Communication Optimizations for Multithreaded Code Generation from Simulink Models

Abstract

Communication frequency is increasing with the growing complexity of emerging embedded applications and the number of processors in the implemented multiprocessor SoC architectures. In this article, we consider the issue of communication cost reduction during multithreaded code generation from partitioned Simulink models to help designers in code optimization to improve system performance. We first propose a technique combining message aggregation and communication pipeline methods, which groups communications with the same destinations and sources and parallelizes communication and computation tasks. We also present a method to apply static analysis and dynamic emulation for efficient communication buffer allocation to further reduce synchronization cost and increase processor utilization. The existing cyclic dependency in the mapped model may hinder the effectiveness of the two techniques. We further propose a set of optimizations involving repartition with strongly connected threads to maximize the degree of communication reduction and preprocessing strategies with available delays in the model to reduce the number of communication channels that cannot be optimized. Experimental results demonstrate the advantages of the proposed optimizations with 11-143% throughput improvement.

Fine-Grained Communication-Aware Task Scheduling Approach for Acyclic and Cyclic Applications on MPSoCs

“…To address the communication challenge, two representative optimization techniques, communication pipeline and message aggregation techniques, have been proposed in [6]. Communication pipeline [6] can hide the receiving overhead by overlapping communication with the subsequent computation. Message aggregation [6], [7] combines the messages with the same sources and destinations to reduce the communication startup time.…”

“…Communication pipeline [6] can hide the receiving overhead by overlapping communication with the subsequent computation. Message aggregation [6], [7] combines the messages with the same sources and destinations to reduce the communication startup time. Nevertheless, the discussions of communication pipeline in the previous work [6] are not sufficient.…”

“…Message aggregation [6], [7] combines the messages with the same sources and destinations to reduce the communication startup time. Nevertheless, the discussions of communication pipeline in the previous work [6] are not sufficient. From the view of time, the description about communication pipeline lacks the consideration of the increasingly common situation that communication time is larger than its related computation time and the cases where using communication pipeline cannot save time.…”

“…From the view of time, the description about communication pipeline lacks the consideration of the increasingly common situation that communication time is larger than its related computation time and the cases where using communication pipeline cannot save time. From the view of topology, the previous work [6] introduces the method to employ communication pipeline on both acyclic and cyclic applications by preprocessing tasks. However, it is unclear which tasks and how many cycles of each task to be preprocessed, which requires theoretical guidance for users.…”

“…The approach works on the fine-grained Simulink model where communications are modeled as sending tasks and receiving tasks with each task including communication startup and transfer operations, and thus it schedules computation tasks, sending tasks, and receiving tasks together to efficiently utilize processors, as well as exploits communication pipeline to reduce communication transfer overhead. To better use communication pipeline, we give a thoroughly analysis of it from both time and topology perspectives based on [6]. After that, we introduce FCATS, which integrates scheduling with communication pipeline.…”