A Review on Software Architectures for Heterogeneous Platforms

Abstract

The increasing demands for computing performance have been a reality regardless of the requirements for smaller and more energy efficient devices. Throughout the years, the strategy adopted by industry was to increase the robustness of a single processor by increasing its clock frequency and mounting more transistors so more calculations could be executed. However, it is known that the physical limits of such processors are being reached, and one way to fulfill such increasing computing demands has been to adopt a strategy based on heterogeneous computing, i.e., using a heterogeneous platform containing more than one type of processor. This way, different types of tasks can be executed by processors that are specialized in them. Heterogeneous computing, however, poses a number of challenges to software engineering, especially in the architecture and deployment phases. In this paper, we conduct an empirical study that aims at discovering the state-of-the-art in software architecture for heterogeneous computing, with focus on deployment. We conduct a systematic mapping study that retrieved 28 studies, which were critically assessed to obtain an overview of the research field. We identified gaps and trends that can be used by both researchers and practitioners as guides to further investigate the topic.

Refactoring Software in the Automotive Domain for Execution on Heterogeneous Platforms

“…First, we studied the available literature on the topics of software deployment [7] and software architectures [8] for heterogeneous platforms in order to obtain an overview of the research area and identify the challenges, concerns, and gaps in research. In this stage, we identified a number of approaches to realize software architecture when a heterogeneous platform is available.…”

“…Twenty-eight studies discussing concerns of software architectures for heterogeneous computing were identified previously in [8]. These studies typically propose solutions to specific problems, rather than a holistic framework to aid in the process of migrating to heterogeneous platforms.…”

An open computing language‐based parallel Brute Force algorithm for formal concept analysis on heterogeneous architectures

“…As already mentioned, there is a great need for computational resources to process high volumes of information. A survey 27 showed that the use of heterogeneous architectures with accelerators (FPGA/GPU) has proved to be quite promising in several aspects related to big data. For instance, a previous work also used custom accelerators for a Hadoop MapReduce framework 16 …”

Arquitetura Heterogênea CPU+FPGA para Análise Formal de Conceitos

“…Contudo, estudos com FPGA mostram ser bastante promissores para mineração de dados e tratamento de um grande volume de dados. Um levantamento de diversos trabalhos que mostraram o uso de arquiteturas heterogêneas, referindo-se ao uso de diversos dispositivos em conjunto, como CPU, GPU, FPGA, foi realizado em [Andrade and Crnkovic 2019]. Sendo assim, no trabalhoé apresentado o estudo de 28 pesquisas relacionadasà computação heterogênea, de formaà realizar críticas construtivas da relevância e problemas que estaárea de pesquisa pode oferecer.…”