That's what I get for buying such complicated shit

Last night I tweaked a couple of things in Vista to try and boost its performance a bit (not that it is lacking in performance, but because I am a hopeless tweaker). I changed the VM settings, disabled and changed a couple of services, and modified the BIOS settings a bit. It booted up just fine, and worked great until I fired up WoW to get ready for last night's raid -- and my video crashed hard.

I went through and meticulously backed out all of the changes I had made and fired up the game again, with the same result. Hm. Even though they should not have affected the video drivers at all, I halfway suspected it might have been caused by the game add-ons that I had updated just prior to running it. I considered stripping the game of add-ons to see if that would help, but I decided that I would first try updating the video drivers -- the ones on the nVidia site were a few months newer than the ones on my machine. Couldn't hurt, could it?

This time the entire system crashed when I tried to run the game, to the extent that for the first time ever since I went to Vista, I had to use the power button to reset the system. I booted back up again, and I noticed that I was getting occasional video glitches during the boot process. Rather than trying to run the game again, I ran PC Wizard to see what was going on with the internals of the machine.

Core voltages looked fine. CPU usage was nominal. CPU temperature was fine. GPU temperature was... ack! The GPU was clocking in at a balmy 117 degrees Celcius. I copped a glance into the top of the case and saw what one does not want to see when they are running a water-cooled system; there were large air pockets in the water hoses, and the fluid was definitely not flowing. I powered down the system immediately, because I suspected that the pockets were actually steam where the fluid inside the GPU jacket was boiling off from the heat.

I know that the cooling system was working when I got up yesterday because I make it a habit of glancing at the hoses for air pockets every other time I use the system, and everything looked fine in the morning. I ran a few tests, and as far as I can tell, the fan is still working but the pump is dead. I checked the manufacturer's site, and they have a three-year free replacement warranty. Time to dig around for my receipt. I hope I can get more than six months out of the replacement.

None of the reviews of this cooler mentioned anything about problematic pumps, so I am left wondering if it got damaged in transit. It was about -32 on the day it got delivered, and the unit was frozen solid when it arrived. Hmph. Next time, damn the noise, I'm getting a fan.

Video card update...

I pulled out the video card (swapped it for my emergency-backup X1300) and water cooler and drained the latter with the intention of sending it back to the factory for a working replacement. On a whim, I hooked the emptied cooler back up to one of the power leads to assure myself that it had not mysteriously fixed itself in the process of being emptied...

... and apparently it had mysteriously fixed itself in the process of being emptied. As soon as I powered the system up I could hear the pump humming as it sucked air. I could feel it vibrating to the touch as well - something that it had not been doing when it was in the case. I fetched the video card again and snapped the hoses together to confirm that it was working, and the pump obligingly started pushing air through the lines. I popped the cap and refilled the fluid reservoir, and after a lot of alarming gurgling and labouring, it went right back to work as if nothing had ever happened.

It is back in the case again, and the GPU is running at its normal idling temperature of 54C. I have no idea what happened, but I am going to keep a closer eye on it until I feel that I can trust it again. I am not too concerned about it overheating when I am not here; my power settings shut off the display after 20 minutes, and the card doesn't generate a lot of heat when it is idling. Still, this is the worst kind of component failure. I would rather have something break outright so that I can send it back for a replacement. If it cuts out again I will send it back regardless and advise them that, whether it works or not on arrival, it had quit working twice in my system.

Hopefully I don't need to do that.