Protected application decomposition is limited by the performance of the local interprocess procedure call implementation. In this paper, we measure the performance of a new IPC implementation, and show that the cost of IPC can be reduced to the cost of raw memory copy plus a small overhead. Even on machines with poor context switching performance, our implementation compares favorably with bcopy() for surprisingly small payloads.
|Number of pages||6|
|Journal||International Workshop on Object Orientation in Operating Systems - Proceedings|
|Publication status||Published - 1996 Dec 1|
|Event||Proceedings of the 1996 5th International Workshop on Object Orientation in Operating Systems - Seattle, WA, USA|
Duration: 1996 Oct 27 → 1996 Oct 28
ASJC Scopus subject areas
- Hardware and Architecture