How to investigate a PC that's recently started crashing

IMissMyDCast · 31 Aug 2014 at 10:39

A couple of years ago I built my own (very basic) PC from scratch (as a novice - never done this before). It went well - no problems until 2 or 3 months ago. All of a sudden, the machine has started powering itself down - no warnings, just bam - machine shuts down.

I have thought maybe memory problems or overheating CPU. My other thought is related to the HD - which I didn't buy at the time, I used my existing 1TB disk.

The PC details are as follows (most items bought from Overclockers):-

CPU: AMD A10-5800K Black Edition 3.80GHz (Socket FM2) APU Trinity Quad Core

CPU Cooling: Used the fan that came with the CPU.

RAM: GeIL EVO Leggera 8GB (2x4GB) DDR3 PC3-17000C11 2133MHz

Motherboard: Gigabyte GA-F2A75M-D3H A75 Socket FM2 VGA

Power Supply: OCZ CoreXtreme 500w '80 Plus'

Op System: Windows 7 64 Bit

The problem has mainly occurred in the following scenario's:-

Playing Quake Online while having Chrome running.

Transferring pictures from Camera to PC via Picasa

This is my home PC and it doesn't get used for much taxing stuff - I don't game on it other than above and the odd flash game.

I would very much appreciate some advice on where/how to start investigating what is causing this. Any suggestions much appreciated.

Philtor · 31 Aug 2014 at 10:46

Does it go through the shutdown procedure first or simply turn off immediately? Have you checked the temps? The likes of Hardware Monitor or Core Temp can do this by the way.

esK` · 31 Aug 2014 at 11:28

Try memtest to check ram could be a fault somewhere.

IMissMyDCast · 31 Aug 2014 at 11:40

Trying this now

Philtor said:
Does it go through the shutdown procedure first or simply turn off immediately? Have you checked the temps? The likes of Hardware Monitor or Core Temp can do this by the way.

Thanks Philtor, trying this now (installed Core Temp) - although, a newbie question on this, can you set it to regularly record the temperature, otherwise do I play quake while also viewing the Core Temp app? I am probably being dumb!

(it doesn't go through any shut down procedure - immediate turn off/power down - and with no warning).

IMissMyDCast · 31 Aug 2014 at 11:41

And will try this!

esK` said:
Try memtest to check ram could be a fault somewhere.

Thanks, will get this and give it a go as well!

MofoKilla · 31 Aug 2014 at 11:53

that ocz psu is prob killing its self.

Idleman · 31 Aug 2014 at 11:53

You might want to check your Event Viewer, particularly, the times it shut down, for any critical warnings. This may give you some indication as to what is causing it.

Svyper1 · 31 Aug 2014 at 12:38

Basically an imediate shutdown can be of these problems:

CPU overheating
HDD error
Memory fault
PSU fault

It doesnt really sound like a PSU problem becuase you are at least being able to use your computer...so this is the last thing you need to look at.

You can check the event viewer - type in start menu search "Event Viewer" and click it. After loading look through the errors. Most likely it will have critical error 41. This just means its powered off unexpectedly...

CPU overheating - get HWMonitor, free and easy. Can quickly see your temps to diagnose the problem. Idling 65+ means the cooler has degraded...

HDD error - you can run a program called chkdisk (check disk). right click on the c: drive, go to properties > tools > check now. tick the box "Scan for and attempt to fix errors". Then click start. A window will pop up and click "Schedule Disk Check". restart your pc and wait for a little while.

Memory fault - Memory can become faulty and the solution is simply to replace the memory. Run a memtest if you want to make sure as said above... if you cant run a memtest then try and see if other things are the problem, if all else doesnt work then replace the memory...

PSU fault - well this could be the case but really you are using your computer and it's more likely something else - particularly its only 2 years and a very basic machine...however you can stress test by using cpu burn and a gfx burn, play music and run a movie at the same time...(low settings in your case) this will max the power consumption. Do this last...

imo it is most likely in priority order:
* Memory
* CPU temps
* HDD

having said that you can check the cpu temps very quickly 1st

IMissMyDCast · 31 Aug 2014 at 13:32

still ongoing!

Cheers guys, lots of posts since I started my tests - so I am a bit behind on what people have suggested since I last looked!

