My PC shuts off while playing games

[ShadySands]

Pretty much as the title says. It usually happens during gaming but I think there may have been a couple times when it cut off and I wasn't gaming. Also, no blue screen and no error messages or anything. Just on one minute and off the next.

 

I think it may be overheating but I wanted to see if any of you guys had any other thoughts of likely culprits because hardware isn't my thing

 

AMD FX-8150

Gigabyte GA-970A-UD3 

16GB DDR3-1600

2 AMD Radeon HD 7770 

850 Watt Corsair CMPSU-850TXV2 

2 120GB SSD  

2TB HD 

 

I dunno if there is any other info you may want. I bought it from ibuypower a little less than a year and a half ago

Edited October 25, 2013 by ShadySands

[PK htiw klaw eriF]

Check for dust on the fans, that can cause them to be less efficient.

 

I bought it from ibuypower a little less than a year and a half ago

Or make sure the power source isn't complete **** and all cables are in good condition, ibuypower delivers a pretty **** product from what I've read.

[ManifestedISO]

Humanoid could probably say in one sentence or less, but if it were mine, I would check that the heatsink/fan on top of the CPU isn't loose. Overheating is the only thing I know of that will shut a system down without warning ... or possibly a power supply unit about to go for a dump. Barring all that, I would just get a can of air and clean out the interior. Dust seems more insidious up there at elevation.    

[Bartimaeus]

If you *think* it's overheating, why not actually check while gaming?

 

I would guess there's probably a problem with a fan/heatsink - probably either the GPU's or CPU's.

Edited October 25, 2013 by Bartimaeus

[Gorth]

Doesn't the Gigabyte MB's come bundled with some monitoring software that does temperature readings from the various chips, including the CPU?

 

I once downloaded and tried a freeware utility called cpuz or some such, which also could create a log over time of cpu performance and temperatures. Might not be the cpu overheating, but there is 3 minutes involved setting it up.

[ShadySands]

If you *think* it's overheating, why not actually check while gaming?

 

I would guess there's probably a problem with a fan/heatsink - probably either the GPU's or CPU's.

 

I did and my CPU is running hot

 

but with my lack of expertise in this area I don't know if that is my only problem or if there could be other things I should also look into

 

but thank you all for your suggestions

[Gorth]

It might be something as "simple" as a sloppy assembly and/or badly applied paste between fan and chip(s). If it's running hot while the fan is actually working, you may have to reapply some thermal paste on it (the offending components).

[ManifestedISO]

It's fun, and easy, to apply a little thermal paste. If you can make kimchi tacos, you can remove and reattach a heatsink/fan. Plus, these guys have an English accent, so you know they can be trusted.

 

[media]http://www.youtube.com/watch?v=olBnEN9YkGc&list=PL6CFEA6C71BB14979[/media] 

[Humanoid]

CPUs these days tend to handle overheating a bit more intelligently than in the past, both AMD and Intel CPUs now should throttle instead of just instantly shutting down if they're overheating. Do you notice any slowdown or other performance related issues occuring before the shutdown?

 

I'd run the PC under load while monitoring temperatures. Run something like HWinfo64 to monitor temps while you test - I'd recommend two tests: something that pushes the CPU/RAM only, and then something that pushes the graphics as well - say OCCT then Unigine Heaven. HWinfo64 has the benefit of being able to monitor live clock speeds and the CPU voltage (Vcore) as well so you can tell if it's throttling before it dies (I think Bulldozer drops to 1.4GHz when it throttles).

 

What's a safe temperature? Chips these days are pretty resilient, I'd only get really concerned at 70+ (but a quick search shows that Bulldozer may begin throttling at 61+). If that's not a problem, then I'd move on to testing your memory specifically using Memtest (you'll need to boot into it from a disc or USB stick). If it passes that then it's a bit trickier because other potentially faulty components are harder to test. The PSU might be at fault, as could be the motherboard. For the latter you can use a multimeter (onboard readings of PSU numbers are horribly unreliable) and test using the back of the ATX connector while running under load (like so).

