Strange PC problems

[WayAbvPar]

Bought this card a few weeks ago. Everything seems to run fine, for the most part. However, I have had the PC spontaneously shut down (monitor goes blank- 'no signal' indicator on monitor- need to push reset key to reboot) several times. Until yesterday, it only happened while playing EVE, and only during combat (happened yesterday with EVE too- right in the middle of a PVP encounter. Lucky me); then I had it happen while playing BF2.

My first guess is that overheating is causing this- when the card is doing a lot of work, it gets hot and shuts everything down. Does that sound accurate? If that is the case, what is the easiest solution? Will un-overclocking the card help? Do I need to buy another fan? Do they have strictly internal fans, or do I need to get a new case with extra fans on it?

If this was a PS issue I figure I would see less predictable behavior, no?

TIA for any assistance.

[Samwise]

It could be an issue of the PS not having enough wattage to run the machine under peak load - when the card starts drawing too much current, the machine loses power and shuts down.  How does your power supply compare to the card you're using?

Heat could also be an issue, though.  If you haven't already, get a can of compressed air and blow the dust out; that solves a lot of problems all on its own.  How many fans does your case currently have?  For best results you generally want one on the front and one on the back (sucking air in the front and blowing it out the back; don't have 'em both pointing in or out).  If you've only got one fan right now, check inside the case (or the manual that came with it) to see if there's a convenient spot to mount an extra fan, and what size fan goes there.  If there's no such spot, you might just wanna get a new case that has such a spot.

[Hoax]

I think Sam is dead on with the Power Supply diagnosis, you can go to some super-techy/geeky forums and find breakdowns of what you need in terms of "rails" (I find it confusing so I will not try to explain) to keep a good video card running.

[Lt.Dan]

Fark, I asked about this in the Video Card Thread and "people" said "blah blah blah, you don't need to worry about this".

I bought the unclocked version of the card so I'm interested in hearing how this works out.  According to the evga forums and newegg customer feedback underclocking may get some relief.  It might also be a psu problem - evidently you need 18A or more on the 12V rail - or a driver issue, but that's just guessing from what people say works and without them mentioning anything about their setup. 

[Samwise]

I think Sam is dead on with the Power Supply diagnosis, you can go to some super-techy/geeky forums and find breakdowns of what you need in terms of "rails" (I find it confusing so I will not try to explain) to keep a good video card running.

My technique is simpler:

1) Look at the box your video card came in.  It says "Requires at least X-watt power supply".
2) Look at the box your case/power supply/PC came in.  It says "Y-watt power supply".

If X > Y, oops.

(There are undoubtedly more complex things that can go wrong, but checking whether X > Y is the most important step IMO.)

[WayAbvPar]

I figured a PS problem would manifest itself more often, but you very well may be right. The card requires 350W minimum...I don't recall specifically picking out a PS when I put the PC together (rather, paid to have someone put it together), so I am wondering if they shorted me with like a 250W? I will check when I get home. I wonder if I can tell without opening the case?

[HaemishM]

Many of them will have some specs listed on the back of the box where you plug the cord into the power supply. If not, you'll have to crack the case to get a gander inside.

[Trippy]

It definitely sounds more like a power supply issue than a overheating one. If you can read the back of the power supply you can usually get the overall power supply rating from there. If your power supply has a printed breakdown of the power supplied to each rail that will usually be on the "top" of the power supply which would usually mean removing it from the case to read it unless your case cover is one big piece that covers the top and the sides (separately removable side panels is the more popular design these days). You can also try unplugging all the unnecessary 12V molex connectors (the same connector type you plugged into you video card) from other devices like your DVD/CD drive(s), and secondary hard drives. This might help confirm whether or not it's a lack of power that's causing the system to turn off.

You can check your NVIDIA control panel to see if it has a temperature monitor display. If not you can try installing the "coolbits" registry hack and checking again. eVGA is a good manufacturer so they should have temperature sensor chip installed on the board.

[Nazrat]

Fark, I asked about this in the Video Card Thread and "people" said "blah blah blah, you don't need to worry about this".

I bought the unclocked version of the card so I'm interested in hearing how this works out.  According to the evga forums and newegg customer feedback underclocking may get some relief.  It might also be a psu problem - evidently you need 18A or more on the 12V rail - or a driver issue, but that's just guessing from what people say works and without them mentioning anything about their setup. 

Here is my quote from the other thread:

I have a 7800GS and my screaming fast, state of the art, Dell Dimension 4450 250 watt power supply only puts out: 14 amps at +12vdc, 22 amps at +5vdc, 18 amps at +3.3vdc.

I have had zero power supply problems in the 2 weeks that I have had the card installed. 

I think someone is trying to find something to complain about. 

I am still having zero issues with the card.  It can't be my state of the art power supply.  Maybe I'm not powering my microwave and blender with the power supply so that there is more juice from the card? 

