PC constantly crashing after 3 good years.

[kidabean]

Hey guys, wonder if you can help me out.

I'll try to keep this detailed but concise, having been at it for the better part of two days trying to figure out what's happened to my pc.

Long story short, built my pc nearly 3 years ago (specs at end of post) and has pretty worked fine all the way through until now.

The main problem is that it now crashes every time I use it - usually (but not limited to) when I'm gaming.

At first I thought it was my GPU, but the drivers are all up to date, plus it only crashes on very specific games.

Crash Scenarios

Starcraft 2 - never crashed once during my 5 month hardcore gaming spree.
Borderlands - again never crashed
Deus Ex Human Revolution - never crashed
Skyrim - occasional crashes
Diablo 3 - occasional crashes

However, it now crashes every single time I play League of Legends. There's no particular time frame in which it'll crash. Early as 10 mins or as late as 2 hours in.

Now, I've also crashed while streaming videos online (e.g Youtube) or worst of all, even while copying files onto my portable hard drive (without any other tasks running in background). The latter made me suspect it's not a GPU problem...

The crashes all involve the same symptoms:

Crash Symptoms

Screen freezes up; image on screen distorted.
If sound was playing, the sound becomes distorted too.

Sometimes this will then be followed by BSOD, other times it won't.

Finally, upon reboot, it'll either boot up fine asking if I want to start Windows normally etc, or it'll not boot up at all. When it doesn't boot up, all fans (including CPU fans) are working fine, the motherboard doesn't beep, there's no signal on monitor and peripherals aren't recognised.

Suspecting the RAM may be the cause I have ran several tests using Memtest86+ during the last 2 days.

If you haven't checked out my specs below, I have 2x2gb sticks of RAM and 4 slots on my motherboard.

From what I've read online, I am to believe that if a stick passes the memtest 100% in a given slot, then it is working fine. However, the results I've had are far from clear...

I've just noticed while typing this, that I may have had the 2 sticks sitting in the wrong pair of slots all this time, but I'll let you guys decide whether this has had any effect the crashes.

Basically, on my motherboard the DIMM sockets are positioned as such:

http://benchmarkreviews.com/images/reviews/motherboards/M4A785TD-M/ASUS_M4A785TDM_DIMM_Slots.jpg

The two black slots are, from left to right, A2 and B2, and the blue slots are A1 and B1. I've always had the RAM sticks in the 2 black slots which I now know are A2 and B2. Nevertheless, below are the test results I've had to date.

Memtest86+ Results

Sticks in both A2 and B2 = Last night, I had 1 pass (all 11 tests), then a second test straight after produced a fail. The failure consisted of an endless stream of red error lines rising by the hundred-thousands in mere seconds until I powered down at near a MILLION errors. Today, 2 tests resulted in no errors but simply crashed/froze during the test, and would not boot up like I mentioned earlier in the symptoms sections

1 stick in A2 = DIMM1 & DIMM2 4xpasses each

1 stick in B2 = DIMM1 & DIMM2 both had 2xfails each last night, but 2xpasses each today. Strangley, whenever I got the stream of red error lines during the failed tests last night, if I moved the mouse the PC would instantly crash/freeze.

Sticks in both A1 and B1 = 2xpasses last night.

I don't understand why I keep getting different results each time I run the tests. Can anyone shed some light into this matter?

Overheating?

I don't suspect overheating to be an issue as all fans work just fine and the few times I've managed to reboot successfully after a crash, I've checked the temperature monitor in BIOS which has the temps for both CPU and GPU to be around 40 degrees celcius.

Faulty Motherboard?

Because of the horrifically inconsistent memtest results, I cannot determine whether the motherboard is faulty, or the DIMM(s). What I can say is that, if I take all the RAM sticks out and try to boot up the PC, I get the same results 100% of the time.... It doesn't boot up, all fans (including CPU fans) are working fine, the motherboard doesn't beep, there's no signal on monitor and peripherals aren't recognised - identical to what sometimes happens after a crash. I don't know if this meant to happen or if this is a vital piece of evidence - I have read from several sources that your motherboard is meant to beep if there is no RAM present.

PC Specs

- AMD Phenom II X2 Dual Core 550 3.10 GHz Black Edition (Socket AM3)
- Saffire ATI Radeon HD 4770 512MB GDDR5 PCI-Express
- Asus M4A785TD-M Evo AMD 785G (Socket AM3) PCI-Express DDR3 Motherboard
- OCS Reaper 4GB (2x2GB) DDR3 PC3-10666C6 1333MHz Dual Channel
- Coolmaster eXtreme Power 650W Power Supply
- Seagate Barracuda 7200.12 500GB SATA-II 16 MB Cache-OEM
- Samsung SH-S223BBEBE 22x DVD+RW SATA Dual Layer ReWriter - OEM

Sorry for the long post, but I'm starting to run out of ideas, so wanted to give you guys as much information to work with.

Hope the info is sufficient, feel free to ask me to elaborate further on any part.

Thanks in advance,
Dave.

 

[swifty_morgan]

remove, pull apart, reapply thermal paste and clean the heat sink fins from the video card. Reassemble and reinstall.

Pull the heat sink and fan from the processor. Clean off and reapply new thermal paste, reassemble. Clean the heat sink and fan and other case fans. Dust build up may occur on the outside of the case blocking the holes the fan draws air through.

don't care that you've been watching temps. if you look in the BIOS after crash the temps are already well on their way down.

