Scalasca

Scalasca is an open-source toolset that can be used to analyze the performance behavior of parallel applications and to identify opportunities for optimization. It has been specifically designed for use on large-scale systems including IBM Blue Gene (such as JUGENE at Forschungszentrum Jülich) and Cray XT, but is also well-suited for small- and medium-scale HPC platforms.

Scalasca supports an incremental performance-analysis procedure that integrates runtime summaries with in-depth studies of concurrent behavior via event tracing, adopting a strategy of successively refined measurement configurations. A distinctive feature is the ability to identify wait states that occur, for example, as a result of unevenly distributed workloads. Especially when trying to scale communication-intensive applications to large processor counts, such wait states can present severe challenges to achieving good performance.

Scalasca is a software available under the new BSD open-source license.

Supported Platforms

The Scalasca tool set has been successfully tested on the following platforms:

 * IBM Blue Gene/P
 * IBM SP & BladeCenter clusters
 * Cray XT5
 * SGI Altix
 * NEC SX-8
 * SiCortex systems
 * various Linux/Intel (x86/x64) clusters

The following platforms have not been tested recently:

 * IBM Blue Gene/L
 * Cray XT3/4
 * Sun Solaris/SPARC-based clusters
 * other NEC-SX systems

However, the supplied makefile definition files might still work on these systems.

Since the code is fairly portable, it is also likely to work on other systems as well with modest effort by someone experienced with the intricacies of MPI libraries and compilers. The supplied Makefile definition files may provide a good basis for building and testing the toolset on other platforms.

What is new in this release:

• This is a bugfix release.

What is new in version 1.4.1:

• Several critical bugfixes and numerous enhancements.

What is new in version 1.3.3:

• Added measurement control API
• Added support for Cray XE6
• Improved support for NEC SX systems
• Resolved trace collection issues on SX-9
• Added support for FTRACE API
• Improved trace analysis support for MPMD codes
• Fixed incorrect handling of MPI_IN_PLACE
• Fixed incomplete tracking of MPI inter-communicators
• All CUBE 3.3.2 improvements

What is new in version 1.3.1:

• Fixed Fortran MPI I/O wrappers for SGI MPT MPI library
• Changed default configuration on Cray XT to explicitly disable MPI I/O wrappers (presumably broken)
• Fixed configuration of AIX POE MPI library
• Fixed configuration when bash >= 4.0 or ksh is used as /bin/sh
• Fixed message statistics for MPI_Sendrecv
• Fixed PDT instrumentation issues with OpenMP codes
• Various minor usability improvements

Requirements:

• Qt