I couldn't get Core Temp to work - kept saying zero temp on everything. Tried a tool called Open Hardware Monitor - which has worked. Ran up Quake Online, with Chrome running (which was the normal scenario for failure, ran it for a few minutes - ran fine, no crashing!!).

However, the monitor app recorded that the motherboard got to 89C ... seems high, but not really sure. Sits at around 35C when idling
The CPU got up to 75C, and sits at around 22C when idling.

I will try the following suggestions from Svyper1 now and report back later (today, hopefully):-

Svyper1 said:
imo it is most likely in priority order:
* Memory
* CPU temps
* HDD

having said that you can check the cpu temps very quickly 1st

I've downloaded something called MemTest86 which I'll try next. If anyone can advise on the temp tests - is that high? I'll report back later on Memtest results!

PS. Event viewer wasn't much help!

Dirk Diggler · 31 Aug 2014 at 14:09

IMissMyDCast said:
Cheers guys, lots of posts since I started my tests - so I am a bit behind on what people have suggested since I last looked!

I couldn't get Core Temp to work - kept saying zero temp on everything. Tried a tool called Open Hardware Monitor - which has worked. Ran up Quake Online, with Chrome running (which was the normal scenario for failure, ran it for a few minutes - ran fine, no crashing!!).

However, the monitor app recorded that the motherboard got to 89C ... seems high, but not really sure. Sits at around 35C when idling
The CPU got up to 75C, and sits at around 22C when idling.

I will try the following suggestions from Svyper1 now and report back later (today, hopefully):-

I've downloaded something called MemTest86 which I'll try next. If anyone can advise on the temp tests - is that high? I'll report back later on Memtest results!

PS. Event viewer wasn't much help!

89C for the motherboard!!? No, far too high. Should be in the 40C range at worst. I would go as far as to say that's a false reading, it's got to be.

Memtest should be run for a couple of passes to rule out memory problems.

Download Aida64 as it has a monitoring facility that you should find useful.

My money is on the PSU...

Idleman · 31 Aug 2014 at 14:52

What case do you have?

IMissMyDCast · 31 Aug 2014 at 20:12

more answers, more questions

Idleman said:
What case do you have?

It's a BitFenix Merc Beta Gaming Case - it's massive for the motherboard I got!

String said:
89C for the motherboard!!? No, far too high. Should be in the 40C range at worst. I would go as far as to say that's a false reading, it's got to be.

Memtest should be run for a couple of passes to rule out memory problems.

Download Aida64 as it has a monitoring facility that you should find useful.

My money is on the PSU...

Once I read your post, I tried Aida64 to test temp, it didn't do a "Max" reached temperature but it certainly seemed to have lower values on closing Quake. I'll try some more of that tomorrow.

I also had a stab at Memtest - which was interesting - it ran for about 20 seconds maybe, then the machine stopped. So I guess that means it is a memory issue??? Tomorrow I will try taking one of the modules out and trying Memtest on it, then trying it on the other module. etc. See what happens!

Dirk Diggler · 31 Aug 2014 at 20:40

IMissMyDCast said:
It's a BitFenix Merc Beta Gaming Case - it's massive for the motherboard I got!

Once I read your post, I tried Aida64 to test temp, it didn't do a "Max" reached temperature but it certainly seemed to have lower values on closing Quake. I'll try some more of that tomorrow.

I also had a stab at Memtest - which was interesting - it ran for about 20 seconds maybe, then the machine stopped. So I guess that means it is a memory issue??? Tomorrow I will try taking one of the modules out and trying Memtest on it, then trying it on the other module. etc. See what happens!

What do you mean by "stopped"? Did it power down or restart?

Svyper1 · 31 Aug 2014 at 21:09

IMissMyDCast said:
I also had a stab at Memtest - which was interesting - it ran for about 20 seconds maybe, then the machine stopped. So I guess that means it is a memory issue??? Tomorrow I will try taking one of the modules out and trying Memtest on it, then trying it on the other module. etc. See what happens!

It shouldnt stop so yeh it looks like a mem issue. Just buy 1600mhz ram, its only like 0.001% slower and save you money compared to higher clocked ram because really the only thing that changes much between them is the looks... tbh i'd recommend corsair XMS3 - load the XMP profile for 1600mhz in bios and all will be fine. I trust corsair more than other brands myself (for ram)...but thats my opinion

The unpredictability of the system restarts tend to be because of faulty ram...

