NCSA Creates World’s Largest LSF-based Parallel Windows NT Cluster
New Windows NT Supercluster Makes High-Performance
Computing More Affordable
Toronto, Canada August 11, 1998 The National Center for
Supercomputing Applications (NCSA) at the University of Illinois at
Urbana-Champaign has used Platform’s LSF Suite to create the world’s
largest LSF-based parallel Windows NT cluster, delivering High
Performance Computing (HPC) from commodity Intel-based PCs and
workstations.
NCSA and researchers with the Alliance Parallel
Computing Team constructed the Window NT “supercluster” from commercial
PCs as part of its research into commodity HPC. NCSA, the leading-edge
site for the National Computational Science Alliance, completed
construction of a 256-processor Windows NT supercluster consisting of 128
dual-processor Pentium II workstations. That supercluster will be
upgraded to 512 processors within the next year.
"LSF is a crucial component of our NT
supercluster," said Larry Smarr, director of NCSA and the Alliance.
"It is our queuing and scheduling system for jobs on the
supercluster, and we are pushing its use to new heights running
distributed jobs. "
LSF provides distributed load sharing and job scheduling for
heterogeneous UNIX and Windows NT computing environments. LSF enables
users to take full advantage of the combined resources of the entire
supercluster for parallel applications that can scale up to the full
cluster. LSF has clearly demonstrated its usefulness in a large-scale NT
environment with parallel applications, accelerating the migration of
mission-critical applications from UNIX clusters to heterogeneous UNIX
and Windows NT environments.
The supercluster supports the Message Passing
Interface (MPI) enabling parallel production applications to be easily
integrated with the LSF Suite. MPI is an industry standard that has been
used successfully for astrophysics, environmental hydrology, finite
element analysis and numerical relativity applications that have
previously been run on NCSA's SGI/Cray Origin2000 or Hewlett-Packard
SPP-2000 systems.
The supercluster uses software called High Performance Virtual Machine
(HPVM), which integrates clusters of Windows NT processors into a
high-performance environment. HPVM, developed by Andrew Chien, a
professor in the University of Illinois Department of Computer Science
and a member of the Parallel Computing Team, enables each node of an NT
cluster to communicate at a bandwidth of just under 80 megabytes per
second and at a latency of less than 11 microseconds. As a result, it
will offer users in the national scientific community a low-cost
alternative to conventional high-performance machines that are typically
used to carry out high-end computational research.
"We didn't see any point in duplicating the
effort of a successful commercial product," said Chien. "With a
modest effort, we integrated LSF with our HPVM software, and it provides
full monitoring and scheduling functions for our cluster."
This new initiative further extends Platform’s LSF
presence at NSCA as it gains widespread acceptance and provides the
potential for cross-platform coordination. LSF is also being used for job
scheduling on NCSA’s Origin2000 and on its SGI POWER CHALLENGEarray
supercomputers.
“Platform Computing is proud to be a part of the
groundbreaking work that is going on NCSA,” said Dave Black, President
and CEO of Platform Computing. “High-performance Windows NT clusters are
ushering in a new era of affordable high performance computing, which is
being enabled by our LSF Suite of tools for Workload Management.
Platform is excited to provide the NT environment the first truly
reliable, scalable, and fault tolerant solution for high-performance
computing.”
About NCSA
The National Center for Supercomputing
Applications is the leading-edge site for the National Computational
Science Alliance. NCSA is a leader in the development and
deployment of cutting-edge high-performance computing, networking, and
information technologies. NCSA is funded by National Science
Foundation, the state of Illinois, the University of Illinois, industrial
partners, and other federal agencies.
The National Computational Science Alliance is a partnership to prototype
an advanced computational infrastructure for the 21st Century and
includes more than 50 academic, government and industry research partners
from across the United States. The Alliance receives core funding
from the National Science Foundation and cost-sharing at partner
institutions.
About Platform
Platform Computing Corporation is the world leader in
workload management. Founded in 1992 and headquartered in Toronto,
Canada, with offices in North America, Europe, and Asia Pacific,
Platform’s LSF Suite enables enterprise customers to manage and support
business-critical workloads in distributed heterogeneous computing
environments.
The LSF Suite is used in more than 60,000 computers in more than 600
major corporations worldwide. It has been embraced as the industry
standard for workload management by major system vendors such as Sun, HP,
Compaq, Silicon Graphics, Digital, Fujitsu, Hitachi, NEC, and Sony. For
more information visit Platform’s web site at
www.platform.com.