Computer Shuts Off / Locks Up With No Warning (long)

[xcheck]

I'm at a loss and looking for any ideas. My system has been running fine since Feb 07, but lately has been acting strangely. It either locks up, shuts off without warning, after 2-4 hours of use.

Windows XP sp2
2gb ddr ram
Western Digital Caviar SE16 WD2500KS 250GB 7200 RPM SATA 3.0Gb/s Hard Drive
80 gb ide hd
nVidia 7800 gt video pci-e
BIOSTAR TForce4SLI Socket 939 NVIDIA nForce4 SLI ATX AMD Motherboard
AMD 64 x2 3800
600 watt power supply

Saturday: The first couple of times it shut down, I got the BSOD. I was playing WoW the first time, then using Paint Shot Pro the second. Of course, the message was near worthless - it referred to a bad IRQ setting. I rolled back the newly-installed mouse drivers.

The 3rd time the screen completely froze. I had to perform a hard reboot. Again, no error message in the log.

The next 2-3 times the pc just shut off with no warning, like someone flipped the power switch. The first time I was surfing, the second time the PC was sitting idle.

I checked the event log, and there was nothing recorded. I did a memory dump, and there was nothing recorded. I checked the cpu temp, it was fine.

Then it shut down without warning about 5 or 6 times on Sunday.

I then checked the hard drive (chkdsk), reformatted, reinstalled windows.

Here's what I have discovered:

It's not overheating (?). The cpu runs at about 52 under full load. The HDs are 22-30c under load. I have a front panel that monitors the temp, and the CPU is set to shut down when it hits 75c

It's not the cpu fan stopping or sizing up. I used a fan throttling utility by nVidia to test it and it was fine (4400 rpm).
The case has two 120mm fans with variable speed controls (front and back), a fan on the cpu. I cleared the fans of any dust and debris, and they were a little dirty, but spin freely.

It's not the ram. I used the Microsoft RAM checker and had no errors.

It's not the hard drive. I used chkdsk, fdsk to correct 2 errors, then I reformatted the SATA drive, reinstalled windows xp. twice. I installed a HD monitoring utility, the temps are great.

It's not the CD/DVD. I was able to burn backups without problems.

It's not the keyboard or mouse, both are brand new.

It's not the sound card - it's on board

It's not the ethernet/lan - it's on board

It's not the router or modem - the other pc is fine.

I had not installed anything for the last month other than new mouse drivers, which I removed. The problems continued on a clean, formatted HD. (it must be hardware related then, right?)

After all the tests, it still shuts off with no warning, no errors. There are no error beeps from the mobo when it reboots. The LED error display on the mobo says 'all systems normal"

There are no viruses or spyware, rootkits or the like.

So I'm left with:

Bad power supply
Bad video card - can a bad video card cause the machine to shut off without any warning?
Shorting out / motherboard & screws - how can this happen if the case doesn't move?

Anyone have any ideas? I've done the google search, yahoo search, newsgroup search...

Thanks guys,

xcheck

 

[DT]

Had some similar problems before and once it was the power supply, and one other time it was the video card. Either can go bad. From first glance, I'd say overheating, but you checked that out.

Have any spare power supplies you can stick in there and see if it still happens? Same with video cards...

 

[chaotika]

In my experience, and I am actually going through this right now, waiting on the new video card to get here..but it sounds like overheating. You said you checked that so my next guess is the video card or power supply or both. Usually when a computer just shuts off by itself it's due to heating issues. Maybe try a new power supply and see if that works or if you can a new video card. If neither one of those is the cause, check your processor..it's a stretch but could be the culprit.

 

[harvey]

MS RAM checker is garbage. You need to check it from from a bootable disk. Floppy or CDROM with http://www.memtest.org/

As someone who has been in the repair business for 15+ years, just a few ideas for you.

1) Power supply. It does not matter how new or what the rating is. It can be bad, and its common its bad. Use a well respected name brand. ALWAYS.

