Skip to Main Content
The development of embedded system has been toward the multicore architectures in the recent years. It raises concerns in the community of supporting programming models and languages to derive maximal performance from the architectures. Among the diversity of models for programming multicore processors, remote procedure call (RPC) is one of the most relevant programming techniques for supporting an explicit parallel programming model. Although such promising programming technique provides an easy way of modeling the applications on multiple processors, it remains an interesting and challenging problem of how to provide an effective system of programming data-intensive applications under the programming scenario of RPC. In this paper, we propose a streaming mechanism called streaming RPC to provide a system for modeling data-intensive and stream-based applications to efficiently utilize the constituents of the multicore processors. Streaming RPC is based on the framework of RPC and implemented as a middleware support to provide a library-based programming model with parallelism by mandatory. We also propose design patterns for the streaming mechanism and present experiences of developing high performance multimedia applications. Experimental results show that our streaming RPC framework is efficient to support multicore programming for multimedia applications.