BANGALORE
NOVEMBER 7, 2006
HP announced the availability of new enhancements to its high-performance
computing portfolio.
The HP Unified Cluster Portfolio now includes a certified original equipment
manufacturer version of Microsoft Windows Compute Cluster Server (CCS) 2003 and
a new HP Cluster Management Utility option. Additionally available is HP Message
Passing Interface on Windows CCS.
The enhancements are part of HP's strategy to deliver industry-standard
high-performance computing (HPC) solutions to customers, HP said in a statement.
The certified OEM version of Windows CCS is fully integrated and supports HP
ProLiant servers and HP BladeSystem solutions. It comes installed and tested on
HP Cluster Platform and Cluster Platform Express configurations.
Windows CCS is designed specifically to make HPC clusters easier to deploy,
manage and integrate with existing infrastructure and tools, allowing scientists
and engineers to focus on science rather than IT. HP enhancements to CCS include
custom installation scripts and documentation that greatly streamline
deployment.
HP Message Passing Interface (HP-MPI), an implementation of the Messaging
Passing Interface standard and critical for running and porting parallel
applications on clustered systems, enables rapid independent software vendor
(ISV) application support for Windows CCS. More than 20 ISVs now support HP-MPI
for Windows, HP-UX and Linux.
HP-MPI supports multi-protocol execution of MPI applications on clusters and
provides a transparent implementation supporting a broad range of switch
interconnects and processor and operating system architectures.
The HP Cluster Management Utility (CMU) is a low-cost, efficient and robust
tool for the management of HPC clusters and compute farms. A simple graphic
interface enables centralized console management, monitoring of cluster-wide and
node-specific metrics, and software installation. CMU enables rapid provisioning
and management of multiple images, making it ideal for a site with frequent
changes to the software configuration across the cluster or on a subset of the
cluster.