A Comparison of C++ Sockets and Corba in a Distributed Matrix Multiply Application

摘要: This project has two primary purposes. The first, is to implement a distributed matrix multiply algorithm using C++ sockets, and Corba objects with the objective of discovering what additional overhead, if any, exists in implementation. Secondly, attempt improve speedup through use stateful servers CONCEPTS: Buffer Deadlock Size Limitations Client-Server Architecture Sockets CORBA Unix Name Service Speedup Efficiency Superscalar Technical Report # NPSCS-98-001 Alpay Duman Matt Schnaidt 6 SUMMARY did Socket implementation multiply. included writing all code (a 2d Array Class, client server for multiply, name server), testing, debugging, making record runs. I this three phases. In phase 1, implemented simple UDP sockets; used debug algorithm, interaction (which called memberServer). 2, converted TCP sockets so that they could reliably transmit messages which exceeded maximum packet size without concern ordering, lost or duplicate packets. finished 2 by measuring time it took do multiplies varied number matrices. 3, changed remembers state cut down on message traffic time. 3 multiplies, again divided report writing, slide development presentation equally. IDL, interface object defined IDL file, debugging one, file definition. For passing array type sequence Interface Definition Language, linear dynamic container. two, invoking deferred synchronous method, non-blocking remote procedure call method. three, special function send_multiple_requests_deferred(), initiates requests parallel. It does not wait finish before returning caller. 7 Table