Suitability of Ada for Real-Time Model Based Vision Applications
Abstract
The purpose of this study was to establish a baseline for assessing the execution efficiency of Model Based Vision (MBV) algorithms coded in both the Ada and C programming languages. To facilitate this study a key algorithm used in The Analytic Science Corporations's (TASC) laboratory MBV system called Fast Pairwise Nearest Neighbor (FASTPNN) was recoded from its original C form to Ada, and benchmarked on both a VAX 11/780 and MIPS MAGNUM 3000 computer system. Comparisons of C execution efficiency versus Ada execution efficiency as well as MIPS MAGNUM 3000 execution efficiency versus VAX 11/780 execution efficiency were made. The benchmark results indicate that when Ada run-time checks are suppressed, Ada and C are roughly equivalent in terms of inherent execution efficiency. Differences between Ada and C execution efficiency can be attributed to deviations in a particular compiler's maturity. Ada run-time checks impose between a 43% and 65% execution penalty when compared with Ada executing with all run-time checks suppressed. Depending on whether C or Ada was the language of interest, the MIPS MAGNUM 3000 demonstrated between a 9 (Ada) to 22 (C) times throughput advantage over the VAX 11/780.
Document Details
- Document Type
- Technical Report
- Publication Date
- Nov 01, 1991
- Accession Number
- ADA245710
Entities
People
- Phillip B. Hanselman
- Robert Kaplan
Organizations
- Wright Laboratory