Given the complexity of parallel programs, developers often must rely on performance analysis tools to help them improve the performance of their code.
HCS Research Laboratory have written a very interesting whitepaper, Parallel PerformanceWizard: A Performance Analysis Tool for Partitioned Global-Address-Space Programming, which includes how one would use PPW in the analysis and optimization of PGAS applications by presenting a small case study with a Quadrics QsNetII cluster using the PPW version 1.0 implementation.
PPW is known to work on many platforms and it has been heavily tested on Linux with the Berkeley UPC compiler and the GCC UPC compiler.
Parallel Performance Wizard (PPW) is a performance analysis tool designed by the University of Florida High-performance Computing and Simulation (HCS) Research Laboratory for UPC and SHMEM programs.
Sharing best practice:
HCS Research Laboratory have written a very interesting whitepaper, Parallel PerformanceWizard: A Performance Analysis Tool for Partitioned Global-Address-Space Programming, which includes how one would use PPW in the analysis and optimization of PGAS applications by presenting a small case study with a Quadrics QsNetII cluster using the PPW version 1.0 implementation.
PPW is known to work on many platforms and it has been heavily tested on Linux with the Berkeley UPC compiler and the GCC UPC compiler.
PPW version 2.0 was released on 01/21/2009.
To download please go to: http://ppw.hcs.ufl.edu/download.html
Image from: "Parallel PerformanceWizard: A Performance Analysis Tool for Partitioned Global-Address-Space Programming", Hung-Hsun Su Billingsley, M. George, A.D. Dept. of Electr. & Comput. Eng., Univ. of Florida, Gainesville, FL, 2008
Posts
No comments:
Post a Comment