ScholarLens Back to search
2011•Jisuanji kexue yu tansuoRequires access

Efficient OpenMP Extension for Embedded Multicore Platform

Tao Liu

Open publisher page 0 citations

Abstract

Developing an efficient programming method for multicore platforms has been a major objective in parallel software research. This paper proposes an efficient implementation of parallel programming OpenMP on embedded multicore platform. For embedded multicore platform, there are limited memory resources. Incorporating limited memory hierarchy into OpenMP is challenging. This paper presents the solution to extend OpenMP custom directive tiling to improve performance. OpenMP with supporting loop tiling is proposed to utilize memory hierarchy. Preliminary performance evaluations show that about 10% performance improvement is gained from the OpenMP implementation on the target embedded multicore platform.

About this research paper

What this paper is about

Developing an efficient programming method for multicore platforms has been a major objective in parallel software research. This paper proposes an efficient implementation of parallel programming OpenMP on embedded multicore platform. For embedded multicore platform, there are limited memory resources. Incorporating limited memory hierarchy into OpenMP is challenging. This paper presents the solution to extend OpenMP custom directive tiling to improve performance. OpenMP with supporting loop tiling is proposed to utilize memory hierarchy. Preliminary performance evaluations show that about 10% performance improvement is gained from the OpenMP implementation on the target embedded multicore platform.

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

Developing an efficient programming method for multicore platforms has been a major objective in parallel software research. This paper proposes an efficient implementation of parallel programming OpenMP on embedded multicore platform. For embedded multicore platform, there are limited memory resources. Incorporating limited memory hierarchy into OpenMP is challenging. This paper presents the solution to extend OpenMP custom directive tiling to improve performance. OpenMP with supporting loop tiling is proposed to utilize memory hierarchy. Preliminary performance evaluations show that about 10% performance improvement is gained from the OpenMP implementation on the target embedded multicore platform.

Key concepts: Computer science, Multi-core processor, Memory hierarchy, Parallel computing, Computer architecture, Software, Programming paradigm, Shared memory

Related papers

Related paper 1Related paper 2Related paper 3
Back to paper searchBrowse research topicsOriginal source
Efficient OpenMP Extension for Embedded Multicore Platform — Research Paper | ScholarLens