Since yesterday, my computer has rebooted at random whilst playing Star Trek Online. Sometimes in the middle of battle, opening an inventory screen, changing instances. Sometimes hours into game play, or minutes into logging in. What is odd that the previous 5 days of beta were relatively flawless crash-wise.
However, what I have noticed is that the graphics card seems unusually hot even minutes into playing, which seems odd considering STO isn't exactly a graphically stressing game compared to Crysis or Modern Warfare 2, both of which run without a hitch.
My computer has been running well, drivers are stable, scanned regularly for viruses and malware, has a good-quality power supply (FSP-manufactured Zalman-branded 500W) to power my system, which has an nVidia 8800GT.
If your PC is rebooting while gaming, it is either 1.) overheating and the video card or CPU shuts down, or 2.) you have defective components.
Check system/CPU temps with software such as SpeedFan, and check video card temps with your ATI Catalyst software, or for NVIDIA cards, with your NVIDIA control panel. You can also use third party software to check video card temps.
My keyboard has an LCD panel that shows CPU usage on all cores. Saw it at 99% last time the PC locked up.
Just tested STO again - GPU temp was 63C CPU was 53C. Then computer rebooted.
I need to test some of my other games and see if I have this problem - I want to deal with hardware problems at the moment. Money is tight.
Those are perfectly normal operating temperatures. In fact, those are actually very good operating temperatures (as long as those are CPU core temps ). Temperature is definitely not your issue, so some other stability problem is.
What I would do is download Prime95 and Furmark (you can just google them). Run Furmark's stability test for about 5 minutes, and see if the computer crashes or if you see artifacts ("snow") on the screen. If that passes alright, then run Prime95 and see if it gives you an error.
Speedfan will also be helpful here because you can monitor voltages. If I had to take a guess, I would say your power supply is undervolting and causing your components to work improperly.
Prime95 and Furmark - I'm kinda not keen to stress my system to the extent that I may damage it
You won't damage your PC by stress or torture tests. If the CPU or GPU are overheating, they both have internal thermal protection routines to shut down if they get to hot (resulting in a reset or power down of your PC).
Prime95 is for stress/torture testing your PC. It just runs a ton of prime number calculations on your CPU and then compares it to known good values. If your PC was overclocked too much to the point of being stable or running too hot and therefore not producing correct results, Prime95 would show errors with the results. Overclockers use it to test how stable their overclocks are.
Futuremark 3DMark (ie: 3DMark Vantage) benchmarking software does a stress test on your GPU(s)/graphic sub-systems. It's doesn't peak your GPU all the time liek Prime95 does to you CPU, but it gives you a good idea of what a long gaming session might do to stability and heat for your video card or built-in GPU.
Cata also mentioned FurMark, though I don't use it. Looks like a 3D torture test (looks good). I use ATI Tool for torture testing GPUs.
SpeedFan monitors your motherboard's and other devices' (hard drive, CPU, GPU, etc.) sensors to report temperatures and, as the name implies, fan speeds (if they're connected with a sensor lead). Overclockers use this to check temps against known values of ranges for specific hardware. If an overclock is running the system a little too hot, a modder might change the cooling system or downclock a bit to run it cooler.
None of these "damage" your PC. It's just software.
Well, while it is true that any software can damage a PC that is cooled poorly enough, generally speaking software cannot, in itself, damage hardware. The fact that PCs have auto-shutdown routines in the event of overheating means that even if a piece of software could potentially damage a computer, the computer is designed to prevent that. Of course, the OP already posted temps, and they clearly weren't the problem. The cooling in the system appears to be working quite well.
As for Furmark, it stresses the GPU(s) to about 98-99%, which means it works as well as Prime95 does for the CPU. It renders a lot of fur the way Ati tool does to generate heat/load, only it renders an even more obscene image (in terms of required processing power, that is ), and works great to test hardware. No program will ever pass Furmark for GPU load. In fact, that program gets my GPUs up to almost 90C, while Crysis/3dmark/STO will only ever get them up to the low 60s (65 max), and it does it for both of them, too. I actually just used it to overclock my 5770s this morning, and once I set the memory too high with Ati Overdrive (it only managed to go from 1200 to 1350 before crashing), the resulting system lockup was instantaneous with Furmark, and so it was a quick test to show that I had to back down the speeds a little bit. The same was true for the core clock, and it took only moments to produce artifacting once I hit the limit there, and so this program shows errors a lot quicker than something like 3dmark. Anything that can survive five minutes of a Furmark stability test or benchmark is completely stable for gaming, trust me
Still, it's just software, so it can't damage a computer unless the computer already has absolutely crippling hardware problems.