An Exploration of Microprocessor Self-Test Optimisation Based On Safe Faults

Narang, Anuraag ORCID: 0000-0001-8033-1261, Venu, Balaji, Khursheed, Syed-Saqib ORCID: 0000-0002-5720-0607 and Harrod, Peter (2021) An Exploration of Microprocessor Self-Test Optimisation Based On Safe Faults In: 34th IEEE International Symposium on Defect and Fault Tolerance in VLSI and Nanotechnology Systems, 2021-10-6 - 2021-10-8.

Text DFT_2021_STL_OPTIMISATION_PAPER.pdf - Author Accepted Manuscript Download (174kB) | Preview

Abstract

Microprocessor software test libraries (STLs) must provide maximum fault coverage with minimum overhead. Pruning safe faults, which cannot cause errors in the output of the processor, from the fault list can increase fault coverage without adding test overhead. Applying more application-specific constraints can lead to the identification of more safe faults, and some such constraints are yet to be explored. This work explores the use of signal combination-based constraints alongside well-known constant signal-based constraints for identifying safe faults. Also, for the first time, information on safe faults is utilised during test compaction in order to further minimise test overhead. Results for an OpenRISC processor design show up to 2.33% improvement in fault coverage with the use of the proposed constraints. In one test program, a code segment contributing only to the coverage of safe faults is identified, with its removal providing a 1.09% code size reduction on top of existing compaction techniques. The results may vary for a larger and more complex commercial design with greater scope for redundant logic.

Item Type:	Conference Item (Unspecified)
Uncontrolled Keywords:	software-based self-test, software test library, test quality, test compaction
Divisions:	Faculty of Science & Engineering > School of Electrical Engineering, Electronics and Computer Science
Depositing User:	Symplectic Admin
Date Deposited:	23 Aug 2021 10:06
Last Modified:	28 Feb 2026 23:57
DOI:	10.1109/DFT52944.2021.9568326
Related Websites:	Author Publisher
URI:	https://livrepository.liverpool.ac.uk/id/eprint/3134180
Disclaimer:	The University of Liverpool is not responsible for content contained on other websites from links within repository metadata. Please contact us if you notice anything that appears incorrect or inappropriate.