At-risk memory location identification and management

Abstract

A volatile memory device is configured to self-document by identifying its own bad or at-risk excludable memory locations in a nonvolatile identification embedded in itself, without using additional board real estate. The identification of bad or at-risk memory is readable by firmware outside the device. The device includes volatile memory cells that have respective failure susceptibility values, some of which indicate bad or at-risk memory cells. The memory device also includes read logic and write logic, and may include refresh logic. The identification may be embedded in the device by blowing fuses in an adaptation of self-repair activity, or by writing identification data into a serial presence detect logic, for example. The configured memory device may efficiently, persistently, and reliably provide detailed memory test results regarding itself, thereby allowing customers to accept and safely use memory that would otherwise have been discarded to prevent software crashes.