From nobody Wed Feb 28 15:42 EST 1996
Date: Wed, 28 Feb 1996 15:42:58 -0500 (EST)
From: uid no body <nobody>
To: techreps@cs.buffalo.edu
Subject: techrep: POST request
Content-Type: text
Content-Length: 1381

Comments: castor
ContactPerson: sreejit@cs.buffalo.edu
Remote host: saiph.cs.buffalo.edu
Remote ident: sreejit
### Begin Citation ### Do not delete this line ###
%R 96-03
%U /u4/csstaff/milun/head.ps
%A Chakravarty, Sreejit
%A Thadikaran, Paul J.
%T Which Set of Bridging Faults Should Test Compilers Target?
%D February 28, 1996
%I Department of Computer Science, SUNY Buffalo
%K Bridging Faults, Fault Simulation, IDDQ Testing, Test Compilation, Test Compression,Test Generation
%Y B.1.3; B.2.3; B.3.4; B.5.3; B.6.2; B.6.3; B.7.2
%X Existing test compilers target stuck-at faults. Recent experimental 
evidence suggest targetting bridging faults(BFs) because they model 
between 40-50% of physical defects. For BFs, one suggestion has  
been to extract a small set of BFs using "layout analysis". We argue 
that this is not a feasible option for test compilers thereby making 
a case for "layout independent analysis" of BFs.   
	Opting for "layout independent analysis" implies that  
computer-aided test tools target "all possible BFs". Targetting 
such a large number of faults is generally considered unreasonable. 
We present new experimental data and analyze published data. The 
analysis reveal that for many problems very good techniques have 
already been developed to target such a large number of faults.