Edited October 25, 2013 by Humanoid

[ShadySands]

Thanks for that Humanoid

 

I had been using AIDA64 but I just ran HWinfo64 and got pretty much the same readings while playing Rift

 

my CPU is getting up over 80 degrees and I'm not seeing the clock speeds drop any

 

I plan on doing as the others have said and disassembling/cleaning/adding new thermal paste/reassembling my machine tonight and will let you know if there is any change afterwards

 

Thanks again

[Humanoid]

Now that is unusual. Wonder if the safety (amongst other power-related toggles) is disabled in your BIOS. Thermal protection comes in two general stages, throttling which is to some extent in the user space, and as a final measure, an emergency shutdown which is a fixed function of the CPU. Looks like you've skipped to that last step.

 

I looked up the manual for your motherboard and the relevant setting is "Hardware Thermal Control" under "PC Health Status". You may also want to check the setting of the various C states. C states, which will be listed individually like C1E, C3, C6, etc, are various features that turn off parts of the CPU when not under load - if they're off they might partly explain the non-gaming shutdowns. Consider just resetting to the defaults and retesting, especially if you've fiddled with the settings before, for overclocking purposes or whatnot. It's possible that the system shipped with them off, as power efficiency related features can affect maximum overclocks - this is fair enough for enthusiasts doing it on their own initiative, but it'd be plain irresponsible for a vendor to ship a system like this. (Might explain the negative things Kaine has heard about them - I've never heard of them myself)

 

I'd also likely use this time to update your BIOS, as there may have been issues with older versions which contribute to your problem.

 

But yeah, while that may explain your shutdown problem, it doesn't solve the overheating, so the cleaning and reapplication of thermal paste, etc, is a good idea. And more broadly, dust around your case intakes as well.

Edited October 25, 2013 by Humanoid

[alanschu]

I had similar symptoms back in 2010.

 

In the end my motherboard capacitors had bulged and my continued use of the computer ended up causing some damage to my video card as well.  I had had the computer since about 2006 or 2007.

 

Unfortunately, I don't know if the capacitors were the cause or the consequence, but given that 2007ish timeframe was a less than amazing time for capacitors they may have been of lesser quality.

[Humanoid]

I forget the specifics, but your capacitor issue might have been that one big incident that affected several companies back in the mid-noughties. If I remember right, it all began like a spy thriller would: in a case of international industrial espionage, a Chinese company induced an employee of a Japanese electronics firm to defect to them, taking knowledge and plans with him. In a case of, somewhat literally, Chinese whispers, some of the staff working for the Chinese company themselves defected - this time to Taiwan. And at this point, this fundamentally design had been mangled by being copied poorly twice over. But Taiwan being HQ to all the big PC component makers, this copy-of-a-copy product, being somewhat cheaper than the original, found its way into hundreds of different products very rapidly. Their rapid and predictable failure cost these companies billions collectively.

 

The kicker? Numbers I've seen about the 'savings' that switching to the cheap capacitors per product quoted figures of 10-50 cents, not for each capacitor, but in total savings per complete product. Yeah.

 

 

EDIT: Some post-referencing courtesy of Googling - Grauniad and Wikipedia.

 

Even now the aftereffects can be felt. Wondered why Gigabyte now suffixes their motherboard product codes with "UDx"? Because there was deep mistrust after the events of the above amongst the consumer base, it became necessary to market these products as being "ultra durable", as in, "we're using good caps now, honest guv".

Edited October 26, 2013 by Humanoid

[ShadySands]

Seems to be running cooler now but I haven't had a chance to put it through the paces yet

 

I took the downtime to go ahead and install Windows 8.1 for some unknown reason

 

Should have everything loaded back on it tomorrow and will test it more thoroughly

 

Thanks again for all the help

Edited October 26, 2013 by ShadySands

[alanschu]

Yeah, after the fact I learned quite a bit about capacitor dramas

[Keyrock]

If it winds up not being overheating that's the problem, then the next most likely cause would be the PSU going through its death throes.  If available, the easiest way to check is to switch it out for a known good PSU and see if that remedies the problem.

