a desktop PC mystery, identify the failing component

[jeffw_00]

I have a well-kept (built myself, keep TSRs to a minimum) white-box WIN XP machine (XP home SP3). HW is an ASUS P4P800SE MB with a 3GHz dual-threaded Intel P4 processor. Network and audio are integral. Video card is a knockoff EVGA 7600GT.. Power supply is upgraded to 500W. AVG free and Spybot run every day.My kids use it to play games. On certain games (but not others) after they play for a while (10-min to an hour), a BSOD quickly flashes and the machine reboots, before windows loads the BSOD flashes again and it reboots again, and it loops on this pattern. Right after this happened, I disconnected the power connector from the video card and the machine booted fine.Most times, when this happens, we just turn the machine off for a few hours and that also fixes it fine.However, replacing the video card had no effect.Replacing the power supply had no effect.It happens on "Oblivion" and "rollercoaster Tycoon (original version").It doesn't happen on "Eve", "City of Heroes" (even after hours of play). The machine also gets some use for business purposes (scanning, data entry), and has no failures in this usage.It's on 24/7 and never fails when idle. I've vacuumed it out, and all the fans are running fine. I'd prefer that the AGP socket grab a little tighter, but the screw keeps the card solidly in place. I thought sure that swapping the video card or the power supply would fix it. I have nothing else to swap. any ideas?Thanks!/j

[Peachy]

A couple of things to check.1) memory. Run memtest+2) CPU heatsink/fan. Make sure it's seated properly with the proper amount of thermal paste3) run prime95 and check for system overheating. This will often unmask bad memory.4) make sure you have the latest graphics card drivers installed.

[jeffw_00]

A couple of things to check.1) memory. Run memtest+2) CPU heatsink/fan. Make sure it's seated properly with the proper amount of thermal paste3) run prime95 and check for system overheating. This will often unmask bad memory.4) make sure you have the latest graphics card drivers installed.

Hi Peachy....1) I forgot to mention that I disabled quick-boot and ran the extended memory boot test for several days - never a problem. I -could- replace all the memory (i have extra SIMMs).2) graphics driver - good idea - there was a newer one from nVidia with "bugs fixed" - maybe that will help.3) overheating - if it was a heat problem, why would disconnecting the power connector from the video card fix it instantly? (kind of rules out memory issue too?)thanks!/j Edited June 27, 2009 by jeffw_00

[kkehoe]

Having the Stop error code(s) might help a lot. Disable auto-restart in System Properties > Advanced > Startup and Recovery.Kevin

[jeffw_00]

good idea!

[Gus K]

3) overheating - if it was a heat problem, why would disconnecting the power connector from the video card fix it instantly? (kind of rules out memory issue too?)

When trouble shooting it's best to not assume and start with the basics rather than over think. It certainly sounds like a heat issue, I'd monitor the temps to get that out of the way. Certainly easy enough to do.

[jeffw_00]

When trouble shooting it's best to not assume and start with the basics rather than over think. It certainly sounds like a heat issue, I'd monitor the temps to get that out of the way. Certainly easy enough to do.

Ok, I'm not explaining myself well. I have it failing, repeatedly. I quickly shut it down, disconnect the video power connector, reboot, and it's fine. I do again (reconnecting the power), and it fails. If it was overheating why would it have booted fine when I disconnected the video connector?/j

[Gus K]

If it was overheating why would it have booted fine when I disconnected the video connector?

For me checking temps would be the first step, cpu, case and gpu. I'd want to know for sure. The power plug issue might also be heat related. You'll never know if you don't check first.

Most times, when this happens, we just turn the machine off for a few hours and that also fixes it fine.

That's a classic heat issue temporary fix... you are letting it cool down.Good Luck

[jolphil]

Hi,This also sounds to me a classic overheating problem but my thoughts are to the power supply(even the one you replaced)..After reading power supply reviews from New Egg and this website for Power supply reviews..(http://www.hardwaresecrets.com/page/power).. I have always thought that one was like another..But..This is not so...What I have read is that just Because you put in a higher wattage supply will not only not cure a problem but in fact cause one..Many symptoms are BSOD's when playing games..Some PS makers flat out lie about specs and omit some specs that are critical..Some rate at 25 deg. C but your case temp is usually 40 to 50 deg. C when the computer is working hard..My advice would be to do as suggested by first monitoring heat...Even to the extent of opening the case and blowing a fan on the computer to determine if it is a heat related issue or not..You can also monitor the temperature of the power supply exhaust to determine what it is a idle and at game playing..I don't know exactly what your problem is but my guess it's like others mentioned..A heat related problem..Good luck in troubleshooting.jolphil

[jeffw_00]

thanks guys - drivers seems to have made a difference. I set it up to freeze on the next BSOD and will report. if it fails again it makes sense to look at the heat issue (maybe start by taking off the skins and seeing if it still recurs. I never liked this particular case, it always seemed a bit light on venting - On the other hand - there's not much in the box - one hard disk, a fanless video card, and the MOBO...). waiting to see if it fails againthanks for all the comments!/jawfully weird though, that I can "cool" it instantly by disconnecting the power from the video card...