I have a CDROM drive, DVDROM drive, hard drive, video card, sound card, etc. using the same power supply. 

[Hoax]

Overheating will not cause crashes most of the time, it will just melt the freaking card at some point.  I had a card where the cheapo on-board fan had died for I have no idea how long, I burned myself on the card not knowing this it was running so hot.

I put a new fan on there but I guess the damage had been done and the card died a few months after my discovery.

Power supply stuff OTOH sucks, but isn't too hard to fix.  I forget what the PSU I bought was but I really like it I'll post back later in case you need a starting point recommendation.

[Flood]

Just to chime in:

1. Power Supply
2. Heat issues
3. Memory issues

In order from most likely to least, with memory being a long shot.

As mentioned before it's probably caused by your machine not getting enough juice while under load from gaming.  The general concensus on a minimum power req for most "casual gaming" rigs is 480 watts.  And caveat that's 480 watts from a reliable PSU maker.  I have what I guess is considered an "enthusiast" grade computer and I run a 700 watt PSU, and I don't even have an SLI rig (anymore).

If your comp was running fine, what did you change between when it was running ok and when it started giving you problems?  By process of elimination you already have named your new video card as something you changed.  Since I doubt you have a bad card (it wouldn't reboot you if it was a bad card anyway) I'd follow the advice above about checking all the power connections on your mobo and vid card (if applicable), take the side off your case to remove the question of heat, and see what happens.

*shrug* My hunch is it's your PSU, or a related power drawing issue.


Some clickies for good PSU's and an interesting link about powering your rig:

http://www.dfi-street.com/forum/showthread.php?t=29079

http://www.newegg.com/Product/Product.asp?Item=N82E16817151024

http://www.home2000.net/client/fspgroupusacom/proddetail.asp?linenumber=195


 Good luck!

 

[Trippy]

My first guess is that overheating is causing this- when the card is doing a lot of work, it gets hot and shuts everything down. Does that sound accurate? If that is the case, what is the easiest solution? Will un-overclocking the card help? Do I need to buy another fan? Do they have strictly internal fans, or do I need to get a new case with extra fans on it?

Un-overclocking may help but that's cause your card is seriously overclocked (460 MHz vs 375 MHz) and is probably drawing more power because it.

[Engels]

Overheated video cards tend to not 'reboot' your system; they just fry, give graphical glitches, and die, with your computer otherwise humming along happily as if nothing had happened.

I'd not only check PSU, but Heat issues with your CPU. I know Everest used to come with a heat sensor for both GPU and CPU, but I'm sure there's some other aftermarket software to see if your CPU is running too hot. If your CPU is Intel, it will 'reboot' on an overheat, instantly 'fixing' the problem. Intels also run hotter than AMD for the most part.

The reason you may not have seen this behavior before is that your vid card is putting out additional heat, as well as creating more of an airflow situation than you may have had before. If you do not have adequate airflow, say above 35 cfps (cubit feet per second) you may have a steamy hot jungle and the first thing to scream for mercy is an Intel CPU.

Make sure you've vaccumed the inside of your comp lately, and that all your fans are operational. I do mean all of them. There really is no skimping on this.

Finally, installing a new PSU is not a big deal at all; one of the simpler things to do. If you can afford it, spend a few extra bucks on a good brand. http://www.pcpowercooling.com/home/ have a good reputation.

[Samwise]

It is technically possible for a hot video card to shut down the system.  If nothing else, most systems have a temperature probe on the CPU that automatically shuts the system down if it gets too hot.  So if the video card was generating too much hot air that wasn't getting circulated out of the case fast enough, it could contribute to the CPU overheating and cause the system to shut down.

Still think the power supply is the most likely culprit though.

[Trippy]

If your CPU is Intel, it will 'reboot' on an overheat, instantly 'fixing' the problem. Intels also run hotter than AMD for the most part.

Pentium 4 class CPUs will *not* reboot when they overheat -- they will automatically throttle down their speed.

[Engels]

It is technically possible for a hot video card to shut down the system.  If nothing else, most systems have a temperature probe on the CPU that automatically shuts the system down if it gets too hot.  So if the video card was generating too much hot air that wasn't getting circulated out of the case fast enough, it could contribute to the CPU overheating and cause the system to shut down.

Still think the power supply is the most likely culprit though.

Pretty much what I was saying. And I didn't know that about Pentium 4 class CPUs, I stand corrected! Since I've been an AMD owner since back before Socket A, I may have just been misinformed about Intels. In any case, I heard they had some kind of safety measure AMDs didn't.

[Samwise]

If your CPU is Intel, it will 'reboot' on an overheat, instantly 'fixing' the problem. Intels also run hotter than AMD for the most part.

