Yes, I have a way that I smoke out bad hardware. I've just put the hardware-testing routines I use into http://www.augart.com/stress-hardware/. All it does is compile the Linux kernel from source 100 times. It compares the output each time to make sure it's identical. I've had hardware that was just barely flaky -- GCC would segfault once out of every three or four kernel builds. But that was enough to cause unpredicatable errors in my system. The documentation on stress-hardware is only in the comments, but the description of it here should be enough to get you going. I also use ECC ram and turn on the ECC checks in the BIOS. I have to admit that they've never gone off so I don't know what would actually happen if I were to have bad ECC RAM. The ram I wrote stress-hardware to smoke out was not ECC RAM. If you use stress-hardware, let me know how it works for you. --Steve Augart Matt Stamm wrote:
Any other suggestions on reliable memory testing for Suse Linux systems?