Date of Original Version



Technical Report

Abstract or Description

The primary purpose of the Ada Embedded Systems Testbed (AEST) Project at the Software Engineering Institute (SEI) is to develop a solid in-house support base of hardware, software, and personnel to permit the investigation of a wide variety of issues related to software development for real-time embedded systems. Two of the most crucial issues to be investigated are the extent and quality of the facilities provided by Ada runtime support environments. The SEI support base will make assessments possible of the readiness of the Ada language and Ada tools to develop embedded systems. The benchmarking/instrumentation subgroup was formed to:

1. Collect and run available Ada benchmark programs from a variety of sources on a variety of targets.

2. Identify gaps in the coverage and fill them with new test programs.

3. Review the measurement techniques used and provide new ones if necessary.

4. Verify software timings by inspection and with specialized test instruments.

This report documents the results obtained from running Ada performance benchmarks on a DEC VAXELN MicroVAX II using the DEC VAXELN Ada compiler. The benchmarks were the University of Michigan Ada benchmarks and the ACM SIGAda Performance Issues Working Group (PIWG) Ada benchmarks (excluding the compilation tests). The benchmarks focus largely on the execution time of specific features of the Ada language; they do not, for example, measure the efficiency or the size of the generated object code. A brief description of the benchmarks and the test environment is followed by a discussion of some problems encountered and lessons learned. The results obtained from running the entire Michigan and PIWG benchmark suites are contained in the appendices to this report. Note that the caveats discussed in the body of the report must be borne in mind when examining these results.