A proposed parallel software testing paradigm
J.L. Walker · 2002
The author presents a testing paradigm for parallel software based upon the technique of validating the software directly according to its behavior. It is believed that this paradigm will provide reproducible information about the validity of parallel software in an efficient manner. It provides for controlled execution of the software in order to determine the software's ability to produce correct outputs in a wide variety of situations. It also provides for passive observation and validation to determine the software's reliability in its operational environment.>