2) RAM. As I said, use Memtest. Nothing else can be trusted. Sometimes its as simple as having to swap the RAM's positions in the motherboard.

3) BIOS. Make sure you have the latest motherboard BIOS and also do not overclock until you figure out what the problem is.

4) Drivers. Make sure you have the best drivers, not always the latest drivers. Also use the base Nvidia video drivers. Not the "name brand drivers" from whoever made your card.

5) If your system found 2 errors on your HDD, then its simply going bad. You will need to replace it. HDD's almost never have sectors go bad and then just stop going bad. Chances are this is where your problem is.

 

[xcheck]

Wow, thanks for all the advice everyone.

I suspect it's the psu too. I thought I bought a reputable brand, it wasn't some cheapy one because I've heard IT pros say in the past "don't build a $2000 machine and put a $20 psu in it". What sucks is that it was a gift (lol) and non-returnable. All the cables have sleeves, lots of airflow though the case.

I will definitely try the memtest first, then throw in a new PSU.

The HDD error - well, I didn't receive any errors after those first ones on subsequent chkdsks. The drive is brand new, Western Digitial SATA. Does that make a difference?

The mobo and video card are nVidia. I used all the drivers that came with the mobo cd when I first built it, but the video drivers are at least 4 months old. It ran Vanguard with no problem for a couple of months, so that's why I'm thinking may it's not the Video Card...

The heat...could it be the PSU over heating? 2 hdds, 1 dvd, that big video card, dual core amd, 2 gb ram... Nothing else is over heating, and I've tried a few temperature proggies and they all place the cpu/case in the same range.

Regardless, I really appreciate all the help!

 

[JimJohnson]

This is a dumb question but when you get your BSD / Reboot. Have you actually gone into the bios and checked your temp?

I know you said it runs at so and so and this and that under full load. But have you actually checked in the bios what your numbers are after you just crashed?

 

[cobold]

My bet is placed on the CPU temp

I did have a similar issue which was caused by my CPU getting to warm. You might want to check you CPU temp under full load.

You can " warm up " you CPU by using Prime95 which is available at: http://www.mersenne.org/

To control the CPU temp while running windows you can use the CPUCooL
which display's the temperatures, fan speed, voltages, system values (Data input and output) and SDRAM's contents, cooling for Windows
can be found here:
http://www.cpu-cool.de/

As my CPU was getting heat issues and the system did shut down (after running the PC for 6 month with no issue) by chance. I reinstallled the CPU cooler and changed the cooling inside the case. Now I reduced the temperature by 30 degree Celsius ;-)

 

[JimJohnson]

30 degrees Celsious ;-)

You mean 30 Celsious right? like from 80C to 50C

Can not really use both at the same time.

 

[WickedMofo]

Beginning to wonder if that is my problem on my 2nd computer. Having some issues running LOTRO. Everything else runs fine -- EQ 1280x1084 resolution, same with Vanguard. Try to load LOTRO and it gets me to the server select then crashes (fades to black and then total lock up). But every 1 in 25 times of trying I can get into LOTRO. I set the settings to LOWEST graphics. Of course it is an NVIDIA card so that explains 1/2 of the problem (I hate NVIDIA with a passion). Going to try switching monitors (flat screen might be cause of it?). And also going to try replacing the Video card with an ATI 1650 PCiE card and see what happens.

I did notice the card (GEForce 6800 GT) runs very hot, so going to toss a few mini fans there too, and get the canned air. Hopefully can get back on tonight and 2 box again.

