Upgraded graphic card now PC shuts off.

[RChappo]

Hello,

I have a Gateway DX4800 desktop PC (Pentium Dual Core E5200, 6GB DDR RAM) and today I installed a Galaxy GeForce 210 1GB DDR2 graphics card to upgrade from the integrated graphics card. I disabled the integrated graphics in Device Manager and downloaed the latest drivers for the card. Now my PC is shutting down sometimes whenever I watch a video or play a game. It would do this this with the integrated graphics if I was moving around a lot on Google Earth or playing Football Manager (which is why I upgraded). I can see that the graphics run smoother when the game is on and they don't stutter like they did when it would cut out before. I had a look on Speedfan and the GPU temp seems to be ok when it goes off (it was at 53 degrees C when it last shut off). I figured that seeing as the new card has a built in fan overheating the GPU wouldn't be the issue - I'm not a heavy gamer. Is this a PSU thing? The PSU on the PC is 300W - the graphics card box recommends 300W or higher, but am I maxed out and should I replace the PSU? Or is it something I forgot to do in the setup that's making this happen. Or did I buy the wrong graphics card for this system. Would appreciate some advice. Thanks!

 

[has22fas]

1) Either Heat issue with CPU/GPU
2) The power supply isnt good enough for your system on load.
3) Driver Issue download latest drivers for your card and operating system
4) Manually disable onboard GPU in BIOS

Tell me how you get on with them

 

[RChappo]

Hi has22fas...thank you for your reply :

1) I don't think that heat was an issue with either the CPU or the GPU - I used Speedfan to monitor the heat and it shut off when the GPU temp was 53 degrees. I had watched it with the integrated graphics card get to higher temps than that (ie in the 80's). I didn't see anything in the CPU temperature readings to make me suspect that that could be an problem either.

2) The box with the card recommended 300W or higher. But I do have some usb devices on this PC - external hard drive, printer, a couple of mp3 players and cameras I connect via usb etc....would this cause me to need more power?

3) I downloaded the latest drivers for my card from the nvidia website.

4) I didn't do this, but when I looked in the BIOS the only setting I could see pertaining to video/graphics was set to 'PCI' - the card I bought was installed in a PCI-E slot. I did disable the integrated graphics card in Device Manager though.

As the evening went on the PC shut off a little more often - even when just browsing the web and this forum. But during the day I left the PC on idle for a couple of hours and it seemed fine - I went out and my wife did some web browsing on it with no problems. I'm back on my integrated graphics card now for the time being.


Again thanks for your reply and patience...I'm pretty much a novice when it comes to this stuff...

 

[jaguarmatt]

Make sure that you have the card firmly in place.
It should be held in place using at least 1 screw and should be locked into place with the PCI-E latch.

 

[RChappo]

Hi jaguarmatt,

I just checked and the card it secure with 1 screw and it's locked in the PCI-E slot with the latch.
Incidentally I'm back on the integrated graphic card and the GPU is now running at 93 degrees - but it has not shut down at the moment!

 

[jaguarmatt]

93 degrees C or F?

 

[Twoboxer]

As mentioned earlier, this is almost always a power or heat problem. And if your psu can deliver 300W, its probably enough to drive that 210. I know you don't think it's heat related, but let's examine temps more carefully.

Please download CPUID's Hardware Monitor (for cpu/gpu temps), Furmark (stress vid card), and Prime95 (stress cpu).

Run Furmark and watch gpu temps. Report max temp or temp at failure.
Run Prime95 (check option to detect rounding errors) and watch core temps. Same report.
If no failure, run both stresses simultaneously and report.

If the temps reported by Hardware Monitor don't look bad, you likely have a psu problem.

 

[RChappo]

Again. Thank you for your replies.

Twoboxer:

I d/l and ran the tools you suggested and here are my findings.

(All temps in CELCIUS / GeForce 210 card enabled)

Furmark : Ran the GPU stress and didn't get a failure. The max temp of the GPU was 57 during the test and the actual temp of it at the end of the test was 51.

Prime 95 Torture Test : Ran this and got a failure between 3 and 5 minutes into the test. The Core temps at the failure were 50 degrees. The core temp was 36 degrees just prior to the start.

