Cost-quality trade-offs of approximate memory repair mechanisms for image data

Qianqian Fan; Sachin S. Sapatnekar; David J. Lilja

doi:10.1109/ISQED.2017.7918355

Cost-quality trade-offs of approximate memory repair mechanisms for image data

Qianqian Fan
, Sachin S. Sapatnekar
, David J. Lilja

Electrical and Computer Engineering

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

2 Scopus citations

Abstract

The traditional approach for increasing yield in large memory arrays has been to eliminate all hard errors using repair mechanisms. However, the cost of these mechanisms can become prohibitive for cheaper memories, which have higher error rates. Instead of completely repairing faulty cells, this paper introduces new approximate memory repair mechanisms that only partially repair both CMOS DRAMs and STT-MRAMs. By combining redundant repair with unequal protection, such as skewing the limited spare elements available for repairing faults towards the k most significant bits, and a hybrid bit-shuffling and redundant repair scheme, the new mechanisms maintain excellent output quality while substantially reducing the cost of the repair mechanism, particularly for increasingly important cluster faults.

Original language	English (US)
Title of host publication	Proceedings of the 18th International Symposium on Quality Electronic Design, ISQED 2017
Publisher	IEEE Computer Society
Pages	438-444
Number of pages	7
ISBN (Electronic)	9781509054046
DOIs	https://doi.org/10.1109/ISQED.2017.7918355
State	Published - May 2 2017
Event	18th International Symposium on Quality Electronic Design, ISQED 2017 - Santa Clara, United States Duration: Mar 14 2017 → Mar 15 2017

Publication series

Name	Proceedings - International Symposium on Quality Electronic Design, ISQED
ISSN (Print)	1948-3287
ISSN (Electronic)	1948-3295

Other

Other	18th International Symposium on Quality Electronic Design, ISQED 2017
Country/Territory	United States
City	Santa Clara
Period	3/14/17 → 3/15/17

Bibliographical note

Publisher Copyright:
© 2017 IEEE.

Keywords

Approximate memory repair
CMOS DRAMs
STT-MRAMs
unequal protection

Publisher link

10.1109/ISQED.2017.7918355

Find It

Find It at U of M Libraries

Cite this

Fan, Q., Sapatnekar, S. S., & Lilja, D. J. (2017). Cost-quality trade-offs of approximate memory repair mechanisms for image data. In Proceedings of the 18th International Symposium on Quality Electronic Design, ISQED 2017 (pp. 438-444). Article 7918355 (Proceedings - International Symposium on Quality Electronic Design, ISQED). IEEE Computer Society. https://doi.org/10.1109/ISQED.2017.7918355

Cost-quality trade-offs of approximate memory repair mechanisms for image data. / Fan, Qianqian; Sapatnekar, Sachin S.; Lilja, David J.
Proceedings of the 18th International Symposium on Quality Electronic Design, ISQED 2017. IEEE Computer Society, 2017. p. 438-444 7918355 (Proceedings - International Symposium on Quality Electronic Design, ISQED).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Fan, Q, Sapatnekar, SS & Lilja, DJ 2017, Cost-quality trade-offs of approximate memory repair mechanisms for image data. in Proceedings of the 18th International Symposium on Quality Electronic Design, ISQED 2017., 7918355, Proceedings - International Symposium on Quality Electronic Design, ISQED, IEEE Computer Society, pp. 438-444, 18th International Symposium on Quality Electronic Design, ISQED 2017, Santa Clara, United States, 3/14/17. https://doi.org/10.1109/ISQED.2017.7918355

@inproceedings{89d96f1eb85d489aa65eeebeefce8d20,

title = "Cost-quality trade-offs of approximate memory repair mechanisms for image data",

abstract = "The traditional approach for increasing yield in large memory arrays has been to eliminate all hard errors using repair mechanisms. However, the cost of these mechanisms can become prohibitive for cheaper memories, which have higher error rates. Instead of completely repairing faulty cells, this paper introduces new approximate memory repair mechanisms that only partially repair both CMOS DRAMs and STT-MRAMs. By combining redundant repair with unequal protection, such as skewing the limited spare elements available for repairing faults towards the k most significant bits, and a hybrid bit-shuffling and redundant repair scheme, the new mechanisms maintain excellent output quality while substantially reducing the cost of the repair mechanism, particularly for increasingly important cluster faults.",

keywords = "Approximate memory repair, CMOS DRAMs, STT-MRAMs, unequal protection",

author = "Qianqian Fan and Sapatnekar, \{Sachin S.\} and Lilja, \{David J.\}",

note = "Publisher Copyright: {\textcopyright} 2017 IEEE.; 18th International Symposium on Quality Electronic Design, ISQED 2017 ; Conference date: 14-03-2017 Through 15-03-2017",

year = "2017",

month = may,

day = "2",

doi = "10.1109/ISQED.2017.7918355",

language = "English (US)",

series = "Proceedings - International Symposium on Quality Electronic Design, ISQED",

publisher = "IEEE Computer Society",

pages = "438--444",

booktitle = "Proceedings of the 18th International Symposium on Quality Electronic Design, ISQED 2017",

}

TY - GEN

T1 - Cost-quality trade-offs of approximate memory repair mechanisms for image data

AU - Fan, Qianqian

AU - Sapatnekar, Sachin S.

AU - Lilja, David J.

PY - 2017/5/2

Y1 - 2017/5/2

N2 - The traditional approach for increasing yield in large memory arrays has been to eliminate all hard errors using repair mechanisms. However, the cost of these mechanisms can become prohibitive for cheaper memories, which have higher error rates. Instead of completely repairing faulty cells, this paper introduces new approximate memory repair mechanisms that only partially repair both CMOS DRAMs and STT-MRAMs. By combining redundant repair with unequal protection, such as skewing the limited spare elements available for repairing faults towards the k most significant bits, and a hybrid bit-shuffling and redundant repair scheme, the new mechanisms maintain excellent output quality while substantially reducing the cost of the repair mechanism, particularly for increasingly important cluster faults.

AB - The traditional approach for increasing yield in large memory arrays has been to eliminate all hard errors using repair mechanisms. However, the cost of these mechanisms can become prohibitive for cheaper memories, which have higher error rates. Instead of completely repairing faulty cells, this paper introduces new approximate memory repair mechanisms that only partially repair both CMOS DRAMs and STT-MRAMs. By combining redundant repair with unequal protection, such as skewing the limited spare elements available for repairing faults towards the k most significant bits, and a hybrid bit-shuffling and redundant repair scheme, the new mechanisms maintain excellent output quality while substantially reducing the cost of the repair mechanism, particularly for increasingly important cluster faults.

KW - Approximate memory repair

KW - CMOS DRAMs

KW - STT-MRAMs

KW - unequal protection

UR - https://www.scopus.com/pages/publications/85019557345

UR - https://www.scopus.com/pages/publications/85019557345#tab=citedBy

U2 - 10.1109/ISQED.2017.7918355

DO - 10.1109/ISQED.2017.7918355

M3 - Conference contribution

AN - SCOPUS:85019557345

T3 - Proceedings - International Symposium on Quality Electronic Design, ISQED

SP - 438

EP - 444

BT - Proceedings of the 18th International Symposium on Quality Electronic Design, ISQED 2017

PB - IEEE Computer Society

T2 - 18th International Symposium on Quality Electronic Design, ISQED 2017

Y2 - 14 March 2017 through 15 March 2017

ER -

Cost-quality trade-offs of approximate memory repair mechanisms for image data

Abstract

Publication series

Other

Bibliographical note

Keywords

Publisher link

Find It

Other files and links

Fingerprint

Cite this