 

[cobold]

;-)
yes, I managed to reduce the CPU temp from 176 Fahrenheit down to 122 Fahrenheit and my PC shut down went away

 

[DT]

doubt it's the monitor, but *shrugs*

Try an ATI card They work well

 

[WickedMofo]

doubt it's the monitor, but *shrugs*

Try an ATI card They work well

Yeah have a ATI Radeon X1900XT card on my main computer and it hums along nicely. Helps that it takes 2 slots, and 1 of the slots is a built in FAN. So keeps itself cool all the time, and I don't hear the fan either. Ordered a similar card for the 2nd computer and am going to take a sledge hammer to the old 6800 gt tomorrow after I install the new card. Will send pictures.

 

[DT]

Well, it's fun talking about uber pc equipment when I run a POS athlon 900mhz

 

[Owned]

What Harvey said!! The locking up and the BSD are good signs of Ram issues. If you are over-clocking it more than likely is timed incorrectly and/or you have mixed ram timings. Nvidia utilities are not exactly correct, use your bios to determine true setting or temps.
Also, is this by chance a pieced box? If i was to guess with the Biostar board and the specs it is a gateway. If it is a gateway it more than likely may be the board. I have replaced many gateway BioStar boards that have been on gateway computers. All had pretty much the same symptoms as you. The shutdown and freeze up would come as soon as a game would load or when it would tax the cpu for more than 2 minutes. Just my 2cp. Good luck and i hope you get it corrected.



Also, the HDD is mechanical, the age means nothing. It could go bad at anytime for any reason. When it goes bad ..stick a fork in it. ***disclaimer*** Only if the power supply is disconnected prior to inserting prongs.

 

[hoppy96pgtbs]

( o Y o )

Had same problem on my PC. Turned out to be the Video card. It was just randomly turning off until one day it wouldnt turn back on. The VRam went bad in the card. Good luck

 

[xcheck]

This is a dumb question but when you get your BSD / Reboot. Have you actually gone into the bios and checked your temp?

I know you said it runs at so and so and this and that under full load. But have you actually checked in the bios what your numbers are after you just crashed?

Not dumb at all. Actually yes, once; it was about the same as my front panel, no more than 52C. I played WoW at max settings for about 30 minutes, in Org, tabbed out, checked the readings from the front panel and the nVidia utility, both said 55c. That's why i'm so puzzled...


cobold - I'll have to try those out, thanks man!

What Harvey said!! The locking up and the BSD are good signs of Ram issues. If you are over-clocking it more than likely is timed incorrectly and/or you have mixed ram timings. Nvidia utilities are not exactly correct, use your bios to determine true setting or temps.
Also, is this by chance a pieced box? If i was to guess with the Biostar board and the specs it is a gateway. If it is a gateway it more than likely may be the board. I have replaced many gateway BioStar boards that have been on gateway computers. All had pretty much the same symptoms as you. The shutdown and freeze up would come as soon as a game would load or when it would tax the cpu for more than 2 minutes. Just my 2cp. Good luck and i hope you get it corrected.

Also, the HDD is mechanical, the age means nothing. It could go bad at anytime for any reason. When it goes bad ..stick a fork in it. ***disclaimer*** Only if the power supply is disconnected prior to inserting prongs.

The ram is matching, brand new 1gb sticks. I have already burned an iso of that memtest, will have a go when i get home. I do not overclock, I've never really had to, and I don't know enough about it.

The pc is completely custom built. Thermaltake case, biostar mobo, amd, Kingston Ram (i think), nVidia video card. I hope to god it's not the mobo, newegg doesn't carry it anymore.

HDD--will you always hear that clicking sound before it goes bad? Maybe I should try another HD to see if that's the problem. Thanksfully I made one of those slipstream Windows Xp cds, so installation takes 20 minutes.

I'm gonna do the ram test first, then try updating the bios, then try removing the secondary HD and the CD/DVD burner. If it still crashes, i'll get a new PSU, then mobo..

Thanks! Thanks everyone...I hope I figure it out.

 

[Owned]

If by chance it is your mobo you could replace it with ABIT KN8 sli. 939 pin and could upgrade to a dual core with base bios...also i think that comes with a flash bio's update utility. I love the board as far as 939 goes. Easy install and cheap.

