGA 8800GT

From BIOS, my temps are:

CPU- 24C
SYSTEM- 28C
v core- 1.316v
DDR18V- 1.88V
+3.3V- 3.312V
+12V- 12.239V
And my GPU normally sits idle in windows at around 65C.

 

[rolli59]

The PSU size wise certainly is big enough for your system and since your temps are fine (90°C for the GPU is OK but not great) then everything points to power issues.

 

[someonewhoknowsalittle]

I don't know what benchmark you were running so don't know what's causing the problem.

To troubleshoot the issue, first make sure ALL your power and data cables are well connected in your computer, and the EVGA card is well seated in the PCI-E slot while the computer's power is OFF and the power cable connecting your computer to the power outlet in the wall is DISCONNECTED. Then if any of the cables seemed loose, redo your "Benchmark" test. If you still have the same problem, with your computer off and power disconnected again, replace the PCI-E power cables from your PSU to your EVGA card, with known good PCI-E cables you can borrow from a friend, and try to reproduce the problem once again with your "Benchmark" test. If it still causes the same "reset" problem, I suggest to do the following:

Then, run some free diagnostic tests to isolate the resets may be caused by your CPU or RAM.

Download Prime95 from the MajorGeeks website for instance, and run it without running any other intensive program on your computer. Choose "just stress testing", then run the "Small FFTs" option for at least a few hours (12 hours is considered enuf by most people) to test the stability of your CPU.

Then, stop it and run the "Blend" test (just stress testing again), again for hours, to test to see if the CPU and RAM in combination are running stable.

Also, one of the best tests for RAM alone, is to run MEMTEST86. Download the free bootable version of MEMTEST86 Version 4.20 from MajorGeeks or another reputable website and burn it to a CD, then reboot with said CD in drive (with CD/DVD drive at higher priority in startup list in your BIOS than other drives), then run MEMTEST to see if your RAM is still good.

If your rig passes all Prime95 and MEMTEST86 tests, then cpu and ram have been ruled out as culprits. Then it may be your PSU or video card. Have you measured the voltages of your PSU by using software under "load" conditions. The voltage readings you have in your bios are when your PSU is at idle/lowest power conditions so they do not mean that the 3.3 volt and 12 volt reading under load (80-100% wattage usage of capacity of PSU) are still within the normal tolerance ranges. I don't know if your brand of PSU is considered quality. Maybe someone else can inform us on that.

To test the video card, if you have a friend with a known good and stable system who has at least a 650 watt quality (80+ certified) PSU and who lets you, swap out his/her video card(s), and swap your video card in and run the same game you were having problems running with your computer. If his/her computer reboots like yours did, then the problem is probably caused by your EVGA 8800GT.

 

[vwj]

Thanks for the replies. I've been meaning to download these stress programs, but haven't got the time yet. This is why I'm running these game benchmarks as quick alternatives. BTW, the games I'm using are FarCry2 and Just Cause 2 for these benchmarks, but mainly Just Cause 2.

I think it'll be difficult to reliably and efficiently reproduce the reboots that I'm having, since it's all random. It could take as little as a minute or it can take 2 hours.

Within the next few days, I hope I'll have the time to run these tests.

 

[vwj]

I've managed to run a stress test for my GPU. I used FurMark, with everything at stock. I ran the 'stability' test and the first two runs gave me restarts no more than 90 seconds into it. I continued running this test and the max that I'm able to get is about 5 minutes with my GPU temp @ 105 C. According to Nvidia, the max temp for this card is 105 C, and I didn't want to go past that since I was afraid I could damage the card. However, I've read that you should have it running for around 30 minutes to really know if the GPU is running stable or not; but I don't think I can even leave mine running for quarter of that length since it gets hot so fast.

I also used FurMark's benchmark test a few times, and I was able to run them for the full length duration every time, which is approx. 1 minute each time.

I haven't stressed out my CPU or RAM yet, and I hope to do within the coming up days, since I hear they are supposed to take hours to appropriately access them.

 

[rolli59]

The cooler on your GPU is clean and the fan is spinning up under load?

 

[vwj]

This is a bit of a long post. I've managed to run tests on all 3 major parts.

