Computer shuts down abruptly

My new pc usually works but it just shuts down while I play games 3d intensive.
Not always, after a few minutes only with some games.
I thought it was temperature related but I checked, the only thing getting hotter is one of the 2 sli graphic cards but it doesn't reach the limit.
Once the pc even shut down with a game in background and just while I clicked the logging icon in the nvidia temp log program, and the video card was also quite cold.
As far as games go, it happened with serious sam 2 and trackmania nations.
I opened the case but this doesn't change anything since it's not the temperature.
I also checked fans and voltages, everything looks ok.
After it shuts down I have to switch it off from the power source and back on and wait a while before it is able to restart.

The abit mobo has a led with a code, the mobo is still on when it shuts down. The code is 8.2 when it shuts down.

Cables and connections are all good (including the video cards additional power plugs) or it wouldn't happen only with games.

This is my rig:

Enermax 600W Al055 Noisetaker Sli enabled
Athlon fx-60
2 512 Mb 7800gtx xxx edition from xfx
2 gb ram corsair
6 raid hard drives (4 maxtor diamondmax 10 250 gb sata2, 2 western digital raptor 36)
lg dvd writer
sound blaster x-fi
ABIT an8 32x mobo
Lian-Li Big Tower Silent 73slb

No overclocking!

Please I need urgent help.
Daniele BrunengoIT Consultant, Web DesignerAsked:
Who is Participating?
 
MarkConnect With a Mentor Commented:
If you disable SLI and remove one card can this error be reproduced?
0
 
tdlorimerCommented:
I had a similar problem with my alienware desktop. I ended up downloading and paying for Spyware Doctor and Registry Mechanic which seems to have fixed the problem (at least for now). It surprised me when it found over 300 spyware products (which fee AdAware did not find). I would recommend at least running one of the free ad removal softwares and make sure your virus scanner is upto date. I'm running McAfee but again there are others available.

Spyware Doctor and Registry Mechanic - http://www.pctools.com/

Tom
0
 
Daniele BrunengoIT Consultant, Web DesignerAuthor Commented:
I use nod32 and it's updated and I check for spyware continually, I use spyware doctor, adware and ewido.
Was it games only for you also?

Anyway I have new info, it happened some minutes ago: in Serious Sam 2 I had put a weigth on the keyboard to make it rotate while I was eating. I checked several times, no problem. Then I came back and it seemed to be freezed. I moved the mouse, nothing. Then suddendly there was a beep from the case, all was functioning again, I removed the weigth and hit alt-tab and the system shut down.

I checked in the events log and it said there was a problem with i8042prt.
The description was (rough translation) "exceeded capacity in the circular buffer which memorizes data coming from the MOUSE (buffer dimension configurable in system registry)"

Actually every time it shut down I was using the ps2 mouse, I'm now trying to use it under usb but I have very few hopes.

Could it be a power supply problem? Maybe all the stuff I have needs more than 600W and when I run intensive applications the system requires more watts and maybe that's when it shuts down. Just a thougth, could it be?
0
Worried about phishing attacks?

90% of attacks start with a phish. It’s critical that IT admins and MSSPs have the right security in place to protect their end users from these phishing attacks. Check out our latest feature brief for tips and tricks to keep your employees off a hackers line!

 
tmj883Connect With a Mentor Commented:
http://www.extreme.outervision.com/psucalculator.jsp
Although the Enermax is a fine power supply, it appears that your "monster" could suck all the juice from Niagra Falls. Just the basics give over 700Watts.
This is most likely the problem as you stated.
T
http://www.sharkcageonline.com/power.asp
0
 
computerfixinsConnect With a Mentor Commented:
Sure sounds like a p/s

Keep in mind ohm law,  the hotter your "rig" runs the less efficent the power supply is.  A 15 degree difference can drain as much as a 100 watts.

First try
Download a stress test util of your choice, sandra, 3dmax, etc and try to reproduce issue.

A good test is too dis-connect the un-needed perphials, like spark mentioned with vid card.  Also try un-hooking your extra hdd's, etc.

Run stress test again...

Also could be overheating / buggy ram.  Might want run some low level tool to check memory.






0
 
Daniele BrunengoIT Consultant, Web DesignerAuthor Commented:
I tried disconnecting 2 sata hard disks and I could play much longer, more than half an hour serious sam 2. Then the pc shut down when I hit alt-tab to check temperatures, but the temperatures seemed to be lower since the case was open. Anyway even with lower temps and 2 less disks at the end the system shut down. So my guess is there is a spike when the system gets too hot and so the p/s shuts down.

