Efficiency analysis of importance sampling in deep submicron STT-RAM design using uncontrollable industry-compatible model parameter

Taehui Na, Hanwool Jeong, Seongook Jung, Jung Pill Kim, Seung H. Kang

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

In this paper, we first analyze the efficiency of importance sampling (IS) method in spin transfer torque random access memory (STT-RAM) design with industry-compatible model parameter. Commonly used normal fitting method cannot estimate the yield accurately unless an output distribution follows the Gaussian distribution. The efficiency of IS method is significantly degraded when industry-compatible model parameters are used because most variables affected by process variation are not controllable. With industry-compatible 45-nm model parameters, Monte Carlo HSPICE simulation results show that the required number of simulations to satisfy error rate less than 5% should be greater than 50,000.

Original languageEnglish
Title of host publication2015 IEEE International Conference on Electronics, Circuits, and Systems, ICECS 2015
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages400-403
Number of pages4
Volume2016-March
ISBN (Electronic)9781509002467
DOIs
Publication statusPublished - 2016 Mar 23
EventIEEE International Conference on Electronics, Circuits, and Systems, ICECS 2015 - Cairo, Egypt
Duration: 2015 Dec 62015 Dec 9

Other

OtherIEEE International Conference on Electronics, Circuits, and Systems, ICECS 2015
CountryEgypt
CityCairo
Period15/12/615/12/9

Fingerprint

Importance sampling
Torque
Data storage equipment
Industry
Gaussian distribution

All Science Journal Classification (ASJC) codes

  • Electrical and Electronic Engineering

Cite this

Na, T., Jeong, H., Jung, S., Kim, J. P., & Kang, S. H. (2016). Efficiency analysis of importance sampling in deep submicron STT-RAM design using uncontrollable industry-compatible model parameter. In 2015 IEEE International Conference on Electronics, Circuits, and Systems, ICECS 2015 (Vol. 2016-March, pp. 400-403). [7440333] Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/ICECS.2015.7440333
Na, Taehui ; Jeong, Hanwool ; Jung, Seongook ; Kim, Jung Pill ; Kang, Seung H. / Efficiency analysis of importance sampling in deep submicron STT-RAM design using uncontrollable industry-compatible model parameter. 2015 IEEE International Conference on Electronics, Circuits, and Systems, ICECS 2015. Vol. 2016-March Institute of Electrical and Electronics Engineers Inc., 2016. pp. 400-403
@inproceedings{99f9a0479b6e4660926162dfb9943991,
title = "Efficiency analysis of importance sampling in deep submicron STT-RAM design using uncontrollable industry-compatible model parameter",
abstract = "In this paper, we first analyze the efficiency of importance sampling (IS) method in spin transfer torque random access memory (STT-RAM) design with industry-compatible model parameter. Commonly used normal fitting method cannot estimate the yield accurately unless an output distribution follows the Gaussian distribution. The efficiency of IS method is significantly degraded when industry-compatible model parameters are used because most variables affected by process variation are not controllable. With industry-compatible 45-nm model parameters, Monte Carlo HSPICE simulation results show that the required number of simulations to satisfy error rate less than 5{\%} should be greater than 50,000.",
author = "Taehui Na and Hanwool Jeong and Seongook Jung and Kim, {Jung Pill} and Kang, {Seung H.}",
year = "2016",
month = "3",
day = "23",
doi = "10.1109/ICECS.2015.7440333",
language = "English",
volume = "2016-March",
pages = "400--403",
booktitle = "2015 IEEE International Conference on Electronics, Circuits, and Systems, ICECS 2015",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
address = "United States",

}

Na, T, Jeong, H, Jung, S, Kim, JP & Kang, SH 2016, Efficiency analysis of importance sampling in deep submicron STT-RAM design using uncontrollable industry-compatible model parameter. in 2015 IEEE International Conference on Electronics, Circuits, and Systems, ICECS 2015. vol. 2016-March, 7440333, Institute of Electrical and Electronics Engineers Inc., pp. 400-403, IEEE International Conference on Electronics, Circuits, and Systems, ICECS 2015, Cairo, Egypt, 15/12/6. https://doi.org/10.1109/ICECS.2015.7440333

Efficiency analysis of importance sampling in deep submicron STT-RAM design using uncontrollable industry-compatible model parameter. / Na, Taehui; Jeong, Hanwool; Jung, Seongook; Kim, Jung Pill; Kang, Seung H.

2015 IEEE International Conference on Electronics, Circuits, and Systems, ICECS 2015. Vol. 2016-March Institute of Electrical and Electronics Engineers Inc., 2016. p. 400-403 7440333.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

TY - GEN

T1 - Efficiency analysis of importance sampling in deep submicron STT-RAM design using uncontrollable industry-compatible model parameter

AU - Na, Taehui

AU - Jeong, Hanwool

AU - Jung, Seongook

AU - Kim, Jung Pill

AU - Kang, Seung H.

PY - 2016/3/23

Y1 - 2016/3/23

N2 - In this paper, we first analyze the efficiency of importance sampling (IS) method in spin transfer torque random access memory (STT-RAM) design with industry-compatible model parameter. Commonly used normal fitting method cannot estimate the yield accurately unless an output distribution follows the Gaussian distribution. The efficiency of IS method is significantly degraded when industry-compatible model parameters are used because most variables affected by process variation are not controllable. With industry-compatible 45-nm model parameters, Monte Carlo HSPICE simulation results show that the required number of simulations to satisfy error rate less than 5% should be greater than 50,000.

AB - In this paper, we first analyze the efficiency of importance sampling (IS) method in spin transfer torque random access memory (STT-RAM) design with industry-compatible model parameter. Commonly used normal fitting method cannot estimate the yield accurately unless an output distribution follows the Gaussian distribution. The efficiency of IS method is significantly degraded when industry-compatible model parameters are used because most variables affected by process variation are not controllable. With industry-compatible 45-nm model parameters, Monte Carlo HSPICE simulation results show that the required number of simulations to satisfy error rate less than 5% should be greater than 50,000.

UR - http://www.scopus.com/inward/record.url?scp=84964831694&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84964831694&partnerID=8YFLogxK

U2 - 10.1109/ICECS.2015.7440333

DO - 10.1109/ICECS.2015.7440333

M3 - Conference contribution

AN - SCOPUS:84964831694

VL - 2016-March

SP - 400

EP - 403

BT - 2015 IEEE International Conference on Electronics, Circuits, and Systems, ICECS 2015

PB - Institute of Electrical and Electronics Engineers Inc.

ER -

Na T, Jeong H, Jung S, Kim JP, Kang SH. Efficiency analysis of importance sampling in deep submicron STT-RAM design using uncontrollable industry-compatible model parameter. In 2015 IEEE International Conference on Electronics, Circuits, and Systems, ICECS 2015. Vol. 2016-March. Institute of Electrical and Electronics Engineers Inc. 2016. p. 400-403. 7440333 https://doi.org/10.1109/ICECS.2015.7440333