At a Glance
- Infiniband, shared memory, GbE, 10 GbE, and Myrinet MX communication support
- Sun Grid Engine plug-in
- Supports third-party parallel debuggers Totalview and Allinea DDT
- Integration with Sun Studio Analyzer
What's New
- Process affinity options, providing improved performance for many HPC applications
- Additional compiler support -- including Sun Studio, gnu, PGI, Intel, Pathscale
- Suspend / resume support
- Improved intra-node shared memory performance and scalability
- Infiniband QDR support
- Automatic path migration support
- Relocatable installation
Support
For more information about purchasing support for Sun HPC ClusterTools, visit our
Sun HPC ClusterTools Support Services page. Need Help Now?
Access SunSolve for technical support resources, documentation, software updates and other services. If your product is under support coverage with Sun, be sure to login to gain full access to premium content.
|
Spotlight
Sun HPC ClusterTools 8.2.1 is based on the Open MPI 1.3.4 release.
Parallel Development
Sun HPC ClusterTools is a high-performance implementation of the Message
Passing Interface (MPI) standard for Sun x86 and SPARC based systems. HPC ClusterTools supports multiple interconnect fabrics including Ethernet, Gigabit Ethernet, Myrinet, and InfiniBand, providing high-bandwidth, low-latency communication for MPI processes. HPC ClusterTools takes full advantage of performance features in
InfiniBand, including support for IB QDR, multi-rail IB, and Mellanox ConnectX. It utilizes shared-memory communication between processes within a node to achieve optimal performance.
HPC ClusterTools is based on Open MPI and is a full MPI-2 implementation, including MPI I/O and one-sided communication. The Open Run-Time Environment (ORTE) in HPC CusterTools provides a basic set of parallel job management facilities, and includes plug-in modules to support:
Sun Grid Engine - allows parallel jobs to be launched and maximizes utilization of shared resources. Portable Batch System (PBS) - a job scheduler that allocates network resources to batch jobs on networked, multi-platform environments.
|