I tried stress testing the system by running prime95, fastpi and cpuburn at the same time together with a nvidia demo made for 7800gtx. The temps got higher and higher til it shut down. Anyway with just prime95, fastpi and cpuburn the system didn't shut down so it looks like it must be a graphics intensive application.

Still have to try with disconnecting a graphic card and check memory at low level.
0
 
tmj883Commented:
Check the distribution of your 12V power rails...see that things like your graphics cards and HDDs are on different power bundles, so as to not overload an individual rail. T
0
 
scrathcyboyCommented:
If the system randomly shuts off at any stage, but especially when you are playing games, you will probably find out that the RAM REFRESH speed (i.e. the CAS latency of the RAM) is too slow for your motherboard.  This has nothing to do with PC3200 or PC2700, it is expressed as CL2.0, CL2.5 or CL3.  I suspect your RAM is CL3 and that is MUCH too slow for the NVIDIA chipset and fast graphics cards,  it needs to be CL2.5 and some souped up VGA cards need CL2.0 on system RAM, otherwise the card sucks instructions before the system RAM has refeshed, and they system freezes or reboots.  Sadly, a bad windows configuration can cause the same problem.  i.e. a full reinstall.
0
 
computerfixinsCommented:
gpu's overheating?  

I've heard rumors that the 7800's run a little warm.

Might try a slot cooler.
http://www.atacom.com/program/atacom.cgi?cart_id=6364135_24_62_32_210&USER_ID=www&PAGE_LIST=video_acc_copy

Also be sure to maxmize airflow in case, if your wiring looks like spaghetti, thats a bad thing, buy some zip ties + alot  of patience can greatly improve airflow.  Also those over-priced round ide and floppy cables make a huge difference over the big ugly gray ribbons.  
http://www.coolerguys.com/idecables.html
http://www.coolerguys.com/cablemangement.html

Also be sure you have all fans poiting in right directions.  (make sure all fan slots are filled)

I usaully put the front fan so it sucks air in, usaully a 120mm.  Then I equip the back two 80mm going out.

Also might want to verfiy that power-strip / wallwart / ups / etc isnt faulty, just as a sanity check.  Plug computer directly into wall if possible.
0
 
Daniele BrunengoIT Consultant, Web DesignerAuthor Commented:
To scratchyboy: the problem is not the system freezing or rebooting but simply shutting down with no warning. It all loses power except for the p/s fan and the mobo.
The ram is corsair guaranteed cl 2.0 and configured as such in the bios.
I'm still waiting to take out one graphics card and reconnect the plugs in another order, trouble is I've just been to hospital for a hernia and I have to get someone expert enough to do it for me...
0
 
Daniele BrunengoIT Consultant, Web DesignerAuthor Commented:
To computerfixins: can't use a slot cooler there's no space between the two enormous graphic cards. This is surely part of the problem: they're so big they almost touch so one of them gets quite hotter than the other (the one with the fan pointing towards the back of the other card).
I have no flat ide cables, just a floppy cable and a round dvdrw cable.
The fans are pointing in the right directions and I've already tried changing the power plug...

I was thinking about trying to place a fan on the side panel, to push out the hot air forming between the two cards.
0
 
matt_wardCommented:
Youve not got any BIOS caching or VGA Caching enabled in your bios?

If you have switch them off.
0
 
Daniele BrunengoIT Consultant, Web DesignerAuthor Commented:
Some news, I used a ram diagnose program, it freezed so I reset the cmos jumper and reconfigured the bios from scratch, then I was able to run 3dmark06 at 1600x1200 and cpuburn at temperatures of 60 C (cpu), 70 C (p/s), >76 C (one graphic card). They ran for four hours with no problem.

BUT then I played serious sam 2 for half an hour. The pc didn't shutdown but it restarted with no warning message. Anyway this must be something different because it put a minidump file in the windows directory.

I don't know if someone among you knows what to do with it, anyway here's the minidump file:

http://www.matefisica.it/mini.zip
0
 
Daniele BrunengoIT Consultant, Web DesignerAuthor Commented:
Looks like the problem changed from "pc shuts down" to "system resets" or "game shuts down".

Need for speed most wanted shut down after a few minutes but windows stayed on, no bsod. In the event registry I found that there is no description of the event but it says to use /AUXSOURCE= flag to recover the description. Don't know what it means.

