On the application of dynamic scan chain partitioning for reducing peak shift power

Sobeeh Almukhaizim, Shouq Alsubaihi, Ozgur Sinanoglu

Research output: Contribution to journalArticle

Abstract

Scan-based testing of integrated circuits results in significant switching activity during the shift operations, dissipating excessive power levels. When such levels are beyond the peak power level under which the chip can functionally operate at, it may lead to an unexpected behavior of the design, resulting in a yield loss. One of the most effective solutions to reduce peak shift power is to partition the scan chains into multiple groups, wherein a single group is active at any time instance within a shift cycle. The partitioning of the chains into groups can be performed statically, i.e., per test set, or dynamically, i.e., per test pattern. In this work, we address the application of dynamic scan chain partitioning for reducing peak shift power. First, we address the application of dynamic partitioning to test delay faults in at-speed test techniques. Then, we formulate the scan chain partitioning problem via Integer Linear Programming (ILP), in order to evenly distribute the transitions produced by any pattern over multiple time instances within the shift cycle, maximally reducing the peak shift power. Finally, we evaluate the power reduction benefit of dynamic partitioning through an extensive set of experiments using different scan configurations and test set characteristics of benchmark circuits as well as industrial designs. The results indicate that dynamic partitioning provides significant reduction to peak shift power over static partitioning methods, and that the benefit is accentuated in scan architectures with fewer scan chains, test sets with more don't care bits, and designs with larger variances of weight differences for transitions in the scan cells.

Original languageEnglish (US)
Pages (from-to)465-481
Number of pages17
JournalJournal of Electronic Testing: Theory and Applications (JETTA)
Volume26
Issue number4
DOIs
StatePublished - Aug 1 2010

Fingerprint

Product design
Linear programming
Integrated circuits
Networks (circuits)
Testing
Experiments

Keywords

  • Dynamic partitioning
  • Low-power test
  • Peak power
  • Peak shift power
  • Scan chain partitioning
  • Scan-based test

ASJC Scopus subject areas

  • Electrical and Electronic Engineering

Cite this

On the application of dynamic scan chain partitioning for reducing peak shift power. / Almukhaizim, Sobeeh; Alsubaihi, Shouq; Sinanoglu, Ozgur.

In: Journal of Electronic Testing: Theory and Applications (JETTA), Vol. 26, No. 4, 01.08.2010, p. 465-481.

Research output: Contribution to journalArticle

@article{d6b90aa62a5e4b74acf5bdcaf9c52c78,
title = "On the application of dynamic scan chain partitioning for reducing peak shift power",
abstract = "Scan-based testing of integrated circuits results in significant switching activity during the shift operations, dissipating excessive power levels. When such levels are beyond the peak power level under which the chip can functionally operate at, it may lead to an unexpected behavior of the design, resulting in a yield loss. One of the most effective solutions to reduce peak shift power is to partition the scan chains into multiple groups, wherein a single group is active at any time instance within a shift cycle. The partitioning of the chains into groups can be performed statically, i.e., per test set, or dynamically, i.e., per test pattern. In this work, we address the application of dynamic scan chain partitioning for reducing peak shift power. First, we address the application of dynamic partitioning to test delay faults in at-speed test techniques. Then, we formulate the scan chain partitioning problem via Integer Linear Programming (ILP), in order to evenly distribute the transitions produced by any pattern over multiple time instances within the shift cycle, maximally reducing the peak shift power. Finally, we evaluate the power reduction benefit of dynamic partitioning through an extensive set of experiments using different scan configurations and test set characteristics of benchmark circuits as well as industrial designs. The results indicate that dynamic partitioning provides significant reduction to peak shift power over static partitioning methods, and that the benefit is accentuated in scan architectures with fewer scan chains, test sets with more don't care bits, and designs with larger variances of weight differences for transitions in the scan cells.",
keywords = "Dynamic partitioning, Low-power test, Peak power, Peak shift power, Scan chain partitioning, Scan-based test",
author = "Sobeeh Almukhaizim and Shouq Alsubaihi and Ozgur Sinanoglu",
year = "2010",
month = "8",
day = "1",
doi = "10.1007/s10836-010-5159-6",
language = "English (US)",
volume = "26",
pages = "465--481",
journal = "Journal of Electronic Testing: Theory and Applications (JETTA)",
issn = "0923-8174",
publisher = "Springer Netherlands",
number = "4",