I had my PC reverted back to it's integrated graphics card all night and although the system wasn't stressed the GPU temp this morning was 84 degrees - however the system stayed on all night - no shutdown even when I browsed the web with it this morning.

Again , thanks for the advice...will be interested to hear any more suggestions

Richard

 

[Twoboxer]

Failure at 50C means you have at least one problem that should not be heat related.

What was the failure in Prime95? If it was a rounding error, the likely cause is a memory issue (or an unstable overclock, which is not applicable in this case).

If you have more than one stick of memory, repeat the Prime95 test with one stick and then again with the other. If you have 4 sticks, repeat the process as necessary with different combos to identify (1) one or more bad sticks of memory, or (2) a mobo or BIOS settings that cannot support multiple sticks of memory installed. The latter case may require adjustment of BIOS settings (higher voltage, relaxed timings) . . . provided the Gateway BIOS will permit it.

[Before doing the above, it would be useful to download CPUID's CPU-Z to identify what voltage and memory timings are actually being used. With your memory's model number we can compare this to the settings the manufacturer requires. This is unlikely to be the issue, since your system has worked for quite a while. OTOH, some form of system crash may have signalled your BIOS to change its settings from as-shipped, and a simple CMOS clearing may be all that's needed.]

 

[RChappo]

Hi again Twoboxer,

I'm not sure what the failure in Prime95 was - the PC shut off during the test - is there a log somewhere that Prime95 makes of the test data?

As for the memory - I'm really not knowledgable about this stuff. I did d/l and run CPU-Z though and this is what I see regarding CPU and Memory :

Maybe this will info will help in deciding what I should do next?

Richard

 

[Twoboxer]

CPU-Z reports 6GB of memory. So your mobo should have four memory slots, two of them in one color, and two in another color. Each slot of one color should be holding a 2GB stick, and each of the other color should be holding a 1GB stick.

While all 4 sticks are the same physical size, two of them should appear to have less or smaller chips on them. Remove those two sticks, NOTING WHERE EACH CAME FROM, by pressing down the tabs at either side of the socket.

Boot up, and check how much memory you now have - it should be 4GB. If it says 2GB you have removed the wrong two chips, but no matter . . .

Now run Prime95 again and see if it fails.

Then put back the two sticks and remove the other two sticks, boot up, check how much memory you now have, should be 2GB. But no matter . . .

Run Prime95 again.

Please report what has happened, and report what the Manufacturer and Model Number motherboard you have (CPU-Z, Mainboard tab).

 

[RChappo]

Ok..sounds good. I can see the 4 sticks of memory and will run that test tomorrow evening when I get home from work. BTW will I need the new graphics card installed to run the test or will the integrated graphics still give relavant information (I have reverted back to integrated graphics for the time being).

Sorry to ask dumb questions!

 

[Twoboxer]

Don't worry about the questions. The only concern is when people do not take advice, or for some reason don't report the info we need to help.

For this test, we're only concerned with memory. Integrated graphics would be best. If you get no failure while using IGA, repeat using the 210.

 

[RChappo]

Hello again Twoboxer!

I finally had time to run the stress test this evening, here's what I saw:

(testing done with integrated graphics enabled / new graphics card removed)

With 4 GB RAM onboard : PC shut down 10 minutes into the test.

Core Temps @ start = 39 deg C
Core Temps at end = 50 / 51 deg C

GPU temp @ start = 77 deg C
GPU temp @ end = 96 deg C (is this what caused the shut down??)


With 2GB RAM onboard :

I attempted to run this but the screen froze almost as soon as I started. PC slowed right down and temp sensor info all froze / windows could not be moved etc...only the mouse cursor was moving. I could hear the processor working etc but I decided to shut down the PC manually after about a minute.

(Is this one worth repeating?)

Here is my mainboard information:

Once again I thank you for your help so far!

Richard

 

[Twoboxer]

Your "GPU" temps are way out of line. However, I've never personally seen Hardware Monitor report any gpu temp for onboard graphics. (I just checked my wife's i3 530 to verify) Could you capture a screen shot sometime between boot up and failure? I need to see what you're looking at.