[ManifestedISO]

Uh, someone is a mere two posts away from the all-time high.

[ShadySands]

So, I'm still running pretty hot and I managed to hit 84 degrees but my average seems a couple of degrees lower and I'm idling much lower. It does appear that clock speeds are dropping when it starts to get into the high 70s and I haven't had it turn off on me yet

 

I'm starting to look around for a better cooling solution

[ShadySands]

and down she went

[AGX-17]

ibuypower delivers a pretty **** product from what I've read.

I haven't had any trouble with one from them, except for a defective video card (which was nVidia's fault, not theirs,) and which was replaced for free.

[Bokishi]

Happened to me 5 years ago when my PSU was overloaded from SLI

[Humanoid]

While a better cooler is always nice, it's not really getting at the problem since the stock cooler is plenty enough to cool at stock voltages. It could be that your stock voltages are messed up - check both in Windows what the Vcore is, it should be ~1.25V (around that, certainly less than 1.3V). And if you haven't done so already, I recommend a BIOS update (and implicitly, a reset to defaults).

 

Thermal paste doesn't make a big of a difference as your result are indicating, so even if misapplied it wouldn't be the cause of your issues. An incorrectly mounted heatsink could be, but AMD's stock cooler is pretty hard to get wrong - compared to Intel's it's pretty idiot-proof: either the clips are engaged and the lever is locked down, or it's loose. And though also unlikely to occur, check that the fan is plugged into the correct header, the CPU fan one instead of a case fan one (though again, this really ought to be a non-issue even if it's set up wrong).

 

Aside, while it's not particularly useful in diagnosis since we sort of know what the problem is, I'd be interested in seeing what the temperatures are like under load with the side panel open and a desk fan blowing into it.

[ShadySands]

After playing for about 30 minutes I'm showing my max CPU temp at 79.4C with the average at 70.5C and Vcore is showing a max of 1.376V with an average of 1.298V

 

I forgot to mention that I update the BIOS yesterday when I cleaned out my computer and reapplied thermal paste. Also, the cooler isn't stock, it's some liquid cooling system ibuypower was giving away when I ordered my machine.

 

That temp was with the side off but I don't have a fan to see if that makes more of a difference than the ~5 degrees I'm already seeing

[Humanoid]

Huh, is it one of those all-in-one watercooling kits like the Corsair H50 or the Antec Kuhler? Almost sounds as if there's a fault with it.

 

Not a fan of these units in general, it's almost as if they exist purely for people to be able to claim they're water cooled - a shallow claim indeed when they're outperformed in both temperature and acoustics by traditional air cooling. But that's a personal gripe of mine, and unrelated to the fact it should be working far, far better than it is. Could be anything from a faulty pump to a simple crack in the plastic retention mechanism.

 

I mean to a certain degree you get some flow through convection, which explains why it's not immediately shutting down (which would be the case if you had an unconnected waterblock installed, or even a small fanless heatsink). But yeah, at the cost of spending some cash I'd rather just ditch that thing and go for an entry level tower cooler which will be better in almost all regards.

Edited October 27, 2013 by Humanoid

[samm]

After playing for about 30 minutes I'm showing my max CPU temp at 79.4C with the average at 70.5C and Vcore is showing a max of 1.376V with an average of 1.298V

Temps and max voltage are way too high. Firstly, disable Turbo in mode in your BIOS, as this requires a massively higher voltage (your 1.376 probably) and does only marginally add to performance. Secondly, as stated by Humanoid, get a cooler that works. Not only will the PC be more stable with lower temperatures, but it will also be faster, because the FX will throttle at these high temps.

 

Regarding cooler, I'm using a Megahalems Rev. C on my FX 8350, and it does a reasonable job cooling it, even with a little overclocking. Just make sure the  cooler you get will fit in your case, and on AMD Socket AM3. Everything will be better than what you use currently.