DocumentCode
2584529
Title
Software-based self-test for pipelined processors: a case study
Author
Hatzimihail, M. ; Xenoulis, G. ; Psarakis, M. ; Gizopoulos, D. ; Paschalis, A.
Author_Institution
Dept. of Informatics, Piraeus Univ., Greece
fYear
2005
fDate
3-5 Oct. 2005
Firstpage
535
Lastpage
543
Abstract
Software-based self-test (SBST) for processors and processor-based systems recently captured the interest of test technology researchers and practitioners due to its several advantages over traditional hardware built-in self-test (BIST). In this paper, we demonstrate for first time the full applicability of a recently proposed SBST methodology to a publicly available complex RISC processor implementation which includes a full pipelined architecture consisting of five pipeline stages, hazard detection, data forwarding and exceptions handling. We first show that the straightforward application of SBST routines developed for the nonpipelined version of the RISC processor can only reach a fault coverage less than 85% in the fully pipelined model. Then, we identify and classify areas with poor testability and provide solutions that extend the SBST methodology and achieve fault coverage more than 95% for this complex processor implementation.
Keywords
automatic test software; built-in self test; electronic engineering computing; integrated circuit testing; microprocessor chips; pipeline processing; reduced instruction set computing; RISC processors; complex processor implementation; fault coverage; full pipelined architecture; pipelined processors; processor-based systems; software-based self-test; Automatic testing; Built-in self-test; Circuit faults; Circuit testing; Computer aided software engineering; Hardware; Hazards; Informatics; Pipelines; Reduced instruction set computing;
fLanguage
English
Publisher
ieee
Conference_Titel
Defect and Fault Tolerance in VLSI Systems, 2005. DFT 2005. 20th IEEE International Symposium on
ISSN
1550-5774
Print_ISBN
0-7695-2464-8
Type
conf
DOI
10.1109/DFTVS.2005.63
Filename
1544553
Link To Document