Home

About

Advanced Search

Browse by Discipline

Scientific Societies

E-print Alerts

Add E-prints

E-print Network
FAQHELPSITE MAPCONTACT US


  Advanced Search  

 
Computer Science Research Group Announces Compiler Release at Supercomputing 2011 in Seattle The High Performance Computing Tools (HPCTools) research group, part of the Department of
 

Summary: Computer Science Research Group Announces Compiler Release at Supercomputing 2011 in Seattle
The High Performance Computing Tools (HPCTools) research group, part of the Department of
Computer Science at the University of Houston, has released a new version of the OpenUH compiler
during the recent Supercomputing Conference in Seattle, Washington. OpenUH is a branch of the open
source Open64 compiler suite for C, C++, and Fortran 95/2003 developed by the HPCTools group.
Key features of the new release of OpenUH include full support for the OpenMP 3.0 industry standard as
well as support for Coarray Fortran. OpenMP is a specification for a set of compiler directives, runtime
library routines, and environment variables that can be used to specify shared memory parallelism in
Fortran and C/C++ programs. It is supported by a broad range of compiler vendors including Intel,
Oracle, IBM, and Microsoft (www.openmp.org). OpenMP 3.0 introduced tasking features that now allow
the parallelization of applications exhibiting irregular parallelism. The support for the tasking features in
OpenUH was implemented in collaboration with researchers from Tsinghua University in China. Coarray
Fortran is a recent extension of the Fortran standard, adopted as part of Fortran 2008, used for parallel
processing both on shared memory systems and clusters. The OpenUH Coarray Fortran implementation
is one of the first open source implementations.
In addition to OpenMP tasking and Coarray Fortran support, the new release includes support for the
DARWIN framework and a new version of the Dragon tool, both also developed by the HPCTools group.
DARWIN is a dynamic optimization framework that collects information about a running OpenMP
program. It includes components for tracking and monitoring performance events, and utilities for
analyzing the collected data. Dragon enables users to get a better understanding of their source code by

  

Source: Azevedo, Ricardo - Department of Biology and Biochemistry, University of Houston

 

Collections: Biology and Medicine; Environmental Sciences and Ecology