Edited June 29, 2009 by jeffw_00

[Peachy]

I'm surprised that a fanless video card even has a power connector. Generally it doesn't need one if the chip is cool enough to only have a heatsink. Remember that by adding power to the video card you are also making it work more, hence it generates more heat. And since your video card has no fan I'm guessing the GPU is overheating.

[jeffw_00]

So what I'm using for the VideoCard is an EVGA (or similar no-name) I got off of newEgg that has nVidia 7600GT chip. Point is, I have 2 of these that are very similar, but one has a fan and one is fanless, and regardless of which is plugged in the failure mode is the same. I use a lot of fanless cards and don't agree with the assumption "oh, no fan, so it must be overheating"/j

Edited June 29, 2009 by jeffw_00

[Gus K]

I'm surprised that a fanless video card even has a power connector. Generally it doesn't need one if the chip is cool enough to only have a heatsink. Remember that by adding power to the video card you are also making it work more, hence it generates more heat. And since your video card has no fan I'm guessing the GPU is overheating.

And fanless vid cards are not made for gaming. Best used for normal, non cpu intensive, computer use.

[jeffw_00]

Hi Gus, while I appreciate all your posts, they do confuse me. Why would someone manufacture a card based on nVidia 7600GT chip and not expect people to use it for gaming?How 'bout we just wait and see if it fails again, before this post gets too long in philosophical issues? 8-}/j

[jeffw_00]

Ok - it failed again (in the middle of the night). This is the first time it failed when no one was using it (it might have been running a backup). It froze on BSOD as desired (I wrote it down) and this morning pulling the video power connector doesn't help, needs to cool down, I think. BSOD reads in part.."A problem has been detected....." IRQL_NOT_LESS_OR EQUAL"if it's the first time....."STOP: 0x0000000A (0xF773F980,0x000000FF,0x00000000,0x804DCBF5)I guess after it cools down I'll run it without one skin for a while, that should drop the internal temp significantly, I think? And then see if it fails again. if not, I need to add another (or bigger) fan....If there's an easy way to monitor temp when it fails - I can try that too. Thanks guys/j

Edited July 3, 2009 by jeffw_00

[crp]

http://www.resplendence.com/whocrashedmight help

[jolphil]

Just a top of the head thought..I have been toying around with the purchase of a IR temperature probe..That way I can examine the CPU Heatsink temp, and in your case,the Video card Heat sink temp. You can even check you MoBo's chipset temp or the hard drive temp,as well a memory stick.The beauty is that it can give some indicationof hotspots without actually touching anything.. I have seen them offered for about 65$US.. It may be a costly one time purchase if you do not have use for it anywhere else around your home but I work on my own car,truck ,tractor and AC unit not to mention my 4 computers so in my case it may be worth it.. Now back to your elusive problem...I think you are taking the right approach to find the problem..Even if you just carefully touch each suspected item (one hand on the chassis) to see if anything is extremely hot will give you a clue..If nothing is very hot, then I would look at the PS output voltages to see if they are reasonable..As mentioned you are on the right track..Good luck,jolphil

[striker]

Ok - it failed again (in the middle of the night). This is the first time it failed when no one was using it (it might have been running a backup). It froze on BSOD as desired (I wrote it down) and this morning pulling the video power connector doesn't help, needs to cool down, I think. BSOD reads in part.."A problem has been detected....." IRQL_NOT_LESS_OR EQUAL"if it's the first time....."STOP: 0x0000000A (0xF773F980,0x000000FF,0x00000000,0x804DCBF5)I guess after it cools down I'll run it without one skin for a while, that should drop the internal temp significantly, I think? And then see if it fails again. if not, I need to add another (or bigger) fan....If there's an easy way to monitor temp when it fails - I can try that too. Thanks guys/j

Maybe these can be of help ( I hope) :http://support.microsoft.com/kb/314063http://support.microsoft.com/search/defaul...log=LCID%3D1033

[jeffw_00]

Thanks - they actually suggest....an overheating problem ! 8-}(problem with 3rd party SW or HW)I think that's where this is going to end up. Unfortunately, the power-connector fix took me down the wrong paththanks/j

Edited July 5, 2009 by jeffw_00

[tommyj12]

There are several utilities to check cpu and gpu temps.(rivatuner, or evga precision) (core temp, speed fan) I would put it under full load and monitor with them. Two other things I would do first is update the mobo chipset drivers, and lastly update to the latest directx. Since it is only a few games that cause it are they fully patched? I am betting a DX update will fix it if so.

Edited July 5, 2009 by tommyj12

[jeffw_00]

Machine is current. Most offending game is "oblivion".

