The Njal Supercomputing Cluster
Home | Research | People | Pictures | Hardware | Get an Account | FAQ | Documentation
The Njal Cluster
News
February 18th, 2007
Installed 56 Xeon 64bit CPUs, clocking in at 2GHz each
September 8th, 2006
njal:~ > uptime
12:33:46 up 391 days, 21:43...

August 8th, 2006
Acquired an additional 52 CPU cluster, based on 26 Intel Dual Core Xeon Processors.
February 18th, 2006
Constructed an AMD Opteron dual-core machine; our new webserver.
September 7th, 2005
7 compute-nodes have been repaired.
April 6th, 2005
Recovered from air conditioning failure. Replaced SCSI controler on file server. Upgraded memory in master node.
February 28th, 2005
Memory upgrade to all compute-nodes and file servers completed!
February 14th, 2005
SB power glitch; Njal is back online.
January 12th, 2005
System updated and installed memory upgrades on nodes a1-a20.
November 16th, 2004
njal:~ > uptime
12:17:45 up 365 days, 15:50...

November 9th, 2004
Check out the Ganglia web frontend!
May 10th, 2004
Kernel update to all compute-nodes.
April 07th, 2004
Restored nodes a6 and a27. Completed system updates.
March 12th, 2004
All systems are fine. Installed Intel compilers in Feb.
January 1st, 2004
New system is running very well; added fftw.
December 1st, 2003
System has been repaired and upgraded.
November 12th, 2003
System is down until further notice.
October 6th, 2003
The new PDU arrived and a1-a22 are back up.
August 27th, 2003
PBS has been fixed.
August 20th, 2003
HP ProCurve has been replaced. Njal is back up with reduced capability until the replacement PDU arrives. PBS is also malfunctioning.
August 14-15th, 2003
Power failure burns out a PDU (Power Distribution Unit) and the HP ProCurve switch. Knocked offline are 22 nodes.
July 19th, 2003
Libgoto - a high-performance BLAS by Kazushige Goto - has been installed.
June 1st, 2003
The library problem with mpiCC on nodes b1 and b2 has been fixed.
May 8th, 2003
We have replaced the HP Pro Curve. Njal is operational again.
May 1st, 2003
The HP Pro Curve Switch has failed; we are waiting for a replacement from Racksaver.
February 23th, 2003
System overhaul completed: Open PBS v2.3, MPICH v1.2.5, LAM/MPI v6.5.9, and Mpiexec.
February 5th, 2003
GSL (GNU Scientific Library) is now available.
January 24th, 2003
CVS is now available on Njal.
October 1st, 2002
The new Njal web site is up.
September 10th, 2002
Michael Martino has joined the system administration team.
July 22nd, 2002
The new storage servers are now on-line.
June 26th, 2002
The official Njal web site is up.
June 1st, 2002
Joseph Spadavecchia is now the system administrator of Njal.
May 12th, 2002
Upgrade is mostly complete, new storage servers are still off-line.


Last updated: 2/21/2006

About Njal
Njal is a 160-processor HPC cluster installed in Professor Konstantin Likharev's Nanoelectronics Device Reseach Group at the Department of Physics, Stony Brook University, NY.

The system was installed in the late 2001 using a grant from DoD's Defense University Research Instrumentation Program (DURIP), obtained via the AFOSR Quantum Electronic Solids program (Dr. H. Weinstock).

Njal was developed by Joseph Spadavecchia and Michael Martino, who have been managing the system since 2002. Currently, the system is based on a combination of Gentoo Linux and Xen paravirtualization.

Critical help in the acquisition, installation, and upgrading of the original cluster configuration was provided by Simon Folling, Drew Youngreman and Alex Korobka. This old configuration was retired in 2003.

In 2006 additional funding for a major hardware update was provided by the Center on Functional Engineered Nano Architectonics (FENA).

In 2008 the direction of Njal was handed over to Prof. Marivi Fernandez-Serra.

Performance
12/48 N/P Xeon 100Mbit HPL Linpack: 75.5 GFLOPS
40/80 N/P PIII Myrinet HPL Linpack: 47.8 GFLOPS
80/160 N/P PIII 100Mbit ScaLAPACK: 73.3 GFLOPS
Copyright (C) 2002-2007 Konstantin Likharev and Joseph Spadavecchia.