CPU temps are quite obvious really - you stress the CPU in some way eg games or hot day or another program and shuts down...not to mention everything goes slow before this happens...

HDD error is obvious whenever you're trying to use the HDD for something. The event viewer will be full of HDD type error codes...

The PSU is only the 1 at fault when really you're maxing it out or it doesnt start at all...Too many people on here view the PSU as a weak spot and overspend, when infact its quite reliable, as with faulty ram like yours or HDD failures every component has a chance to be faulty but really, saying immediately it's the PSU is actually kinda ignorant...

you can test each ram if you want but really it's much easier to get another set of ram...but its youre call

IMissMyDCast · 1 Sep 2014 at 13:33

Svyper1 said:
you can test each ram if you want but really it's much easier to get another set of ram...but its youre call

I kinda agree with you - but I'm far too cheap to just buy ram without doing some more checking!

I am also slightly worried that maybe it is an issue with some memory related connections in the Motherboard? I know I am ignorant of the details in this area, but I think if I at least test both modules it will at least confirm that it is a specific module.

Dirk Diggler · 1 Sep 2014 at 14:14

IMissMyDCast said:
I kinda agree with you - but I'm far too cheap to just buy ram without doing some more checking!

I am also slightly worried that maybe it is an issue with some memory related connections in the Motherboard? I know I am ignorant of the details in this area, but I think if I at least test both modules it will at least confirm that it is a specific module.

Test the RAM with memtest. Why spend money on new RAM if it isn't causing the problem.

Greboth · 1 Sep 2014 at 14:25

IMissMyDCast said:
I kinda agree with you - but I'm far too cheap to just buy ram without doing some more checking!

I am also slightly worried that maybe it is an issue with some memory related connections in the Motherboard? I know I am ignorant of the details in this area, but I think if I at least test both modules it will at least confirm that it is a specific module.

If Memtest failed (I assume it did by your it stopped comment a few posts up)It may be the DIMM slots however it is far more likely to be the RAM sticks themselves. If you want to double check though, you can run with only one stick of RAM. Run Memtest until it passes or fails, move the stick to another DIMM slot and test again. Obviously if both sticks fails in 1 slot but pass in the others stands a chance of being motherboard but if it is your RAM I would bet you would get errors with one stick regardless of which slot it is in.

Threepwood · 1 Sep 2014 at 14:26

Open the case and check for dust clogging the cpu heatsink, just by eye.

Download another hardware monitoring program, did I read that your cpu is incompatible with core temp? Try 'CPUID HWMonitor' instead.

David Bisset · 1 Sep 2014 at 14:51

Just powering off without error/bluescreen etc sounds a lot like PSU or power-delivery on the mobo to me, surprising it didn't even manage an entry in event viewer though. Overheating CPU would give a desktop message before dying on you unless it was a very sudden overhead which it doesn't sound like (though that can cause restarts in memtest too) so I'd move that to the bottom of my suspects list.

On RAM... that RAM is quite fast. I assume you're using XMP to set it to run that fast? You could try turning XMP off so it's running at stock and see if that resolves your issues for now. This wouldn't actually confirm it's the RAM/slot causing the issue as you'll draw less power doing this too.

With memtest dying I'd say a faulty HDD is very unlikely as you can run memtest with no HDD even connected (try unplugging it to make sure!)

My bet would be PSU, it's the most common part to fail and the symptoms fit, but as checking the RAM more doesn't require buying new parts (I'm assuming no spare PSU to test with) it's probably worth trying a few more diagnostic steps.

( I'm assuming there is nothing visibly wrong with any of the components and all fans are spinning? Especially the PSU fan I guess. )

Dirk Diggler · 1 Sep 2014 at 15:57

David Bisset said:
My bet would be PSU, it's the most common part to fail and the symptoms fit, but as checking the RAM more doesn't require buying new parts (I'm assuming no spare PSU to test with) it's probably worth trying a few more diagnostic steps.

I think PSU too. It happened to me recently, just random shutdown and restart. Luckily I had a spare to test with at the time.