[jeffw_00]

curiouser and curiouser. For the last 2 weeks, it's failed exactly once a week, sometime in the middle of the night, on Thursday night. During the 2nd week, the skins were off. Now, thursday night is when the chkdsk, reboot, and then full backup automatically run, but when, on Friday evening, I see the BSOD (same message), turn it off, wait no more than 3-4 minutes, turn it on, it runs fine, and I can run the chkdsk and backup (ImageForWindows2) without incident.still a mystery, I think.../j

Edited July 11, 2009 by jeffw_00

[jolphil]

Hello again,I have often wondered why i got the BSOD on my computers..I did not realize what went on inside with regard to the specific computer and the software that was running at any instant until I read this article..HereThis gives a spread of different computers from an ordinary office box to a high end gaming computer..The first few pages gives the methodology used but then shows in each case what happens when different programs were run and the demands it puts on the computer..Will this solve your problem? Maybe not, but it will start you thinking on how and what to do ..I for one found it very enlightening..Jolphil

Edited July 11, 2009 by jolphil

[Ed_P]

It could be due to a drop in the house's voltage at that hour, either something happening in the house or at the utility company. The freezer kicks on at the same time the refrigerator kicks on at the same time the A/C kicks on at the same time the hot water heater kicks on at the same time the heater in the fish tank kicks on etc. Or the power company switches distribution lines or is working on the service or etc.Unfortunately you will need a volt meter than supports recording the readings to track it.It sounds like you leave the pc on 24/7 which is a concept I've never understood. People don't leave their lights on 24/7 or their tv on 24/7 or their car running 24/7 so why the pc? Most illogical.

[Specmon]

Jeff,Your EVGA 7600GT card has one of the GPUs that were involved in the huge defective solder bump mess that nVidia got itself into. Almost everything they made over a couple year period was involved. Heat related failure rates for cards that were in notebooks and stressed cards in gaming rigs like yours were ( and continue to be) very high. nVidia has already lost more than $200 million because of this. I remember that when the issue was fully realized, it was thought it could cost nVidia billions or even cost the company. Naturally nVidia has done everything it could to resist replacing all defective parts, (as these parts span ALL production over a couple year period) but there are still tons of these GPUs out there, and they continue to fail.My son runs a gaming rig with two SLI 8800GT cards and he had the same problem you have. It started infrequently, but become more troublesome. It took many hours of frustration to isolate the problem and then determine which card it was. We knew when we bought them that this problem could arise, but PNY sold the cards with a 5 year warranty.But you can't assume because your problem persists after you exchange one 7600 card with a different one, that the problem is NOT the GPU on the original card. Almost every 7600 GPU can be made to fail under stress. They were ALL made with defective solder

Edited July 11, 2009 by Specmon

[jeffw_00]

Hi Ed1) I doubt it's a voltage issue. Ever since I showed the pwr company how -bad- our voltage was a few years back, which resulted in new service to the neighborhood, we're well in spec all the time, and we have very little that 'kicks in' and when it does it's all new energy-efficient stuff. 2) I leave them on 24/7 because a) with monitor powered down, draw is minimalB) I do maintenance tasks (backups, virus/malware scan, defrag, etc) at nightc) lifespan of electronics, fans, disk drives,etc is somewhat proportional to number of on/off cycles.my theory is that somehow the stuff that the kids play gets the machine into some weird state, and then the scandisk or backup triggers the fault..../j

[jeffw_00]

interesting Specmon. I can argue it both ways.1) I had both cards in use, in different machines, for over a year before the product started. But the 2nd card only started failing when I swapped it for the first one, after the first one failed. 2) when the kids go back to school in a month, I could swap in a lower-end card (or use the on-board video) and see if the problem goes away - might not be a bad thing to try.However, previously, I swapped cards only once, and only briefly. Perhaps now that i have this 'weekly failure' thing I'll swap again and leave swapped and see if at least behavior changes..../j

[Ed_P]

c) lifespan of electronics, fans, disk drives,etc is somewhat proportional to number of on/off cycles.

See that's the part that always gets me, if people believe that then they would leave their monitors on, their tvs on, their fans on, even their cars on 24/7 also, but they don't. The logic is only applied to their pcs and that to me is illogical.As for doing maintenance in the off hours that I understand, and pcs I set up to do that are also set up to shutdown when they are finished their maintenance tasks.I'm not trying to argue with you, or change your mind, I'm just expressing a personal feeling and it's based on shutting machines off that are not in use, for many years, and not having the problems you describe.

[jeffw_00]

I do power down the monitors. But since I have backups and maintenance going, use 1 machine for home control. one machine receives VMs, etc - just easier to leave them all up.../j

[goretsky]

Hello,Can you move any of the internal expansion cards and wiring away from the video card so that there is more space for air to flow around it? Also, is it possible to mount a fan internally that blows air over the video card's component side?Regards,Aryeh Goretsky