Building a scalable and portable message-passing library for embedded multicore systems

Authors:
Shih-Hao Hung;Po-Hsun Chiu;Chi-Sheng Shih
Affiliations:
National Taiwan University, Taipei, Taiwan;National Taiwan University, Taipei, Taiwan;National Taiwan University, Taipei, Taiwan
Venue:
Proceedings of the 2011 ACM Symposium on Research in Applied Computation
Year:
2011

Citing 11
Cited 0

Parallel program design: a foundation

Parallel program design: a foundation
A high-performance, portable implementation of the MPI message passing interface standard

Parallel Computing
OpenMP: An Industry-Standard API for Shared-Memory Programming

IEEE Computational Science & Engineering
Industry Trends: Chip Makers Turn to Multicore Processors

Computer
Design space exploration for multicore architectures: a power/performance/thermal view

Proceedings of the 20th annual international conference on Supercomputing
Implementation and evaluation of shared-memory communication and synchronization operations in MPICH2 using the Nemesis communication subsystem

Parallel Computing
Software Standards for the Multicore Era

IEEE Micro
Cache Hierarchy and Memory Subsystem of the AMD Opteron Processor

IEEE Micro
Designing and Implementing a Portable, Efficient Inter-core Communication Scheme for Embedded Multicore Platforms

RTCSA '10 Proceedings of the 2010 IEEE 16th International Conference on Embedded and Real-Time Computing Systems and Applications
A portable, efficient inter-core communication scheme for embedded multicore platforms

Journal of Systems Architecture: the EUROMICRO Journal
Parallel Architecture Core (PAC)--the First Multicore Application Processor SoC in Taiwan Part I: Hardware Architecture & Software Development Tools

Journal of Signal Processing Systems

Quantified Score

Hi-index	0.00

Visualization

Abstract

Multicore architectures have become increasingly popular for embedded systems, but diversified inter-core communication schemes provided by various vendors haveimpacted the portability of multicore applications on embedded systems and slowed down the software development. In general, parallel applications written with the message-passingparadigm are relatively portable. While MPI programscan run across a variety of systems, from PC, clusters, to supercomputers, the MPI libraryis simply too large for embedded systems. Previously, we designed a library called MSG, which emphasizes on the portability and efficiency of multicore applications on resource-constrained embedded systems. The work described in this paper further reduces the memory usage for many-core systems by incorporating two scalable buffer management schemes. We also present mathematical models and experimental results to evaluate the performance of the library. Our results show that our library is both portable and scalable, as it has been ported to several embedded platforms and outperforms MPICH2, a popular MPI library that has been optimized over the years. The findings also provide some insights on the design of communication architecture for a many-core system.