Hamilton Chris-cham <chris.hamilton@xxxxxxxxxxxxx> writes: > I haven't read a lot about bad RAM, but I had the impression that if > you had bad RAM, your failures wouldn't be so predictable. In my > case, the failure happens predictably at the same spot every time > (assuming I don't change the configuration). I agree that that seems less likely to be a RAM problem, although it is certainly still possible. > How can I prove whether this is a memory problem? Run a memory tester, such as http://www.memtest86.com/ I don't know whether that will work on an x86-64 platform, but it may. Or AMD may provide one with the system. But first see if you can run the failing command manually, and have it fail in the same way. If you can, add -xv options to ksh to see precisely which command is failing. That may help. Ian
