SCI-Clone/32 has been shown to meet the objectives hoped for from a distributed real-time system. It allows the addition of processing nodes to an existing distributed or monolithic system. The system builder can incorporate additional features into his design without the need to reconfigure his existing hardware or software. The distributed system is fully supportive of functional development being done on a stand alone system. If the structure of the common partition is well defined software may be designed and tested on a single node then simply plugged into the total configuration for final testing. The architecture only broadcasts the update of a common variable and this is simultaneously received by all other nodes. All reading of variables and fetching of instructions is done locally within each node. Therefore the traffic on the reflective memory bus is dramatically reduced over other distributed methods. The bandwidth of the bus is very high and hence adding further nodes is unlikely to generate so much cross-talk to affect the overall performance.
SCI-Clone/32 computer system for real-time computing
SCI-Clone/32-Rechnersystem fuer Realzeitbetrieb
1986
4 Seiten, 3 Bilder
Conference paper
English
SAE Technical Papers | 1974
|Clone detection in automotive model-based development
Tema Archive | 2008
|