2006•Kluwer Academic Publishers eBooksOpen access

Asynchronous Integration of Coarse-Grained Reconfigurable XPP-Arrays Into Pipelined Risc Processor Datapath

Jürgen Becker, Alexander J. Thomas, Maik Scheer

Open full text 0 citations

Abstract

Nowadays, the datapaths of modern microprocessors reach their limits by using static instruction sets. A way out of these limitations is a dynamic reconfigurable processor datapath extension achieved by integrating traditional static datapaths with the coarse-grain dynamic reconfigurable XPP-architecture (eXtreme Processing Platform). Therefore, a loosely asynchronous coupling mechanism of the corresponding datapath units has been developed and integrated onto a CMOS 0.13 μm standard cell technology from UMC. Here the SPARC compatible LEON processor is used, whereas its static pipelined instruction datapath has been extended to be configured and personalized for specific applications. This allows a various and efficient use, e.g. in streaming application domains like MPEG-4, digital filters, mobile communication modulation, etc. The chosen coupling technique allows asynchronous concurrency of the additionally configured compound instructions, which are integrated into the programming and compilation environment of the LEON processor.

Open-access reader

About this research paper

What this paper is about

Nowadays, the datapaths of modern microprocessors reach their limits by using static instruction sets. A way out of these limitations is a dynamic reconfigurable processor datapath extension achieved by integrating traditional static datapaths with the coarse-grain dynamic reconfigurable XPP-architecture (eXtreme Processing Platform). Therefore, a loosely asynchronous coupling mechanism of the corresponding datapath units has been developed and integrated onto a CMOS 0.13 μm standard cell technology from UMC. Here the SPARC compatible LEON processor is used, whereas its static pipelined instruction datapath has been extended to be configured and personalized for specific applications. This allows a various and efficient use, e.g. in streaming application domains like MPEG-4, digital filters, mobile communication modulation, etc. The chosen coupling technique allows asynchronous concurrency of the additionally configured compound instructions, which are integrated into the programming and compilation environment of the LEON processor.

Why it matters

A significance statement is not available in the OpenAlex record.

Key contribution

A contribution statement is not available in the OpenAlex record.

Method / approach

Method details are not available in the OpenAlex metadata.

Main findings

Findings are not separately available in the OpenAlex metadata.

Limitations

Limitations are not available in the OpenAlex metadata.

Applications

Application details are not available in the OpenAlex metadata.

Available abstract

Nowadays, the datapaths of modern microprocessors reach their limits by using static instruction sets. A way out of these limitations is a dynamic reconfigurable processor datapath extension achieved by integrating traditional static datapaths with the coarse-grain dynamic reconfigurable XPP-architecture (eXtreme Processing Platform). Therefore, a loosely asynchronous coupling mechanism of the corresponding datapath units has been developed and integrated onto a CMOS 0.13 μm standard cell technology from UMC. Here the SPARC compatible LEON processor is used, whereas its static pipelined instruction datapath has been extended to be configured and personalized for specific applications. This allows a various and efficient use, e.g. in streaming application domains like MPEG-4, digital filters, mobile communication modulation, etc. The chosen coupling technique allows asynchronous concurrency of the additionally configured compound instructions, which are integrated into the programming and compilation environment of the LEON processor.

Key concepts: Datapath, Computer science, Asynchronous communication, Computer architecture, Reduced instruction set computing, Embedded system, Parallel computing, Instruction set

Related papers

Back to paper searchBrowse research topicsOriginal source
Asynchronous Integration of Coarse-Grained Reconfigurable XPP-Arrays Into Pipelined Risc Processor Datapath — Research Paper | ScholarLens