Dying GPU or something else?

[bigbob1122]

I've got a 3 year old ATI HD5670 in a Dell E510. Obviously this isn't the latest and greatest hardware but I use this as an extra computer. I have a few older games installed on it like Left 4 Dead, Portal, and GTA IV. They have all ran well over the past few years but lately I've been having issues with this card. I'm assuming it's dying on me, but I'm not entirely sure.

When playing a game weird artifacts will start appearing on screen or the game will just lock up completely requiring a reboot. When just browsing the internet sometimes I'll get strange color patterns on the screen or the computer will lock up. The screen will then go black for a few seconds and then come back on and I can continue browsing as if nothing happened.

I tried switching back to my old card, a crappy ATI x600, and it worked fine. I couldn't use it to play games since it's so weak but I didn't experience any freezing at all. I then switch back to the HD5670 and used the ATI driver cleaner tool to do a fresh install of the drivers. I've tried a few different drivers for this card but they all have the same result. Games will run fine for a few minutes but start freezing after playing for a little while. I'll also get weird artifacts and tearing under normal web browsing. The card's idle temperature is around 40 degrees.

Any ideas? I'm assuming this video card is just about gone but I want to be sure the problem isn't somewhere else.

I've attached a few pics of the issue. Strangely, a Youtube video I was watching at the time continued to play just fine while the rest of the window was entirely unreadable.

 

[FireManDude]

artifacts can mean that a card is over heating ! before rulling that the card is dying do this:
open the card up and reapply thermal paste!,it might have dried out or went ineffective
Also thoroughly clean the heat sink and fans!

 

[Shaun o]

To confirm if the card is overheating you can use software to monitor the temps.
While in a game, it should create a log file of the temp of the card to a text file so you can look at it.

If a few years old and the system has not been dusted out check for dust build up against the front end of the Gpu heat sink if bad air flow could be the cause.

Replacement of the thermal paste is also a good idea as it becomes less effective as it ages due to heat ect like the other poster said.

You should always inspect the card at least every six to eight months and give it a clean.
Remember a pc tower is like a vacuum cleaner sucking air in one end and dumping the hot out.
So you get a heck of a lot of dust in there over a 12 month or more period.

The only other cause of instability would be a Psu of an age of 5 years or more.
Again as time goes by they become less efficient and present problems of a system freezing or locking up due to improper power ditribution with age.

Do the first step give the card and system a good dust out.

If it still locks or freezes then you should be looking at testing the system with another power supply.

It does look like a over heat of the gpu or memory of the card.
If you get a message of the game stopping and the driver failing to respond, it is the card and the driver terminating 3D operation to protect the gpu from burning out.
Saying in terms it`s getting too hot for normal operation.

If both cards exhibit the same symptom then a Psu.
Also check the cause is not coming from the cpu overheating also.

 

[bigbob1122]

Interesting that you mentioned the CPU. Under normal web browsing when the computer locks up and the screen goes black I can hear my CPU fan pick up speed for a few seconds. Once the screen comes back on it slows back down to normal idle. So it sounds like CPU activity spikes during that time. If the CPU is overheating could it cause the problems I'm experiencing?

How would I go about checking the CPU temps? A few months ago I tried "Core Temp" but it wouldn't give me any sort of temperature reading, only usage percent.

I clean the dust out of the computer once a month so I know that dust buildup isn't the cause.

My CPU is a 3.2Ghz Pentium D.

 

[bigbob1122]

Did a little more observing and found that the graphical errors seem to increase as CPU usage increase. I can't find a tool to give me the temps from my CPU and I'm not sure if it supports that to begin with. I did open up Task Manager and kept an eye on the CPU usage. When the computer froze and the screen went blank for a few seconds the CPU usage was 100%. I started noticing artifacts at around 70% CPU usage. Under normal browsing the CPU idles at around 10% usage and I notice no glitches or artifacts at all.

I'm going to go ahead and see about replacing the thermal paste on the CPU and maybe add another fan since I have enough room for one.

 

[Shaun o]

The processor you mention tends to run hot, if using an intel stock cooler.
If you say the fan speed kicks up, then you get lag spikes in a game.
It is a sign that the cpu is clocking its self down to cope with too much heat like said.

So you need to check the seating of the cooler and its tim application.
Failing that you may have to buy an after market cooler to keep the temps in check.

Try this for reading the temps.
http://www.techpowerup.com/realtemp/

I suspect the cpu is down clocking it can go from 3.2ghz to about 800 due to thermal throttle management of the cpu.
The settings for the thermal throttle trigger should be found in the bios of the mobo you have.