Friday 31 August 2012

Hardware Related PC Crashes

Computer crashes happen and leave most users wondering what to do. It's very common to blame an application, whether you go big and blame Windows or whichever program was running in the foreground at the time of the crash - or maybe you think you might have a virus.

The fact is that yes, software can cause issues like the dreaded Blue Screen of Death, but so too can hardware. Hardware errors are not something commonly thought about unless a user is over clocking, like I mentioned in a previous blog post or two. Obviously when you over clock it's important to make sure your computer is stable, but often when a computer is at stock settings people simply don't think about it.

Additionally, if a computer is running well for a year or two, and then all of a sudden crashes begin to occur, it's easy to wonder about that last program you installed. Sometimes people even go so far as to completely format the hard drive and reinstall Windows. This might help performance and certain PC issues, but it might not address the bigger problem - especially if it continues to happen.

The truth is that PC hardware can have defects. Sometimes they are defective from the factory, other times they simply get worn out by heat and power (electromigration for example). This is an interesting article on the subject, written by Wired.com: http://www.wired.com/wiredenterprise/2012/08/chip_errors/

Unfortunately these things can happen in any part of the system, be it the CPU, DRAM, GPU, or even the motherboard. Not everything can be checked, but there are tools to help rule out some issues. If your computer is crashing, here are some things to check.

First and foremost, verify your RAM. Memtest86+ is probably the best program for doing this. To run Memtest, you first need to put it on either a DVD (ISO version) or on a USB stick (USB installer version). Then reboot your PC with the DVD or USB inserted. You may need to go into your BIOS (typically hit Delete key at startup screen) and make sure that DVD or USB is selected as the boot drive before your hard drive.

Memtest will then auto start and begin running. It takes quite a while to run through a single test so make sure you have time to let the PC run on it's own.

Hard drives can also have errors on them. Sometimes it is physical errors, sometimes it is corrupt data errors. The first thing to do to check a HDD for issues is to run a Check Disk scan in Windows. You can see how to do this in Windows 7 here: http://www.sevenforums.com/tutorials/433-disk-check.html

The CPU is the next possible culprit. There's two pieces of software I personally like to use to check CPUs, but there's many others available too. These are the most common tools used by over clockers to check stability.

First up is Intel Burn Test. Running it at the normal setting for about 20-30 runs should verify that the CPU is pretty ok. It's also possible to check communication with RAM by setting it to a High stress setting. Using more RAM will cause the test to run a lot longer, so for 4G or more RAM I'd suggest no more than 10 runs. If an error shows up here but not in Memtest or on a low stress run, it's probably a RAM issue but not a severe one. Adding slighly more RAM voltage (say, from 1.5V to 1.55V) can likely fix it.

If the issues are only occuring during either HD video playback or while gaming (or other graphic intensive situations) then the culprit might be your graphics card. It could also be a motherboard issue, so a simply solution to try is to switch the GPU into a different PCIe slot. However, to test the GPU, run a program like Furmark. There's 2 settings to try out, first off is max resolution, full screen, with no Anti Aliasing enabled. Allow it to run for 3 to 5 minutes. The second setting to try is similar to before, except with 4xMSAA enabled. This will stress the GPU memory more. Finally, if both pass but you still think there's a GPU issue, you can try running various GPU benchmark programs like 3DMark 11, 3DMark Vantage, or Unigene Heaven.

In the case of any hardware issues at default settings, you are most likely eligible for a product exchange either from the place you bought the product, or else directly from the manufacturer. You will have to contact their support team, describe the situation, and hope for the best. Consult your product warranty info/user manual that came in the box.