Pentium 4 class CPUs will *not* reboot when they overheat -- they will automatically throttle down their speed.

If throttling down the speed doesn't do the trick, though, won't they eventually give up and shut down?  My siblings had similar problems that eventually turned out to be heat-related (and were fixed with a can of compressed air).

[Trippy]

My first guess is that overheating is causing this- when the card is doing a lot of work, it gets hot and shuts everything down. Does that sound accurate? If that is the case, what is the easiest solution? Will un-overclocking the card help? Do I need to buy another fan? Do they have strictly internal fans, or do I need to get a new case with extra fans on it?

Though it is more likely to be a straight up power supply issue you can check how hot things are getting if you have the right monitoring tools.

The NVIDIA display driver comes with a simple temperature display assuming your card has a temperature monitoring sensor like I mentioned above:



Notice how there's a temperature warning check box which should give you a warning if something bad is happening and there's a threshold value which is supposed to slow down your GPU if things are getting way too hot.

If you have a decent motherboard with temperature sensor chips on it you should be able to monitor the key temps with software like this (for those that are curious my MB has a heat pipe passively cooled chipset which is why there's no fan attached):

[Trippy]

If your CPU is Intel, it will 'reboot' on an overheat, instantly 'fixing' the problem. Intels also run hotter than AMD for the most part.

Pentium 4 class CPUs will *not* reboot when they overheat -- they will automatically throttle down their speed.

If throttling down the speed doesn't do the trick, though, won't they eventually give up and shut down?

Yes it will eventually give up and shutdown but that usually indicates something catastrophic has happened like your CPU fan has stopped spinning or your heat sink somehow fell off. If you are playing a game, though, you should see your frame rate drop dramatically as the CPU throttles down before it eventually gives up.

[Trippy]

Pretty much what I was saying. And I didn't know that about Pentium 4 class CPUs, I stand corrected! Since I've been an AMD owner since back before Socket A, I may have just been misinformed about Intels. In any case, I heard they had some kind of safety measure AMDs didn't.

Yeah the Socket A CPUs don't have any built-in temperature sensor -- it is up to the motherboard to provide thermal protection which many do not. That meant you could very easily fry your Athlon/Athlon XP if you didn't install your heat sink properly and you didn't have a quick finger on the power supply off switch. Even on the MBs that do have a thermal protection, the temperature sensor is usually this little probe sticking up from the center of the CPU socket and you had to make sure it was bent upwards enough to touch the CPU packaging. Ah those were the days...

[WayAbvPar]

Thanks for all the replies so far- lots of good info here. I just got off the phone with the place I bought the case (and the rest of the parts from)- it sounds like I probably have a 380W PSU in there right now. Upgrading to an Antec 500W (the same brand I have now) will run about $110. Does that sound about right?

As far as the cleaning goes, the case got a pretty throrough cleaning when I put the card in a few weeks ago, so factor should be less likely.

Other info- I have an AMD 3200+ (IIRC); definitely not an Intel chip. The only change to the set up was an extra gig of RAM a few months ago, and the new card a few weeks ago. I have never seen the issue happen before the new card was in.

[Trippy]

Thanks for all the replies so far- lots of good info here. I just got off the phone with the place I bought the case (and the rest of the parts from)- it sounds like I probably have a 380W PSU in there right now. Upgrading to an Antec 500W (the same brand I have now) will run about $110. Does that sound about right?

Depends on which model of Antec PS is it. The "SmartPower" version is $80 at newegg:

http://www.newegg.com/Product/Product.asp?Item=N82E16817103937

The more robust 550W "TruePower" version runs $86 at newegg with rebate:

http://www.newegg.com/Product/Product.asp?Item=N82E16817103931

Other info- I have an AMD 3200+ (IIRC); definitely not an Intel chip. The only change to the set up was an extra gig of RAM a few months ago, and the new card a few weeks ago. I have never seen the issue happen before the new card was in.

The extra RAM will draw more current and depending on how the power supply is setup that may reduce the available current on the 12V line.

[Engels]

380 really should be plenty of wattage unless you're running a buncha hard drives in raid or some such. Still, the thing with PSUs is that the cheap ones have voltage fluctuations that can degrade over time and eventually you end up with the PSU sending a bit more or less voltage than you want, and that causes all sorts of havoc that's hard to identify.

Also, have you checked your 'event viewer' under administrative tools or Manage (right click on my computer)? It may have recorded the fault although that's sometimes impossible to decifer.

[Strazos]

Watch yuorself with that Asus Prode program....it can get buggy sometimes. Also, it can force reboots before anything even reaches a maximum threashold (ie - temp).

[Trippy]

Watch yuorself with that Asus Prode program....it can get buggy sometimes. Also, it can force reboots before anything even reaches a maximum threashold (ie - temp).

