I recently built a computer and any game I would run would randomly crash at some point in time. On top of that I would get random BSoD's.
I figured it was my nvidia card conflicting with my ATI motherboard so I got an ATI card but same thing happened after that. I even reformatted windows thinking old nvidia drivers were causing conflicts.
Here are my specs.
2.80 gigahertz AMD Phenom II X6 1055T
256 kilobyte primary memory cache
512 kilobyte tertiary memory cache
Multi-core (6 total)
First I'd recommend to get into the BIOS and do a Load Optimized Defaults (or Load Defaults, depending on your BIOS). Once that is done, manually configure your RAM to the specs on the stickers on your RAM modules. After doing this, try playing one of your games, to check for stability. If you get BSOD, continue diagnosing with the following suggestsion.
Load Windows. Download and install the following free software:
1. FurMark - GPU tester
2. CPU-Z - Use this to check what your CPU and RAM are operating at
3. Prime95 - CPU/RAM tester
4. HW Monitor - Displays what your hardware temps and voltages are at
5. GPU-Z - like CPU-Z and HW Monitor, but for your GPU
6. SpeedFan - use this to get a second opinion on what CPU-Z shows.
7. MemTest86+ - A RAM tester
MT86+ is software that needs to be burned to cd (or made bootable on a flash drive) and is used independently of Windows. To use it, put the bootable media into your computer and reset the computer. The test will run automatically. Note that this test may take awhile; 8-10 hours, unless errors are found. Also note that it is best to test one RAM stick at a time.
Since BSOD are most commonly related to RAM, it is best to start your diagnosing with MT86+. If no errors are found, proceed.
Ideally, what you should do is pull you system apart and rebuild it outside of the case; this is called breadboarding. Breadboarding will rule out any case shorts as the reason for failure(s). Granted, most electrical shorts cause a "no boot/no display" issue, but still, it's good to be 100% sure.
Run both CPU-Z and HW Monitor and check your temps/voltages, make sure they're where they should/need to be. If not, shut down and verify you've installed the questionable components.
After verifying temps and voltages are correct, move on to GPU-Z. Again, check temps, make sure your GPU isn't getting too hot. If all is well, proceed. If card is running too hot, check to see if the fan is spinning and check the aux power cable to the card.
Now its time to run a stress test on the GPU. Run FurMark. Using this software will open a window in fullscreen mode and will cause your GPU temps to rise, significantly. Don't be alarmed if you see GPU temps in the mid 80's. If your system crashes while this is running, chances are the GPU is the cause. If you don't experience any difficulties, you can stop running this after about 20-30 minutes.
Next, use Prime95. This tests stability under 100% load on the CPU. Naturally, 100% load on the CPU means that you'll be testing RAM along with it, but keep in mind the focus is on the CPU. If any errors are detected, you will see a message instructing you to look at a log file created by Prime95.