4 gig video card, MSI Nvidia Gforce GT730. low end stuff really but I wanted to add more ram into the GPU area of the machine last year. Overall specs, 16gigs ram, quad i5 Sandy Bridge configuration. I know, should have gone for the Ivy Bridge and I still may. WinX because what the hell else would I use?
A few weeks ago the video started shutting down while the computer kept running. It would be fine again after restarting. Could figure it out at first. It only did it on the one guilty pleasure I indulge in. Asphalt8: Airborne. Yep. A car racing game which includes online racing against other players, most of which are in the Eastern World.
Now, this started as soon as an update was installed by the game maker, Gameloft. Naturally I thought they did something that was taxing the system to a greater degree. I gave up playing the game but the problem still bugged me in my head. A few less weeks ago I started running some of the software that came with the GPU and did a stress test. It's going through the test and what I noticed was when it reached 100ºC, it would shut itself down. That's why I couldn't play.
Keep in mind, this is my main machine for photo editing so I was concerned the problem would get worse.
I took out the GPU and ran bare naked on the MB video. Yes I noticed a difference in speed of the video and the quality looked like it was off just a hair. I run video to a 42" monitor. A TV actually via VGA. Tried HDMI but the picture quality from the computer just wasn't there. VGA is razor sharp and the colors are dead nuts accurate which is important for someone who does what I do.
Upon removing the fan and the heatsink from said GPU, I noticed it was pretty much set up like a CPU on a MB, thermal grease and all. The first thing that stuck way out was how the thermal grease looked like it had separated. Like oil and water separates. Maybe it broke down. Maybe the heatsink wasn't making good contact. Maybe it broke down because the heatsink wasn't making good contact.
I had already went to the locally owned computer store, which we are lucky to have and grabbed a couple of small fans intended for GPU usage that were also rated 2000 RPM higher than the stock factory fan. These were on clearance for 2 and 3 dollars and came with thermal paste, copper heatsinks and a cord to directly connect the fan to the main power supply. After much thought and consideration, I ended up cobbling both fans directly on to the heatsink forcing air down upon it. While I was at it, I attached a 4" fan to the other side of the board and wired it in to the case power supply feeding 12 volts.
Now keep in mind that the connectors are the old school, bulky white 4 prongers. I had soldered all the wires and connectors so that all I would have to do is mount the card and plug the fans in. I'm looking inside that case, (yes I built this machine about 14 months ago) when I notice that I only have the newer flat connectors and no converters (if they even make those). Ha, well, the soldering iron was still hot. I had to cut into the extra connector from the optical drive and tap into the yellow and black wires there.
Am I done yet? So I solder, tape, route, clear, route again before I finally get to start the beast up. Everything spins! No smoke! No noise! No complaints from the bios or OS! The start screens lights up and it goes to the desktop. I run the stress test again and and this time it starts at 33ºC and maxes out at 54ºC. Room temp was about 24.4ºC.
So now I'm off to the races! I start the game up and it runs, and runs, and runs, and runs ... very awesome! I check the temp after 45 minutes and it peaked at about 55ºC or 104ºF. So now I wondering if the original 3000 RPM fan, running off the power from the GPU was enough to begin with or was it thermal paste breakdown or a bad contact with the thermal paste or is something else breaking down on the board that is causing heat. Did the manufacturer realize that they made it too easy to fix and was that their intention and what will happen in the future?