It was bound to happen within the first 2 years: I thought my video card was dying. Well, more correctly, I though the fan on it was dying. The card, a PNY produced Nvidia GeForce 9800GT, was overheating, reaching temperatures of just over 100 degrees Celsius, in effect cooking itself to death. Even with the fan speed forced to maximum, the unit is idling at about 75-80 degrees, which is already well in excess of safe ranges. With the fan speed set to minimum, the fan actually shuts off completely; not cool considering that is at a speed setting of 35%, not off.
Every once in a while, within minutes of booting, the system would just shut down, blacking out on me for what seemed to be no reason. More often though, the syste would boot fine, and run smoothly for a while, but once a game was running, it would only last a few short minutes before it would black out again. Once I noticed the pattern occurring in games, as it had before, I quickly realized what the issue was: The video card might be overheating. I loaded speedfan to make sure, and sure enough, I was measuring themps of 80 degrees Celsius idle, and logging anywhere between 104-107 before the system would die. Insane, to say the least.
So, the first thought in my head was a simple one: Dust. While I had just cleaned the card a few months ago, I figured I did not get it as clean as I could have and dust might be clogged up in there once again. Sure enough, the heat sink and fan blades did have a nice coating of dust in there, blocking a portion of the airflow.. but logically, not enough to actually cause the card to behave in such a way.
I thought for a while longer, and noticed that, while the system was running, the system fan would sometimes not run at all, or would be oddly slow. I figured that the fan itself was clogged with a small dust ball inside, causing enough resistance to, in some cases, stop the fans movement completely.
I was left with a system that, basically, was unstable: if the fan was not going to run, then the system would only have a few minutes before the video card would shut down, and I would have to reboot. If the fan would spin, unless I ran open case, and left a fan blowing air to the card, the system would only have a few minutes of game time before the inevitable shutdown. If dust was the issue, my only option would be to clean the card, and hope that it would run cooler, and be stable, after the cleaning.
So. I went ahead and purchased some compressed air and some pipe cleaners, went home, and began to clean. I cleaned each blade on the fan, and poled and prodded my way through the heat sink, trying to loosen every bit of dust I could find. Gave it a few bursts from the compressed air can, and repeated the process a few times. Swabbed over the surfaces with a pipe cleaner soaked in rubbing alcohol, and after letting it dry, finally placed the video card back in the system.
I turned the system back on, and ran speedfan: The idle temperature reading went from around 80 to 73… which seems to be the average for this type of card. I would discover later, while talking to a friend, that the 9800GT chip is a rather hot running chip, designed for double-slot expansion cards. My card, however, is a rather thin, single slot unit. This, combined with a rather tiny fan on the unit, means that it simply WILL run hot; there is no way around it with default cooling.
I continued to test the card, running the Milltown level on Left 4 Dead 2. My logic was that this level would tax the card just hard enough to get a good reading, but not be anything too critical, risking actual damage. The card held steady, and I was able to finish the level without issue. Checking the logs, the card held at about 104 degrees… still amazingly hot, but stable regardless. My friend would later tell me that, while that is hot, it is nothing dangerous to the card, and that the blackouts I had experienced in the past were actually the video card itself shutting down to prevent damage: the system would not know what to do with the video card dropping out like that, and so, it would simply lock up, looping the last bit of sound data it was working on. Pretty clever, actually, and good to know that the card was protecting itself, and the system.
In the end, I decided the best option to do would be to keep the video card clean, dusting it about once a month… and from there, getting a proper case fan to lower the ambient temperature by a few degrees. Every bit helps, right?
Incientally, the night of all this, I went ahead and purchased a replacement card, a GeForce 430. Obviously, with the temperatures running fine on my 9800GT, I will be returning this card. However, the guy who helped me with this card… well, there is a story to that one. More coming soon.