Anyway since I changed ram config in the bios and the system changed behavior can it be the ram is faulty?
0
 
MarkCommented:
Windows XP will auto restart if it BSOD's. To turn this off Click -->START-->RUN type in sysdm.cpl click -->OK
this will take you to --> SYSTEMS PROPERTIES in here click on the -->ADVANCED tab In here go to the STARTUP AND RECOVERY section and click -->SETTINGS
under the SYSTEM FAILURE section uncheck AUTOMATICALLY RESTART box.

This will allow the BSOD to stay on the screen if the computer restarts.
What memory diagnostic did you use. Memtest is stand alone from floppy and is highly regarded http://www.memtest86.com/
I also think you should check your power consumption http://www.extreme.outervision.com/psucalculator.jsp

0
 
Daniele BrunengoIT Consultant, Web DesignerAuthor Commented:
I have disabled the automatic restart as you said.

I have this error code in system registry:

error code 1000007f, parameter1 00000008, parameter2 80042000, parameter3 00000000, parameter4 00000000.

The memory diagnostic I use is goldmemory which has never failed me, the ram didn't pass the test before changing parameters in bios and now it passes the test.

I will try memtest though.

About power consumption I'm following all other available roads now because right now I can't put my hands inside the pc (hernia), I will look into it if all else fails.
0
 
computerfixinsCommented:
You can verify that the ram you bought is compatible buy going here.

http://www.corsairmemory.com/corsair/configurator_search.html

I would be still more inclined to think it was a p/s.

Some good articles on this site.

http://www.pcpowercooling.com/technology/myths/
0
 
Daniele BrunengoIT Consultant, Web DesignerAuthor Commented:
Memory is compatible, but the fact remains that since I changed some parameters the system doesn't shut down anymore, it just reboots or the game crashes.

Anyway I think it is either ram or p/s from all you guys have told me, I'm going to tinker some more with the ram until a friend of mine comes to help me remove one graphic card.
0
 
MarkCommented:
The link I posted for the power consumption is only a usage calculator. No need to open the box, as you put your hardware components in the appropriate areas and it calculates how big a power supply you need.I did a preliminary calculation with the component info you gave in your question and came up with 703watts, but I was guessing on a couple of the items(#of ram sticks, exact voltage on CPU,etc) and probably left some out.Heres that link again.
http://www.extreme.outervision.com/psucalculator.jsp
0
 
Daniele BrunengoIT Consultant, Web DesignerAuthor Commented:
Yeah I know, sorry. I've tried the link and actually I come up with over 600W too but I tried removing 2 hard disks to no avail. I was talking about opening the box for removing one graphic card, I think it would be the best way to make sure it's p/s related but right now I can't.

Anyway the ram has something wrong, goldmemory is giving me errors, still have to understand if it's faulty or there's something else. I asked on corsair forums they said the default voltage my mobo gives (2.6) is too low and my ram needs 2.7x. Now I'm testing with this new voltage.
0
 
computerfixinsCommented:
Voltage and temps are tied together, what they are really saying is your memory is running to hot (or ineffienct).  Increasing voltage is kind of a scary thing because your basically throwing gasoline on the fire, increasing the temps further causing more resistance, etc  but it may work, just something to keep in mind.  

The problem with the volage calculators is they don;t take tempature / marketing hype into the picture, a typical p/s will run at 50%! less wattage then there advertised claims at 50C.

0
 
Daniele BrunengoIT Consultant, Web DesignerAuthor Commented:
Looks like the problem is solved but I'm going to test for a few more days.

Here's what I've done:
-Reset the cmos jumper on the bios
-Reconfigure everything from scratch (this stopped the shutdowns but left me with games' ctd)
-Increase RAM voltage to 1.7
-Clean uninstall of nvidia drivers and reinstall of latest official drivers

I've played games all day with no problem, I'll let you know if it keeps on going...
0
 
Daniele BrunengoIT Consultant, Web DesignerAuthor Commented:
CORRECTION: the problem is not solved, just better identified.

Actually the p/s is once again the most likely culprit: when I reinstalled nvidia drivers, SLI got disabled and I didn't notice, so I kept playing games basically with just one graphic board active. It was basically the same thing as if I had removed one card. So when I enabled SLI the crashes were there once again...

