Title :
Thoughts on high speed data bus performance
Author :
Ludvigson, Merrill T.
Author_Institution :
Rockwell Int. Corp., Palo Alto, CA, USA
Abstract :
The author believes that average performance calculations do not provide an adequate method for evaluating bus performance and making engineering design decisions in the presence of bursty traffic, because bursty traffic significantly changes those factors that most impact on bus performance. It has been found that very subtle changes may have significant impact on bus performance in unexpected ways. An example is the choice of when to reset the token rotation timers to achieve optimum priority performance. During the PAVE PILLAR program, the Collins Government Avionics Division of Rockwell International recommended that the timers be reset when the token arrives. Others recommended that the timers be reset when the token is sent. The author discusses the reasons for the original recommendation, how a simulation failed to confirm those original conclusions, and how a few simple equations have since been derived which have led to a new set of conclusions. Subtle points pertaining to high-speed data bus characteristics that affect bus performance are discussed
Keywords :
computer interfaces; digital simulation; protocols; PAVE PILLAR program; bursty traffic; engineering design decisions; high speed data bus performance; timers; Aerospace electronics; Analog circuits; Analytical models; Bandwidth; Circuit analysis; Delay; Government; Performance analysis; Performance evaluation; Traffic control;
Conference_Titel :
Aerospace and Electronics Conference, 1990. NAECON 1990., Proceedings of the IEEE 1990 National
Conference_Location :
Dayton, OH
DOI :
10.1109/NAECON.1990.112760