Deriving Efficient Data Movement From Decoupled Access/Execute Specifications

Abstract. On multi-core architectures with software-managed memories, effec-tively orchestrating data movement is essential to performance, but is tedious and error-prone. In this paper we show that when the programmer can explicitly specify both the memory access pattern and the execution schedule of a com-putation kernel, the compiler or run-time system can derive efficient data move-ment, even if analysis of kernel code is difficult or impossible. We have devel-oped a framework of C++ classes for decoupled Access/Execute specifications, allowing for automatic communication optimisations such as software pipelin-ing and data reuse. We demonstrate the ease and efficiency of programming the Cell Broadband Engine architecture using these classes by implementing a set of benchmarks, which exhibit data reuse and non-affine access functions, and by comparing these implementations against alternative implementations, which use hand-written DMA transfers and software-based caching.