Cost Effective Adaptive Voltage Scaling Using Path Delay Fault Testing

Mahroo Zandrahimi, Philippe Debaud, Armand Castillejo, Zaid Al-Ars

Research output: Chapter in Book/Conference proceedings/Edited volumeConference contributionScientificpeer-review

Abstract

Application of manufacturing testing during the production process of integrated circuits is considered essential to ensure the quality of the devices used in the field. However, it is desirable to use the information gathered during the test process to add value to other aspects of the manufacturing process. This paper proposes a method to use path delay (PDLY) test patterns, not only to validate the functionality of the devices, but also as an alternative solution for performance estimation, that can be used for offline adaptive voltage scaling. This approach has many advantages over the currently used industrial performance estimation methods, so-called performance monitoring boxes (PMBs). Using simulation of ISCAS'99 benchmarks with 28nm FD-SOI libraries, the paper shows that the PDLY based approach reduces the inaccuracy of performance prediction from 2.32% (achieved by the classic PMB approach) to 1.85%, without the need for any on-chip monitors.

Original languageEnglish
Title of host publication2018 IEEE East-West Design and Test Symposium ( EWDTS)
Place of PublicationDanvers
PublisherIEEE
Pages1-6
Number of pages6
ISBN (Electronic)978-1-5386-5710-2
ISBN (Print)978-1-5386-5711-9
DOIs
Publication statusPublished - 2018
Event2018 IEEE East-West Design and Test Symposium, EWDTS 2018 - Kazan, Russian Federation
Duration: 14 Sept 201817 Sept 2018

Conference

Conference2018 IEEE East-West Design and Test Symposium, EWDTS 2018
Country/TerritoryRussian Federation
CityKazan
Period14/09/1817/09/18

Keywords

  • adaptive voltage scaling
  • critical path replica
  • path delay testing
  • performance estimation
  • process monitor boxes

Fingerprint

Dive into the research topics of 'Cost Effective Adaptive Voltage Scaling Using Path Delay Fault Testing'. Together they form a unique fingerprint.

Cite this