AMD EPYC ‘Rome’ server processors may stop working after 1044 days of uptime

A bug in the AMD Epyx 7002 server chips, code-named “Rome,” can cause a core on the processor to crash after about 1044 days of uptime (just under three years), according to a revision guide (PDF) from the manufacturer who discovered Tom’s Hardware. If this occurs, the server must be reset before the chip works again. AMD has no plans to resolve this issue. The problem occurs because the core in question cannot wake up from the CC6 sleep state…