The invention provides a method for positioning a fault memory. Each memory corresponds to a
light emitting diode (LED) lamp; in
system startup process, a basic input / output
system (
BIOS) detects memories; and if a memory fault is discovered, an
intelligent platform management interface (IPMI) command is sent to a
baseboard management controller (BMC), the BMC acquires the position of the fault memory according to the IPMI command, and the
LED lamp corresponding to the fault memory is turned on, wherein the memory corresponding to the lighting position of the
LED lamp is the fault memory. The method comprises the following steps of: (1) starting a
system, and detecting the memories by the
BIOS; (2) when the
BIOS detects the memory fault, organizing the IPMI command according to the position information of the fault memory, and sending the IPMI command to the BMC; and (3) receiving the IPMI command sent by the BIOS by the BMC, resolving the position information of the acquired fault memory, and turning on the
LED lamp corresponding to the fault memory.