首页> 外文会议>International Conference on Advances in Circuits, Electronics and Micro-electronics >Evaluating Heterogeneous Architectures based on Zynq AP SOC for Real-Time Video Processing
【24h】

Evaluating Heterogeneous Architectures based on Zynq AP SOC for Real-Time Video Processing

机译:基于Zynq AP SOC的基于Zynq AP SoC进行评估,进行实时视频处理

获取原文

摘要

Embedded systems are known as valid candidates to efficiently support image and video processing algorithms. Their high flexibility, speed performances and low power consumption are mainly due to the joint design of their software and specific hardware portions. The Zynq All Programmable System on Chip, that integrates ARM processor and high performance programmable logic resources, is nowadays often preferred to the more traditional realization platforms, such as Application Specific Integrated Circuits (ASICs) and Digital Signal Processors (DSPs). In this paper, we evaluate two different design strategies, each with its own Zynq-based support platform, giving to the designers useful hints on how to identify the best design choices for the target application. The first support platform presented here also makes use of an embedded operating system (OS); it significantly limits the required design efforts and time-to-market. The second architecture is realized without the OS support, and of course reaches much higher performances than the former, but requiring higher development and verification times. Both platforms exploit a hardware accelerator for the function of interest. As a case study, a simple but complete image processing architecture has been designed by using both the above platforms. Performances measurements revealed that an approximate speed improvement between 4 and 52 times could be obtained with respect to an all-software implementation.
机译:嵌入式系统被称为有效的候选者,以有效地支持图像和视频处理算法。它们的高度灵活性,速度性能和低功耗主要是由于其软件和特定硬件部分的联合设计。 Zynq全部可编程系统上,集成了ARM处理器和高性能可编程逻辑资源,现在通常是更加传统的实现平台,例如应用特定的集成电路(ASIC)和数字信号处理器(DSP)。在本文中,我们评估了两种不同的设计策略,每个不同的设计策略都有其自身的基于Zynq的支持平台,向设计人员提供了有用的提示,了解如何识别目标应用程序的最佳设计选择。这里呈现的第一个支持平台还利用嵌入式操作系统(OS);它显着限制了所需的设计努力和上市时间。第二架构在没有操作系统的支持下实现,并且当然比前者更高的性能,但需要更高的开发和验证时间。两个平台都利用了一个用于感兴趣的功能的硬件加速器。作为一个案例研究,通过使用上述平台设计了一种简单但完整的图像处理架构。性能测量表明,对于所有软件实现,可以获得4至52次之间的近似速度改善。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号