 

[Wragusa]

It sounds like it is overheating.
Run PRIME95 on it overnight.

 

[harvey]

Not dumb at all. Actually yes, once; it was about the same as my front panel, no more than 52C. I played WoW at max settings for about 30 minutes, in Org, tabbed out, checked the readings from the front panel and the nVidia utility, both said 55c. That's why i'm so puzzled...


cobold - I'll have to try those out, thanks man!



The ram is matching, brand new 1gb sticks. I have already burned an iso of that memtest, will have a go when i get home. I do not overclock, I've never really had to, and I don't know enough about it.

The pc is completely custom built. Thermaltake case, biostar mobo, amd, Kingston Ram (i think), nVidia video card. I hope to god it's not the mobo, newegg doesn't carry it anymore.

HDD--will you always hear that clicking sound before it goes bad? Maybe I should try another HD to see if that's the problem. Thanksfully I made one of those slipstream Windows Xp cds, so installation takes 20 minutes.

I'm gonna do the ram test first, then try updating the bios, then try removing the secondary HD and the CD/DVD burner. If it still crashes, i'll get a new PSU, then mobo..

Thanks! Thanks everyone...I hope I figure it out.

For HDD issues you will not always hear it going bad. Also a new drive should NEVER shows errors. I am wondering if it was dropped hard enough during shipping that it messed it up and is having intermittent issues.

Anyway, the order of attack you listed is gtg.

 

[Coffee]

Any time the PC turns itself off, which was stated - regardless of "sometimes bsod" or "sometimes freeze", there are only three options, and its almost always the first two:

1. Bad Power Supply Unit

-PSU in a PC is the highest case of "you get what you pay for". Its also the single component that can cause every other component in your PC to fail, either directly damaging the equipment by allowing a surge through, or slowly degrading it overtime causing minor errors and fuckups all around by allowing a slow stream of improper voltage to certain areas or everywhere over time.

Firstly - get yourself a good tester. Even a shitty tester can determine the differences in the voltages. A few examples of "affordable" ones being:

http://www.xoxide.com/coolmax-24-pin-power-supply-tester.html
*i use this one now

http://www.geeks.com/details.asp?invtid=ATX-2024-PT&cat=PWR
*i used this one as a tech for over 2 years, never failed me

http://www.atxpowersupplies.com/powmax-atx-power-supply-tester.htm
*another example

These are super easy to use. Unplug your PSU from your mobo+devices and hook it up to the tester, turn it on. Most use the common green for good, red/yellow for bad and break it down to show you which areas are failing.

When it comes to your power supply - don't cheap out. Spending the $100-200 (or for those with big budgets, $400-600) is worth it because a $20-40 one can end up costing you the pricetag of your video card, ram, and motherboard over time. Once you break the $100 mark, the usually all come with lifetime warranties, or at the very least 5 years. A few manuf.'s also offer assurance policies that they will replace any hardware damaged behind their power supply firewall. You won't see anything short of a 30day warranty on the cheap ones. Invest with confidence here.


2. Heat / Dust

He said that the case is dust free and there are many fans, but that doesn't ensure that there are no heat issues. There are several good examples of what to do here in this thread to properly monitor your CPU temp. Reading the digits on the front of the case or checking the BIOS are not always accurate at times caused by spiking. Its unlikely the CPU given what you said, so I'd put the money on the video card is overheating if these lockups and freezes only occur when you are running 3d games.

There were a few suggestions on good cards. Don't be afraid to buy a custom fan and replace the garbage stock ones that come with the cards. Even some of the $400+ cards come with terrible fans. Invest in a good replacement fan and they will provide easy to follow instructions on how to crack off the old one and put a high quality one on in its place. The cheap ones will send you a small piece of paper with just pictures and no words, and if one of those comes in the box - don't put it on your card, it won't help much.