Starting the test at 77C would mean it was hot before you exercised the PC at all. 96C is above danger level for a graphics card GPU. Finally, Prime95 does not stress graphics at all lol. So why is the temp high? If we were to believe the temps, you have a motherboard issue.

 

[RChappo]

Hi Twoboxer,

I suspected as much with regard to the gpu temps. When I reloaded the 6GB of RAM and booted up again a few minutes after the stress test I was able to watch some HD video and do other moderate graphical tasks using the integrated graphics with no problem....and I didn't detect any burning smell! LOL

Can you clarify for me what you would like to see a screenshot of please...what do you need to see?

Thanks
Richard

 

[Twoboxer]

I would like to see whatever it was that showed you "the gpu temp was 77C at the start, and 96C at the end".

I've not seen temps reported for onboard graphics. The second problem is the stress test we ran should not have heated up ANY graphics device.

 

[RChappo]

Ok,

Here is a screenshot of the HWMonitor window that I just opened. The PC has been on overnight with no failure...

 

[Twoboxer]

Sorry for doubting you - even an old dog can learn new tricks I guess. Yup, that's your onboard gpu and, yes, its running hot. Duh!

There's really not much you can do about it, except perhaps repair/replace a missing heat sink on the "northbridge", if there is or were one, and if that's the problem (rather than some electrical issue).

An example of a "northbridge" is shown as #4 in this photo:

http://www.pantherproducts.co.uk/Articles/Motherboard/Understanding your Motherboard.shtml

There is usually some kind of heat sink on top of the chip. If it has gone missing, you might see residue of thermal paste/adhesive on the chip top. If there is one, and it is loose, you may be able to re-attach it.

I've reviewed the tests you've reported, and it seems all the failures could be attributed to this chip heating up. Unless I've missed something lol. But the sudden occurrence when you installed the new graphics card still bothers me. Could you have dislodged the northbridge heatsink when installing that card?

 

[RChappo]

Hi Twoboxer,

Thanks for sticking with this!

I guess my onboard graphics have always run hot, it was the reason why I wanted to install a new graphics card in the first place....and looking online I noticed that the graphics capability of this PC model had been a problem for other people too. Even though I don't really use my PC for heavy gaming I had noticed that I had had the odd shutdown when using Google Earth or playing Football Manager for extended periods. Usually you could tell when the shutdown was imminent because the graphics would lag and stutter beforehand (plus I used Speedfan to monitor the gpu temp and saw it get hot). I figured that the graphics card I purchased would would be sufficient for my needs. I don't think that I disturbed anything during the installation of my new card but I can take a look...the new card didn't seem difficult to install and neither was it a tight fit in the case. I can't help thinking that my problem is due to me missing something in the setup. I didn't do anything in the BIOS settings when I installed the new card...just plugged it in and then went into device manager and disabled the integrated card. Initially when I had trouble with the new card and switched back to my integrated graphics I left the new card in but had it set to "disabled" in device manager...but I noticed that it's cooling fan was still running even though it was disabled. Now I have pulled the new card from the PC altogether and I'm having less shutdowns with the integrated graphics even though it's still running hot. Did I miss something obvious....maybe a "dummy's guide to installation of a new graphics card" is in order?

 

[Twoboxer]

Honestly, not that I can think of.

One semi-plausable explanation for the problem may be that when your vid card is installed, the extra demand on your psu lowers the voltage it delivers. Which in turn may raise the current the chip uses. Which may be responsible for generating extra heat when both are installed. But mobo voltage regulation might correct this problem before it reached the northbridge. Dunno, it's beyond my knowledge.

Anyhow, I think the best advice I can give you is that motherboard is failing. Depending on what you use the PC for, or have on it, it might be wisest to make an orderly transfer to a new machine while you can do so. If you *can* get a motherboard replacement from Gateway, it may cost 1/3 - 1/2 what you paid for the system itself. Worth checking though.

 

[RChappo]

Hey twoboxer,

Well thanks again for your help and suggestions. I guess the best thing I can do is get someone to physically take a look at the PC and check it out before I invest in a new mobo and/or new PSU or whatever is required.

Cheers

Richard

 

[Twoboxer]

Good luck - sorry we didn't do better.