首页> 外文期刊>IEEE transactions on very large scale integration (VLSI) systems >Reconfigurable pipelined 2-D convolvers for fast digital signalprocessing
【24h】

Reconfigurable pipelined 2-D convolvers for fast digital signalprocessing

机译:可重新配置的流水线二维卷积器,用于快速数字信号处理

获取原文
获取原文并翻译 | 示例

摘要

In order to make software applications simpler to write and easiernto maintain, a software digital signal-processing library that performsnessential signal- and image-processing functions is an important part ofnevery digital signal processor (DSP) developer's toolset. In general,nsuch a library provides high-level interface and mechanisms, therefore,ndevelopers only need to know how to use algorithms, not the details ofnhow they work. Complex signal transformations then become functionncalls, e.g., C-callable functions. Considering the two-dimensional (2-D)nconvolver function as an example of great significance for DSP's, thisnpaper proposes to replace this software function by an emulation on anfield-programmable gate array (FPGA) initially configured by softwarenprogramming. Therefore, the exploration of the 2-D convolver's designnspace will provide guidelines for the development of a library ofnDSP-oriented hardware configurations intended to significantly speed upnthe performance of general DSP processors. Based on the specificnconvolver, and considering operators supported in the library asnhardware accelerators, a series of tradeoffs for efficiently exploitingnthe bandwidth between the general-purpose DSP and accelerators arenproposed. In terms of implementation, this paper explores thenperformance and architectural tradeoffs involved in the design of annFPGA-based 2-D convolution coprocessor for the TMS320C40 DSPnmicroprocessor available from Texas Instruments Incorporated. However,nthe proposed concept is not limited to a particular processor
机译:为了使软件应用程序更易于编写和维护,一个具有数字信号处理和图像处理功能的软件数字信号处理库是每个数字信号处理器(DSP)开发人员工具集的重要组成部分。通常,这样的库提供高级接口和机制,因此,开发人员只需要知道如何使用算法,而不是他们如何工作的细节。然后,复杂的信号转换成为函数调用,例如C调用函数。考虑到二维(2-D)n卷积器功能对于DSP具有重要意义的示例,本文建议通过对最初由软件编程配置的现场可编程门阵列(FPGA)进行仿真来替代此软件功能。因此,对二维卷积器设计空间的探索将为开发面向nDSP的硬件配置库提供指导,以显着加快通用DSP处理器的性能。基于特定的卷积器,并考虑到库中作为硬件加速器支持的运算符,提出了一系列折衷方案,以有效利用通用DSP和加速器之间的带宽。在实现方面,本文探讨了性能和架构上的折衷,这些折衷涉及可用于德州仪器(TI)的TMS320C40 DSPn微处理器的基于anFPGA的二维卷积协处理器的设计。然而,提出的概念不限于特定的处理器

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