US 12,456,537 B2
Extending dual inline memory module (DIMM) lifespans based on fault characterization and optimized corrective action
Fabrice Aidan, Ramat HaSharon (IL); and Evgeni Krimer, Haifa (IL)
Assigned to Google LLC, Mountain View, CA (US)
Filed by Google LLC, Mountain View, CA (US)
Filed on May 22, 2023, as Appl. No. 18/200,322.
Prior Publication US 2024/0395350 A1, Nov. 28, 2024
Int. Cl. G11C 29/52 (2006.01); G11C 29/00 (2006.01)
CPC G11C 29/52 (2013.01) [G11C 29/783 (2013.01)] 23 Claims
OG exemplary drawing
 
1. A method for extending a lifespan of a dual in-line memory module (DIMM) comprising:
detecting an error associated with a first data read from a memory address of a dynamic random access memory (DRAM) device as a result of a first read transaction, wherein the DRAM device comprises independently controlled data banks, wherein each data bank comprises a subarray of data, and wherein the subarray comprises rows and columns of data stored in the DRAM device;
determining whether the detected error is a correctable error;
based on determining the detected error is correctable, determining a temporal characteristic of the detected error;
generating, based on the temporal characteristic indicating that the detected error is an intermittent error or a permanent error, a fault profile that corresponds to the detected error, wherein the fault profile comprises a spatial characteristic of the detected error,
wherein the spatial characteristic indicates memory locations within the DIMM that are affected by the detected error, and indicates a number of data pins (DQs) within the DRAM device that are affected by the detected error, wherein based on determining the detected error affects a single data pin, re-executing the first read transaction in one of:
a different column and a different row than a column and row where the first read transaction occurred,
the different column and the row where the first read transaction occurred, or
the different row and the column where the first read transaction occurred, and
wherein the spatial characteristic corresponds to corrective action that is used to correct the detected error; and
correcting the detected error based on the corrective action indicated in the fault profile.