Abstract: Computer hardware technology has shown tremendous growth from simple uniprocessor to bigger and faster uniprocessors. In the recent past, these uniprocessors are replaced by multi-processor cores. To match these capabilities, the programming paradigms have also shifted from single-threaded sequential programming to multi-threaded parallel programming. But Legacy systems which were developed primarily to run on uniprocessors were left behind. This paper proposes to transform these applications to run on multi-processors at the same abstraction level (Implementation level) using semi-Automatic techniques. Here the legacy program is analysed based on dependency between functions and its signatures. They are then divided into smaller sub-programs (Parallel threads), which are executed on multi-core machines. The execution time of these restructured legacy programs were analyzed and observed ameliorated efficiency.
Keywords: Parallel threads; Re-engineering; Multicore machines; Slicing
|
DOI:
10.17148/IJARCCE.2021.101251