}

TY - JOUR

T1 - On the application of dynamic scan chain partitioning for reducing peak shift power

AU - Almukhaizim, Sobeeh

AU - Alsubaihi, Shouq

AU - Sinanoglu, Ozgur

PY - 2010/8/1

Y1 - 2010/8/1

N2 - Scan-based testing of integrated circuits results in significant switching activity during the shift operations, dissipating excessive power levels. When such levels are beyond the peak power level under which the chip can functionally operate at, it may lead to an unexpected behavior of the design, resulting in a yield loss. One of the most effective solutions to reduce peak shift power is to partition the scan chains into multiple groups, wherein a single group is active at any time instance within a shift cycle. The partitioning of the chains into groups can be performed statically, i.e., per test set, or dynamically, i.e., per test pattern. In this work, we address the application of dynamic scan chain partitioning for reducing peak shift power. First, we address the application of dynamic partitioning to test delay faults in at-speed test techniques. Then, we formulate the scan chain partitioning problem via Integer Linear Programming (ILP), in order to evenly distribute the transitions produced by any pattern over multiple time instances within the shift cycle, maximally reducing the peak shift power. Finally, we evaluate the power reduction benefit of dynamic partitioning through an extensive set of experiments using different scan configurations and test set characteristics of benchmark circuits as well as industrial designs. The results indicate that dynamic partitioning provides significant reduction to peak shift power over static partitioning methods, and that the benefit is accentuated in scan architectures with fewer scan chains, test sets with more don't care bits, and designs with larger variances of weight differences for transitions in the scan cells.

AB - Scan-based testing of integrated circuits results in significant switching activity during the shift operations, dissipating excessive power levels. When such levels are beyond the peak power level under which the chip can functionally operate at, it may lead to an unexpected behavior of the design, resulting in a yield loss. One of the most effective solutions to reduce peak shift power is to partition the scan chains into multiple groups, wherein a single group is active at any time instance within a shift cycle. The partitioning of the chains into groups can be performed statically, i.e., per test set, or dynamically, i.e., per test pattern. In this work, we address the application of dynamic scan chain partitioning for reducing peak shift power. First, we address the application of dynamic partitioning to test delay faults in at-speed test techniques. Then, we formulate the scan chain partitioning problem via Integer Linear Programming (ILP), in order to evenly distribute the transitions produced by any pattern over multiple time instances within the shift cycle, maximally reducing the peak shift power. Finally, we evaluate the power reduction benefit of dynamic partitioning through an extensive set of experiments using different scan configurations and test set characteristics of benchmark circuits as well as industrial designs. The results indicate that dynamic partitioning provides significant reduction to peak shift power over static partitioning methods, and that the benefit is accentuated in scan architectures with fewer scan chains, test sets with more don't care bits, and designs with larger variances of weight differences for transitions in the scan cells.

KW - Dynamic partitioning

KW - Low-power test

KW - Peak power

KW - Peak shift power

KW - Scan chain partitioning

KW - Scan-based test

UR - http://www.scopus.com/inward/record.url?scp=77956265267&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=77956265267&partnerID=8YFLogxK

U2 - 10.1007/s10836-010-5159-6

DO - 10.1007/s10836-010-5159-6

M3 - Article

VL - 26

SP - 465

EP - 481

JO - Journal of Electronic Testing: Theory and Applications (JETTA)

JF - Journal of Electronic Testing: Theory and Applications (JETTA)

SN - 0923-8174

IS - 4

ER -