Tech Support Forum banner

I believe my Video Card is dying

6332 Views 78 Replies 5 Participants Last post by  greenbrucelee

Starting about 2 weeks ago, I noticed that games were starting to chug. These are games that I used to be able to play completely maxed out, but now the framerates drop and from time to time, the games will totally freeze or skip. Even Counter-Strike:Source has experienced problems.

here's what I think you'll need to know;

Intel Core 2 Quad Q6600 @ 2.40 GHz
Abit IP35 Pro Mobo
Rosewell ATX12V 600w PSU
ATI Radeon HD 4870

I ran ATITools for 90 minutes and got 191 artifacts.

The GPU Temperature is at 68ºC after sitting idle for about a half hour.

Any ideas what's wrong?
(thanks in advance)
Not open for further replies.
1 - 20 of 79 Posts
Rosewell ATX12V 600w PSU)
I believe that is your problem, can you get into bios and post your temps and voltages

Temperatures (what I could find at least)
See less See more
The HD4870 should be about 45C at idle, only reaching 68C under stress. The BIOS readings show that your PSU's +12V output is too low. I would check the graphics fan and replace the Rosewill 650W with a better quality PSU.
According to GPU-Z, my fan is running between 30% and 35%, and GPU temp is fluctuating between 67 and 68ºC. It's been sitting idle for a few hours. The fan looks to be running fine, but there's not a whole lot of air coming out of the back vent. I just cleaned out my system yesterday, so it's not clogged with dust at all.

Also, any suggestions for a better PSU?
You can increase the graphics card fan speed in the Catalyst Control Center. Click 'Graphics' in the top left and select ATI Overdrive, then enable manual fan control, set the slider to about 60-80% and click Apply. Higher fan speeds are louder, so find a compromise between noise and temperature.

Corsair 650W - $70 (after $20 rebate) -

Corsair 750W - $100 (after $10 rebate) -
With my fans at 70%, my card is down to 48ºC. It's a tad loud now, but it'll be worth keeping my card's temperature down.

I'm going to aim for the 750w if I can. Are you sure it's my power supply though? I just don't want to buy a new power supply if that's not the root of the problem. Money's kind of tight for me.
See how you get on with the ATITools test and games with the fan speed increased. If everything's ok, you could drop the speed down to 60% to make it quieter. Mine is 40% and 50C with just Firefox, Notepad and a media player running. Fullscreen games at high settings take it up to about 60% and 65C.

The +12V should be very close to 12.0V in BIOS, and can fluctuate while playing fullscreen games, so 11.65V is not good for an idle computer. Check with a system monitor like Everest or SensorsView to see what it's like under stress while playing. The rest of your voltages look ok, it's just the +12V that is low. This can contribute to high GPU and CPU temperatures and can cause lag in games.

Have you got two case fans, one at the front pulling in cool air and one at the back blowing out warm air? If not, then installing two 120mm fans will help bring down the graphics temperature.

750W would be better if you manage it, as it provides more +12V amps and gives more headroom for future upgrades, but the 650W would still work. Just don't go for one of the really cheap brands. Corsairs are excellent value for money, very reliable, and have a high efficiency rating and a 5 year warranty.

EDIT: I've just noticed that your CPU is overclocked. Set this back to default and check the voltages again in BIOS and during gameplay.
See less See more
I tried running Counter Strike:Source which ran fine, but when I tried to play Left 4 Dead 2 the first time, both the audio and the video skipped. I'm going to run ATITools again later today for another 90 minutes to see what I get.

I ran SensorsView while playing Left 4 Dead 2 a second time. When I first started up SensorsView, it said my +12V was at 13.10V and maxed out at 13.40V while playing. The highest my GPU temperature got while I was playing, was 52ºC. Graphics on the second time were better, but still chugged at spots, especially during fast movements or when a lot was going on (which never happened before, no matter what game.)

I have several case fans actually. Two 80mm in the front (intake), one 80mm on the top (exhaust), and one 120mm on the back (exhaust). Just after getting done testing out games with SensorsView going, my case temperature was at 32ºC.

Also, immediately after I originally posted this, I restarted my computer and went into BIOS. In there still said that +12V was at 11.65V... I forgot to mention when I originally posted the voltages, the voltage level for +12V (8-Pin Connector) went between 11.58V and 11.65V. The voltage for +12V (24-Pin Connector) didn't move from 11.65V.
See less See more
Software system monitors are not as accurate as the BIOS readings, but even taking the margin of error into account, a jump up to 13.4V shows that there's definitely a problem. The safe range is 11.4-12.6V, but with a good quality PSU you'll see it will be closer to about 11.9-12.1V, even under stress.

What are your CPU and hard drive temperatures? These are also powered by the +12V rail, so if they are high and you're getting 13.40V, then the PSU would be the main suspect, especially as Rosewill is not a good quality make (70% efficiency and approx 26A/+12V). The Corsair 650W is over 80% efficient and provides 52A/+12V.

If you've been running your HD4870 on the Rosewill 600W for a while, the poor quality might have damaged the graphics card. You can test the card in another computer with a suitable PSU to confirm this, but I would still recommend upgrading to a Corsair.
My CPU cores range from 28ºC to 33ºC. The highest any of them got to was 35ºC. My HDD is at 34ºC and has stayed right at that level the whole time.

As of right now, +12V is sitting at 13.20V as far as SensorsView says.

I've been running my graphics card with the rosewill psu for over a year... I really hope it hasn't done damage because I wouldn't be able to afford to replace it. I do plan on upgrading to a Corsair as soon as I can.
Your CPU and hard drive temperatures are ok, so it looks like your problem is being caused by the low +12V amps from the Rosewill.

PCIE cards need a minimum of 28A, but 40A is recommended for mid-range cards like the HD4870.

Good luck and let us know how the upgrade goes. :smile:
never buy any power supply with the name rosewill or your heading for trouble, they are one of the wost brans of psu you can get.
I decided to run a ATITools test as soon as I started up my PC today and got zero artifacts.

The thing is... according to SensorsView, my +12V is running at 13.70V.

I'll hopefully be ordering the power supply sometime this week. In the meantime, is there any way to test and see if my graphics card is damaged besides putting it into a different computer? None of my friends have a computer that would be able to use my graphics card.
lower or higher than 10% on 12v is bad.
The only way to properly test a graphics card for damage is to use a suitable PSU. If you don't have access to another computer, you could take the card to a PC repair store and get them to check it out, or wait for your new PSU to arrive.
I got an installed my new power supply today. BIOS says that both +12V are running at 12.10V. SensorView says 12.90V.

I did an artifact test and found none, but my graphics are still sluggish...

There's definitely no improvement on the graphics card even though the voltages are now correct. Is there any way to diagnose what's wrong with the card?
the vram has probably been damaged due to your other psu. Low quality power supplies tend to damage other components when they start to fail. Rosewills are renowned for it just linke winpower, huntkey and HEC are.
Can you test the card in another computer?
Unfortunately none of my friends have computers that would be able to use my graphics card. I'm going to look for a computer repair store, or something of that nature, so I can bring my card in and ask them to run it. Not sure how well that'll go over or if I'll have to pay, but that'll probably be several days from now. I'm surprised there's no sort of diagnostic software for graphics card so you can find out what's wrong.
1 - 20 of 79 Posts
Not open for further replies.