Title :
Reliability of protecting techniques used in fault-tolerant cache memories
Author :
Miremadi, Seyed Ghassem ; Zarandi, Hamid R.
Author_Institution :
Dept. of Comput. Eng., Sharif Univ. of Technol., Tehran
Abstract :
This paper analyzes the problem of transient-error recovery of several protecting techniques used in fault-tolerant cache memories. In this paper, reliability and mean-time-to-failure (MTTF) equations for several protecting techniques are derived and estimated. The results of the considered techniques are compared with those of cache memories without redundancies and with only parity codes in both tag and data arrays of caches. Depending on the error rate under which a cache memory will operate, and the size of the cache memory, one of the analyzed cases could be used. If the transient-error rate is very small or the size of cache memory is relatively small, then a protected with only single parity code is adequate. But for large cache memories or for noisy environment, with high transient-error rate, cache scrubbing or single error correction-double error detection technique (SEC-DED) become essential although the former incurs performance penalty and the latter occupies significant area and incurs complexity and power consumption penalties
Keywords :
cache storage; error correction codes; error detection codes; failure analysis; fault tolerance; cache scrubbing; fault-tolerant cache memories; mean-time-to-failure; protecting techniques; reliability; single error correction-double error detection technique; single parity code; transient-error rate; transient-error recovery; Cache memory; Computer errors; Electromagnetic transients; Equations; Error correction; Error correction codes; Fault tolerance; Protection; Redundancy; Reliability engineering;
Conference_Titel :
Electrical and Computer Engineering, 2005. Canadian Conference on
Conference_Location :
Saskatoon, Sask.
Print_ISBN :
0-7803-8885-2
DOI :
10.1109/CCECE.2005.1557054