Title :
A Tag based solution for efficient utilization of efuse for memory repair
Author :
Ellur, Harsharaj ; Shah, Kalpesh
Author_Institution :
Texas Instrum., Inc., Bangalore, India
Abstract :
Efuse or FuseROMs play a major role in embedded memory BIST and repair flows. The trend of increasing demand for SRAMs in SoCs for graphics, DSP and application processors have resulted in the need for more efuses on die to repair these memories. Efuse bit cell is comparatively large in size and does not scale well at lower technology nodes, thus becoming a floor plan nightmare. All these factors call for an efficient use of available efuses. This paper describes the implementation of an efuse compression scheme that identifies repair registers by tags and allocates efuse to only the repair registers that need a repair code. Linked list based data structure was utilized for storing and retrieving repair codes from the FuseROM. This implementation was designed with the goal of achieving good compression when using incremental repair. Analysis indicate compression ratio as high as 88.5% and a max improvement of 35% over RLE based compression schemes with logic overhead of ~10K gates.
Keywords :
SRAM chips; built-in self test; data compression; data structures; embedded systems; integrated circuit testing; read-only storage; system-on-chip; DSP; FuseROMs; RLE based compression schemes; SRAMs; SoCs; application processors; efuse bit cell; efuse compression scheme; embedded memory BIST; floor plan nightmare; graphics; incremental repair; linked list based data structure; logic overhead; memory repair; repair code; repair registers; tag based solution; technology nodes; Fuses; Maintenance engineering; Memory management; Random access memory; Registers; Sorting; System-on-chip; dynamic allocation; efuse; memory repair; optimization; tags;
Conference_Titel :
Test Conference (ITC), 2014 IEEE International
Conference_Location :
Seattle, WA
DOI :
10.1109/TEST.2014.7035324