GPU: I swapped out my 8800gt for a HD4830. I downloaded and updated the drivers for the ATI. I ran FurMark again with the HD4830 and pretty much the same thing happened: it restarts around the 5 minute mark, albeit it runs a lot cooler than the 8800gt. It also passes the benchmark test every time and continues to restarts during games. So, it seems like the GPU isn't directly at fault here since the same thing happens to two different cards, and it's unlikely that both cards are bad units. In addition, I'm not too familiar about FurMark, but it seems as if it continues on going until the GPU gives up. I have a stock EVGA 8800gt and it heated my GPU past the 105 C mark, which is labeled as the ceiling for this card.

CPU: I downloaded IntelBurnTest to test out my CPU since I didn't want to wait hours and hours to run prime85. I'm running everything on stock anyway since the random restarts occurs whether it's stock or OC. I ran the stability test a few times and every time it checks out cleanly, with decent temperatures (normally tops off at ~50-55 C). So I'm ruling out my CPU.

RAM: I have 2 sticks of 1 GB RAM. I tested it with memtest, but not as thoroughly as some people. I used the recommended clock settings. The first test I ran was with 1 stick, and I ran it for about an hour, completing 3 tests. No errors were found.

Overnight, I decided to test out both sticks and when I woke up in the morning it found 1 error on test 6. Next, I decided to test the other stick that I didn't test the night before, but this time I let it ran for much longer. There was 1 error on test 8, pass 5. I haven't had the chance to run the test more, partly because it takes so long.

I think that the GPU and the CPU should be okay. The CPU passed all its tests. Both GPU couldn't go past the 5 minute mark on FurMark's stability test and I continue to have random restarts, no matter which GPU I use. I think it's unlikely that both GPUs are bad. Regarding the RAM, memtest did find an error when I left it running for a few hours. Again, I'm not familiar with memtest, so I'm not sure whether 1 error is considered disastrous or not. Based on these tests, I'll need help determining whether any of these parts are faulty.

On another note, could it be my motherboard? I don't want to get too off topic, but it might be relevant. The other day, I powered up my motherboard and it was stuck on the MB splash screen. It wouldn't even let me get into BIOS. I unplugged the power cord, made sure all the HDD were connected, etc. but I couldn't get past it.
In short, the only way I could get it past the post screen was when I connected only one HDD SATA cord to the motherboard, and it had to be a specific port. If I connected to another port, or had my other SATA cords connected, I'm stuck at the post screen.
And finally, BIOS won't recognize my cd/dvd drives anymore, nor can windows. Everything is connected firmly and I even switched IDE cables. All drives have power, as I can open and close them.

As you can see, I'm having a lot of issues with my computer at the moment. I'm wondering if everything could be related or if something is the root of all these problems. I'm not sure if swapping out a new PSU will solve my HDD and cd/dvd drive issue.

 

[someonewhoknowsalittle]

RAM: I have 2 sticks of 1 GB RAM. I tested it with memtest, but not as thoroughly as some people. I used the recommended clock settings. The first test I ran was with 1 stick, and I ran it for about an hour, completing 3 tests. No errors were found.

Overnight, I decided to test out both sticks and when I woke up in the morning it found 1 error on test 6. Next, I decided to test the other stick that I didn't test the night before, but this time I let it ran for much longer. There was 1 error on test 8, pass 5. I haven't had the chance to run the test more, partly because it takes so long.

The 2nd stick of your RAM did not pass Memtest. That means it is defective, plain and simple. It is inadvisable to run a computer with any defective RAM because it will introduce errors into your CPU's calculations and you will not be able to trust any result it produces, and applications and the OS may freeze as a consequence.

First, under safe conditions, remove your defective stick of RAM. If it is still under warranty, contact its manufacturer and obtain a RMA and send it in for a free replacement. Then, either wait if you can to receive the replacement RAM or buy a new one (with lifetime guarantee) that is compatible with your MB and same speed and size as the other known good stick of RAM you have. Then, test the replacement stick of RAM alone with Memtest and make sure it passes all tests. Then if it does, retry your FurMark tests with both sticks of RAM before you even think of replacing any other component. There's a good chance all the problems you described stem from that 1 defective RAM stick.