Cold RAM Crashing Computer?

RAM
Most of us have heard that heat can kill a computer. Just browse any online computer store and you’ll find countless computer cooling products. And while keeping components cool makes perfect sense, is there such a thing as “too cool?” Is is possible that my computer is similar to my car and needs a little time to “warm up?” Normally I’d said no to those questions but my current computer problem is starting to make me think otherwise.

Original problem: computer shuts down after a few minutes of usage

About a week ago I turned on my computer. Windows XP completely loaded, and the computer sat there idle for a few minutes ready for use. I was about to check my email when “click”, the computer shut off. I flipped the power switch off and then back on and the computer rebooted. With Windows XP completely loaded I started to check my email and sure enough a few minutes later it shut off again. Confused by the problem, and little time to spare, I gave up and turned to my laptop and carried on with my day.

The next day I went through the same process and encountered the same auto shut off problem. At this point I concluded there was some hardware problem but I didn’t have time to identify the source of the problem. My desktop computer would have to wait till the weekend for further attention.

Computer running once again but new clues to the source of the problem

Over the weekend it took me several attempts to finally get my computer to stay on long enough to update my data backup. Before I started to probe into the source of the problem, I wanted the latest backup of my important data so that in the event the machine completely melted down, it wouldn’t bother me too much.

I actually thought the machine was overheating due to a failed cooling component. I opened the case and checked the CPU fan, case exhaust fan, video card fan, and power supply fan. Everything was working as it should. Next I booted the machine and launched the BIOS setup. I checked all the settings and everything appeared as it should. I let the computer sit idle in the BIOS setup for at least 20 minutes. I thought that if it shut down now, the problem was definitely hardware related. Nothing. The machine happily sat there well beyond what it had when I booted into Windows. Maybe it needed a little more activity to get it to shut down.

I booted from an Unbuntu live CD (popular Linux distribution) and started to use the machine. I browsed the web and tried various applications long enough to where it should have shut itself down. Nothing. The machine happily hummed along as it always had. Now I was starting to think this was a hard drive problem or even worse some issue with Windows XP (I dreaded the thought of having to reinstall XP).

I booted back into Windows and combed through the Event Log to see if anything gave me a clue to the source of the problem. Nothing. I thought maybe it was a fluke and I wouldn’t see the problem again. I doubted that conclusion but knew that the only way to really solve the problem was to be able to recreate the problem at will. Unfortunately, once the computer got running after a few restarts, the problem didn’t return.

Happy that I had at least updated my backup I turned the machine off and decided to continue my research tomorrow.

Memtest86 provides some insight into the problem

MemTest86

The next day I started my computer as I always had and sure enough after completely starting Windows XP and sitting idle for a few minutes, the computer shut down. I spent some time researching my troubleshooting options and decided the next best step was to run Memtest86 and validate that my 2 GB of RAM was not the source of the problem.

Note: Before disaster strikes, take a moment and build some system recovery tools from a machine that has CD burning software capable of creating a CD from an ISO file. My desktop has such software but my laptop does not. I wasted more time than I care to mention on searching for decent and free CD burning software. A frustrating experience.

I ran Memtest86 version 2.11 from a floppy disk. The test got as far as test #3 (moving in versions, 8 bit pattern) before it started to report a failing address at 0006FE1CC74 and gave a RAM size marker of 1790.7 MB. I’m not certain exactly what that means but I came to the conclusion my problem was in my second RAM module. I was overjoyed because replacing that module would be easy and cheap. Nevertheless, I thought I would double check by rebooting my machine and run the test again.

Some version of Murphy’s Law kicked in when I reran Memtest86 because everything checked out. Memtest86 found absolutely no problems with my RAM. One nice feature of Memtest86 is that it will repeat its series of RAM tests over and over until it’s told to stop so I let Memtest86 do its thing. I thought maybe a second run of tests would reveal the problem. Nope. A third pass and everything remained perfect, and the fourth run was just as clean as the first. What happened to my problem?

Completely frustrated I killed the test, pulled out the floppy and let the machine completely reboot. Once back in Windows everything was running as if nothing was ever wrong.

Now what? Next steps for problem shooting this problem

A few computers ago (first generation Pentium machine to be exact) I had a machine that suddenly starting having booting problems. I’d press the power button and the computer would indicate it had power but it would just sit there. If I let it sit for about 5 to 10 minutes and then switch it off and on, it would finally boot the rest of the way. It was as if it needed to “warm up” before it would completely start. I hated the problem but eventually accepted it as a fluke and built in the “warm up” routine into my start up process.

My current problem is starting to feel like that old Pentium machine — given enough time to “warm up” the machine will eventually work as it should. This time, however, I’m not about to “accept” this quirk. I want answers, damn it! My next step to solving this problem will be to try and find a way to recreate it at will. I plan to start the machine every morning with the Memtest86 floppy and see if I can get the memory test to fail. If it does, I’ll make note of the failure point. Then I’ll reboot the machine and rerun the tests. I’ll make note of any failed tests and also see how many times I have to repeat the process before I get a clean Memtest86. Once I can recreate the problem, I’ll figure out my next steps.

I’ll try to update my troubleshooting progress in a new post each week until this problem is finally solved, or I’ve thrown the machine out the window in complete frustration. Also, as always, any suggestions or insights are always welcome.

Updated: August 31, 2014 — 11:34 pm
Peter Herrera's Blog © 2014