So what now? Does this confirm for sure it is a p/s problem or could it still be something else? If it is a p/s problem, what is the best p/s for my system (something available in europe since I'm italian)?
0
 
matt_wardCommented:
This could also be a RAM Timing issue too... ive seen problems like this as well... Is your Motherboard Auto detecting the RAM Timings? Or are you modifying them manually?

I would recommend manually adding your RAM values and see if that helps.

It could be that it is hitting an address space and then losing the plot.
0
 
Daniele BrunengoIT Consultant, Web DesignerAuthor Commented:
I've entered them manually some days ago, they are ok... Do you think it can be ram even if it happens only with sli switched on?

By the way, I've found this:

http://www.extremetech.com/article2/0,1697,1932947,00.asp
0
 
MarkCommented:
I have read through the postings again and checked out the error message you recieved earlier.Here is the MS KB article pertaining to the error http://msdn.microsoft.com/library/default.asp?url=/library/en-us/DevTest_g/hh/DevTest_g/t05_bugs_60_6e8bf56a-1b93-454b-9cf8-ab912bbd2157.xml.asp
 
0
 
Daniele BrunengoIT Consultant, Web DesignerAuthor Commented:
0x0000007F is the same as 1000007f?
0
 
Daniele BrunengoIT Consultant, Web DesignerAuthor Commented:
Yeah I guess it is. I got a blue screen this time but the error was 0x0000008E referencing to nv4disp.dll.

I've found the article (in italian) http://support.microsoft.com/?kbid=325730&sd=RMVP but it doesn't relate because I get the error only with SLI enabled so the solution doesn't work, I've tried all possible drivers on the face of the earth.
0
 
computerfixinsCommented:
This is by far the best p/s, but a bit pricey.
http://www.pcpowercooling.com/home/

As a "sanity" check you might want to install a second copy of windows, just to rule out the possibility of a faulty install, drivers, kernel, etc...

If your running xp just install under c:\windows2, be sure NOT to format, play with paritions :P


0
 
Daniele BrunengoIT Consultant, Web DesignerAuthor Commented:
Sadly here in Italy I can't find a reseller for pcpowercooling but I found this:

FORTRON FX700-GLN, SLI/CROSSFIRE 700watt

It has 4 separate 12v supplies while I just have two on my ps

It is recommended by nvidia, here:

http://www.slizone.com/object/slizone2_build.html#certified_powersupplies

If only I had looked here before building the system... There are very few power supplies going for a 7800gtx 512Mb sli system...

0
 
computerfixinsCommented:
did you see the one on the bottom of the page?

http://www.thermaltake.com/product/Power/PurePower/w0099/w0099.asp

pretty slick,
0
 
Daniele BrunengoIT Consultant, Web DesignerAuthor Commented:
Yeah but once again no reseller...
0
 
computerfixinsCommented:
A hack way of fixing the problem would be to run two p/s's.  Good if you have a spare p/s kicking around.

You can always jump line 14 ps_on (usaully the green one) to the ps_on on your ATX connect that is hooked up to your motherboard.  

Or just ground it, then it will act like an at p/s(power on as soon as you hit the switch in the back)

http://www.techpowerup.com/articles/other/22
0
 
Daniele BrunengoIT Consultant, Web DesignerAuthor Commented:
Interesting article! Anyway the problem is not so much about wattage as current I think. I fear my p/s isn't capable to deliver the necessary amperes to the graphic cards under certain situations and that wouldn't be solved with another p/s I think...

Sadly the links from extremetech and slizone I posted above give me little hope...
0
 
computerfixinsCommented:
well watts is amps; er kind of.

amps x volts = watts

So in theory of you hooked up one p/s to your graphic cards, and one up to everything else...
0
 
Daniele BrunengoIT Consultant, Web DesignerAuthor Commented:
Actually there's a 12V cable on my p/s dedicated to graphic cards but under maximum load it will only carry more or less 20A, while it looks like the cards need even 22-23A at full load. So if the line I take from my old p/s is similar the problem is still there.
The fortron p/s I was mumbling about has four 12V cables, two of which are dedicated to one videocard each. These cables carry 15A, so the sum is 30A which should be more than enough.
0
 
Daniele BrunengoIT Consultant, Web DesignerAuthor Commented:
All right, with the new Fortron p/s all is perfect!
So I'm going to close this one, thanks guys!
0
Question has a verified solution.

Are you are experiencing a similar issue? Get a personalized answer when you ask a related question.

Have a better answer? Share it in a comment.

All Courses

From novice to tech pro — start learning today.