 

[kidabean]

Thanks for the advice, didn't think temps could fall that quick after just a few seconds. I'll certainly give it a shot seeing as thermal paste is a lot cheaper than replacing parts... I'll also download an actual temp monitor that can log the temperature before a crash in the mean time. Is there any software you recommend?

 

[swifty_morgan]

No.

 

[Cazalan]

If you're moving the DIMMs around that much you'll want to clean the contacts with some rubbing alcohol.

Dust is bad too. Get a can of compressed air and blow out the connectors and any fans.

With PC age comes the death of cheap capacitors. Look for those.
http://images.search.yahoo.com/search/images?_adv_prop=image&fr=yfp-t-553&sz=all&va=blown+capacitor

 

[kidabean]

Thanks Caz, the capacitors all look fine thankfully. I'll add the rubbing alcohol and compressed air to my shopping list for tomorrow!

 

[kidabean]

Ok, I cleaned all the fans and heatsink, replaced the thermal paste and blew all the dust off the components and case. Things were working fine for most of the night. I had the temp monitor logging the temp every 10 seconds until my pc eventually crashed after around 5 hours. I rebooted and checked the temp my CPU was at when it crashed and it registered 41C. Surely that can't be too hot?

 

[InvalidError]

I rebooted and checked the temp my CPU was at when it crashed and it registered 41C. Surely that can't be too hot?

41C is insignificant, nothing to worry about until over 70C.

My first guess would be the PSU. If you have another PC you could swap PSUs with, try that.

 

[swifty_morgan]

if you did that and it ran fine for 5 hours we're on the right track. But after crash how long until you were able to read the temp ? and yes, it doesn't take long for the components to cool after no current is going through them. and did you check the case fans and did you clean the out side of the case ( air draw holes ) opposite the intake fans.. ( usually the front are hidden from site.

 

[InvalidError]

if you did that and it ran fine for 5 hours we're on the right track. But after crash how long until you were able to read the temp ?

How long does not matter since he turned sensor logging on and pulled his 41C reading from the last log entry before the computer crashed.

 

[swifty_morgan]

wonder how I missed 4770 ? way under powered for modern gaming. however, for kicks, try uninstalling the HD HDMI sound from the device manager. Don't let windows reinstall after reboot.

 

[kidabean]

wonder how I missed 4770 ? way under powered for modern gaming. however, for kicks, try uninstalling the HD HDMI sound from the device manager. Don't let windows reinstall after reboot.

Yeah the 41C was taken from the log 10 or fewer seconds before the crash so it should be a pretty accurate. All the fans and drawholes are spotless.

I have a feeling the graphics card is the one overheating rather than the cpu. I haven't been able to replace the thermal paste on that yet. As for modern gaming, I don't use it for graphic intensive games - it handled Starcraft 2 perfectly fine on high settings, and even ran Deus Ex: Human Revolution on high settings with no crashes. My current game, League of Legends has graphics akin to Starcraft 2 so in theory it should be able to handle it.

It's a long shot, but perhaps it's recently started to crash so frequently because it's summer and my room gets really hot - especially with the pc on. Is it worth installing a fan on the side panel directly opposite the GPU? As currently I only have top and back.

 

[gamerk316]

...STOP IT!

Seriously, he clearly has either a RAM or failed DIMM slot. I have no idea why everyone is going on a tangent when we have confirmed memory failures popping up. This is NOT a heat issue.

At this point, I would test each stick of RAM, individually, in each DIMM. Minimum 10 passes for each. My suspicion is one of the sticks is faulty, though based on your description, I can't rule out a failed DIMM [though very unlikely, I've seen it happen enough].

The reason to test one stick at a time is to isolate a potentially faulty stick of RAM. If both sticks fail one a particular DIMM, then that DIMM probably failed. If both sticks fail across multiple DIMMs, then I'd suspect either a dieing PSU or a dieing mobo, rather then the RAM itself.

 

[InvalidError]

If both sticks fail one a particular DIMM, then that DIMM probably failed.

DIMM is synonymous with 'stick' so it would be pretty hard to plug a 'stick' in a DIMM.

Replace your mentions of "DIMM" with RAM/DIMM slot.

 

[kidabean]

At this point, I would test each stick of RAM, individually, in each DIMM. Minimum 10 passes for each. My suspicion is one of the sticks is faulty, though based on your description, I can't rule out a failed DIMM [though very unlikely, I've seen it happen enough

Thanks for your input. Regarding the RAM...while I didn't test them 10 times in each slot, I did test each stick individually in each slot.

In one night, I tested DIMM1 in slots A1/A2/B1/B2 and got passes in all tests. I then tested DIMM2 in slots A1/A2/B1/B2 and again got all passes.

I think I have failed to mention that, in the past two days I have had my processor pulled out from under my desk with the side panel taken off. I've also had a fan blowing in my room to keep the room temperature relatively cool, and have not had a single crash since.

I've just finished playing League of Legends in this way for the last 4 hours without a problem - which goes the same for last night too. I just checked the temp on my GPU and it's sitting at 50C (pre-game was 37C). Keep in mind that it's reached this temp while having a side panel off and a fan blowing nearby to keep it and the room cooler than usual. I suspect that, if I put the side panel back on and turned off the fan (as I did when I replaced the thermal paste recently) the GPU will most likely overheat and crash my PC again. Would this be a likely scenario?

I suppose the only way to find out is to do just that, and log the GPU temp before a crash. However the temp monitor I have is AMD's own real time monitor with no option to keep a log. Does anyone know of a good piece of software? Also, I have no clue as to what the normal temp for a GPU should be. To me, 50% does seem a little too high for my liking...considering I've already took measures to lower the temp.