Aims to be a useful grey-hair prevention tool for Linux users (especially sysadmins).
Helps users to pinpoint problem hardware (even if they hadn't yet realised that it was faulty).
Makes use of existing (but largely unused) hardware-provided error-detection facilities.
Tries to do useful stuff with the information (practically anything is better than the status-quo).
Currently handles two main areas - System RAM (Main Memory) Errors, and PCI parity Errors.
Only reports errors at the moment (there are plans to do more intelligent fancy stuff).
Is currently most useful on recent/mainstream x86 chipsets, when used on systems with ECC system RAM.