I don't normally have it running -- I launched it to take those screenshots but thanks for the warning. I stopped running it after I was satisfied that my new computer was running cool enough for my tastes. It also has a few annoyances like not being able to set the fan speed threshold low enough. Notice how my chassis fan is not checked. That's cause it's a 120 mm fan that's been undervolted to spin slower (and quieter) and the software won't let me set a warning level under 600 RPM.

[HaemishM]

If it IS a Power Supply issue, and I'm inclined to believe it is based on my past fuckups with PSU's, changing PSU's is really easy. It's much easier than I thought it would be. For someone who has tried and failed to install a CPU, I'd much rather change a power supply than add a new IDE drive (because IDE cables fucking suck).

[WayAbvPar]

Update- I bought one of these and got it in last night. So far so good- it runs a bit quieter now, and the case seemed a bit cooler (although I didn't get to play very long, so that could change). The real test will be when I get into a combat in Eve. Hard to do at the moment since my combat ship got assploded. Played some BF2 without issue, but I have only seen a problem with that once.

[vex]

You know I picked up that same power supply a couple of months ago and my first thought after I installed it was how quiet it was.  I've been happy with it since.

One thing you can do if you suspect cooling issues is just run it with the cover or side off and direct a fan in there if need be.

To monitor temps and fans I've been using SpeedFan, it's handy and free.

[Lt.Dan]

Well the 7800GS is installed and running.  Had a weird lockup where all I could do was bring up task manager, which showed the system idle, and eventually had to hard reboot.  Also lost my router connection and had to reset it to get the connection back.  No idea if that's related but too much of a conincidence.  Will see how stable things remain but might just go ahead with the psu upgrade anyway.

[WayAbvPar]

Latest update:
After  a few days of no problems, my PC quit again last night after several hours of gaming (Eve and BF2). This time I got a message that the CPU was overheating, and it wouldn't boot. I went to bed. This morning, I booted it up- during the boot, I got a message saying the overclocking was dsabled, and that I could either go into setup and tinker with it or boot with default levels. I booted with the default, and was able to log into Eve and change some skills.

Questions-
Will the lack of overclocking make the card run cooler? I never had heat problems before I installed the new card.
Are there any internal fans that can be mounted on the inside of a case, or do I need to buy a new case altogether?
Why, after going to the trouble of buying a new card that is doing its level best to destroy my PC and wallet, am I not playing Oblivion more often?

[bhodi]

Latest update:
After  a few days of no problems, my PC quit again last night after several hours of gaming (Eve and BF2). This time I got a message that the CPU was overheating, and it wouldn't boot. I went to bed. This morning, I booted it up- during the boot, I got a message saying the overclocking was disabled, and that I could either go into setup and tinker with it or boot with default levels. I booted with the default, and was able to log into Eve and change some skills.

Questions-
Will the lack of overclocking make the card run cooler? I never had heat problems before I installed the new card.
Are there any internal fans that can be mounted on the inside of a case, or do I need to buy a new case altogether?
Why, after going to the trouble of buying a new card that is doing its level best to destroy my PC and wallet, am I not playing Oblivion more often?

1. Yes. Overclocking is running the card at a higher frequency. This requires more power. The amount overclocked however isn't great and you won't be able to get more than a few degrees of coolness. What it DOES do however is reduce the tolerances so that your card can run normally at a higher temperature.

2. Most cases have conspicuous "fan holes" in the front and the back where you can easily mount off the shelf fans. If you open the case up it's pretty obvious where they go. Fans generally come in 2 standard sizes and look like mini box fans. They come with their own 4 mounting screws and can really take the heat pressure off, assuming you have decent airflow out the back. Don't put them in backwards and remember air generally flows front->back.

3. Because oblivion sucks and is lonely.

[HaemishM]

If you have a PCI slot available, they do make PCI-slot case fans.

[Yegolev]

I have great success with leaving my case open.  It is noisy but I prefer that to overheating or paying for extra fans.

[WayAbvPar]

I have great success with leaving my case open.  It is noisy but I prefer that to overheating or paying for extra fans.

Hmmm. I do have a spring loaded door on my case...I might try leaving it crack open and see if that has any effect. It sounds like my card decided to un-overclock itself, so the combination of the two might keep me from making ANOTHER run to the computer store. Until it doesn't work and fries my goddamned CPU.

[Moaner]

I have an eVGA 9800gt and have ran into some overheating problems.  It turns out that fan housing works quite well as a dust collector, which blocks the flow of air to the heat sink.  Considering your card is brand new I would assume this is not your problem, but for future reference you are going to want to remove that plate and either vacuum or blow the dust and hair out of the housing.  It's also something to consider if you are goint to leave your case open.  I ended up closing mine after years of not having a cover on in hopes of keeping dust/hair out of my fan housing.  It's really quite a stupid design.