论文信息 - Scaling PyCSP

Scaling PyCSP

PyCSP is intended to help scientists develop correct, maintainable and portable code for emerging architectures. The library uses concepts from Communicating Sequential Processes (CSP) and is implemented in the Python programming language. This paper introduces a new channel implementation and new process types for PyCSP that are intended to simplify writing programs for clusters. The new processes and channels are investigated by running three benchmarks on a cluster, using up to 512 CPU cores. The results show that PyCSP scales well, and that the introduction of a mobile channel home improves the performance of all three benchmarks.

[1] Brian Vinter,et al. Cluster Computing and JCSP Networking , 2002 .

[2] Peter H. Welch,et al. Communicating Mobile Processes : Introducing occam-pi , 2005 .

[3] C. A. R. Hoare,et al. Communicating sequential processes , 1978, CACM.

[4] Brian Vinter. Next Generation Processors , 2009 .

[5] Brian Vinter,et al. Three Unique Implementations of Processes for PyCSP , 2009, CPA.

[6] Adam T. Sampson,et al. pony - The occam-pi Network Environment , 2006, CPA.

[7] Neil Brown,et al. C++CSP2: A Many-to-Many Threading Model for Multicore Architectures , 2007, CPA.

[8] Thorsten Meinl,et al. KNIME: The Konstanz Information Miner , 2007, GfKl.

[9] Neil Brown,et al. Integrating and Extending JCSP , 2007 .

[10] Brian Vinter,et al. Verification of a Dynamic Channel Model using the SPIN Model Checker , 2011, CPA.

[11] Brian Vinter,et al. PGAS for Distributed Numerical Python Targeting Multi-core Clusters , 2012, 2012 IEEE 26th International Parallel and Distributed Processing Symposium.

[12] Tatu Ylönen,et al. The Secure Shell (SSH) Protocol Architecture , 2006, RFC.

[13] Brian Vinter,et al. PyCSP Revisited , 2009, CPA.

[14] Katherine A. Yelick,et al. Languages for High-Productivity Computing: the DARPA HPCS Language Project , 2007, Parallel Process. Lett..

[15] Katherine Yelick,et al. UPC Language Specifications V1.1.1 , 2003 .