Performance modeling of the Ada rendezvous. Final report

Technical Report · Tue Oct 01 04:00:00 EDT 1991

OSTI ID:5745722

Sterrett, A E; Minei, M K

The Ada Programming Language (United States Department of Defense, 1983) was designed to meet the need for a standard computer-programming language. Ada has the ability to take advantage of microprocessor environments. One feature, known as the rendezvous, allows tasks to synchronize. This very important and powerful feature is poorly understood. Rendezvous performance is a known area of concern, especially if the system is in a multiprocessor environment. In this paper, analytic techniques are developed that will predict the performance of systems using the rendezvous. In a distributed Ada system, the rendezvous provides synchronized communication between asynchronous tasks. A system of this sort would consist of at least two processors, each serving various tasks. We have analytically developed algorithms that determine the average Rendezvous Response Time for a two-processor system. Rendezvous Response Time will be defined as the amount of time one task (i.e., a client task) must wait until its rendezvous request to another task (i.e., a server task) is completed. This approach uses Mean Value Analysis (MVA), analytic extensions to MVA, and elementary queueing theory to decompose the rendezvous into interacting separate models. The notion of a software server that maintains a queue and services rendezvous requests will play an important part in the solution techniques.

• Research Organization: Naval Ocean Systems Center, San Diego, CA (United States)
• OSTI ID: 5745722
• Report Number(s): AD-A-243392/8/XAB; NOSC/TD--2194
• Country of Publication: United States
• Language: English

Related Subjects

99 GENERAL AND MISCELLANEOUS
990200* -- Mathematics & Computers
ADA
ALGORITHMS
ARRAY PROCESSORS
COMMUNICATIONS
COMPUTERS
DATA PROCESSING
DATA TRANSMISSION
DISTRIBUTED DATA PROCESSING
ELECTRONIC CIRCUITS
MATHEMATICAL LOGIC
MICROELECTRONIC CIRCUITS
MICROPROCESSORS
PERFORMANCE
PROCESSING
PROGRAMMING LANGUAGES
SYNCHRONIZATION
TASK SCHEDULING