I’ve seen my share of operating systems and thought I’d seen it all when it came to crashes until about a month ago. The system I was troubleshooting consisted of an ASUS dual processor motherboard with two AMD 1.4-GHz chips, 768 MB of PC 2100 DDR RAM, an IBM hard drive, and an old Viper V770 video card. It ran Windows XP with no problems for about a year. Installations, hardware changes, and even program removals were seamless with XP until the day of “the crash.” But luckily, a program called DocMemory from CST saved me from having to purchase a new motherboard. Here’s what happened.

Random weirdness in XP
The first sign of trouble was when the computer locked up during the screen saver. The user wasn’t too concerned, since he could reboot and log back in. Next, he opened Internet Explorer, only to find that the program window opened and immediately closed. When he opened it again, the system froze. He rebooted. This time, when he tried to log in, the computer rebooted itself. Once the reboot was complete, he noticed his user name was gone, so he logged in under another user’s name when the system rebooted itself again. At this point, he brought the computer to me. Once he described the problem to me, I began to consider the usual suspects: virus, hard drive failure, video card failure, or a corrupted system file.

Troubleshooting gone wrong
My first step was to try to repair the Windows XP installation. Unfortunately, during the text portion of XP’s installation process, immediately after Setup Is Inspecting Your Computer’s Hardware Configuration, I received an message stating An Unexpected Error (-1) Occurred At Line 1562 In D:\nt\private\ntos\boot\setup\arcdisp.c. Press Any Key To Continue.

I tried the process a few more times, thinking this was a fluke, but received the same error message. On the third try, the computer entered a loop where it would boot and then automatically reboot.

At this point, I was thinking the problem was hardware related. I tried a different video card—I got the same type errors and crashes. I then took the hard drive to another computer to check for hardware errors. The hard drive was fine. I checked for viruses and found none. I then formatted the drive and began a fresh installation, only to get the previous An Unexpected Error… message. After multiple iterations between IBM’s hard drive utility and Windows Setup, the installation miraculously succeeded—at least, until the system restarted to complete the process. Then it crashed again. The only possibilities left were motherboard or RAM.

DocMemory diagnoses the problem
A friend of mine suggested I download a program called DocMemory from CST. DocMemory is a 271-KB, “free for a limited time” program that will create a bootable floppy to test RAM. Version 1.45 works for Windows 95/98, and Version 2.0 works for Windows 2000/98 and later. Upgrades are free once you’ve registered for the download.

After I created a bootable floppy with the download (a simple 30-second operation), I booted my system to the floppy. DocMemory opened automatically, recognized my RAM, and gave me the option to run a quick test. Six minutes and 37 seconds later, it identified a problem with my extended memory (memory locations above 1 MB). To make sure the error wasn’t in the memory slot on the motherboard, I switched the RAM to different slots and reran the test with the same results. The program reported which tests failed, the memory address at failure, the expected result, and the actual result. I could then save the test results to my floppy to view later in Notepad. The help file on the Web site included a detailed description of how to interpret the test results. Once I removed the memory stick in question, my system ran flawlessly.

Upon further investigation, I found DocMemory had various options:

  • Burn-In Mode for testing until the user stops the test
  • Multiple test pattern parameters
  • Modifiable number of test loops (from 1 to 9,999)
  • Definable area of system memory to test

Limitations of the software included these issues:

  • It is unable to specify the exact location of DRAM failure.
  • It is unable to detect the exact memory cell that failed.
  • The first memory module must be good for the program to boot.

Overall, I found DocMemory to be a very user friendly, robust program for testing RAM. I had no problems with the download, usage, or understanding of the test results. In fact, if I had taken an extra thirty seconds in the beginning of all my problems to run DocMemory, I would have saved myself hours of troubleshooting.