Center for Piezoelectrics by Design web page

The CPD Opteron cluster

Description of the cluster

The configuration of the Opteron cluster is shown in Fig. 1. It consists of 114 compute nodes connected by Infiniband and gigabit networking. All machines run Red Hat Enterprise Linux 4 and Platform OCS (a vendor supported "Rocks"-like distribution) All compute nodes are are accessed from the front end. A RAID array is directly attached to the front end and contains the home directories and 2 globally accessible scratch spaces of 1 TB each. Also, there is a set of 3 IO nodes which serve as a 1.5 TB parallel file system and is accessible from the nodes and the front end via the infiniband network.

Diagram of the CPD Opteron cluster

 

 

Technical details

The following table lists the details of each machine in the cluster:

Table 1: Front end and compute node details.
# Node name architecture processor memory local disk
1 Dell PowerEdge 2970 pzt.wm.edu Opteron 2220 / 2.8 GHz 8 GB 380 GB
98 Dell PowerEdge SC1435 c1-[1-28],c2-[29-57],c4-[58-86],c5-[87-102] Opteron 2218 / 2.6 GHz 8 GB 120 GB
4 Dell PowerEdge SC1435 c5-[103-106] Opteron 2222 SE / 3.0 GHz 32 GB 420 GB
8 Dell PowerEdge SC1435 c5-[107-114] Opteron 2222 SE / 3.0 GHz 16 GB 420 GB

 

Table 2 shows the various filesystems that users may access within the cluster. The /home, /scr1 and /scr2 filespaces are all mounted via NFS over the gigabit ethernet network. Currently a 10GB quota is set for all users /home directory space. /scr1 and /scr2 are to be used as global scratch space. The global scratch space is to be used for temporarily storing large files. Neither filesystem is backed up and purging is not done automatically. Users with large amounts of data (>100 GB) will be made to reduce their usage when the scratch space starts to get too full (> ~70%).

 

Table 2: Filesystems on the Opteron cluster
name size notes
/home/$USER 1 TB users get a 10 GB quota
/scr1/$USER 1 TB Global NFS scratch 1
/scr2/$USER 1 TB Global NFS scratch 2
/pvfs 1.5 TB Global parallel scratch
/lscr 120 GB / 420 GB Local scratch on compute node

 

 

Installed software

The following lists the compilers, libraries, mpi environments and computational packages that are currently installed on the cluster. Most packages have an environment module in place for users to load. Please send email to Eric J. Walter if you need additional software installed.

Compilers

Command
Description Version(s) Module C C++ Fortran
Portland Group / PGI 7.1, 7.1-3, 7.2-3 pgi/7.1, pgi/7.1-3, pgi/7.2, pgi/7.2-3 pgcc pgCC pgf90/pgf95/pgf77
Pathscale 3.0 pathscale pathcc pathCC pathf90/pathf95/pathf77
Intel 9.1.036, 10.1.012, 10.1.015 intel/9.1, intel/10.1, intel/10.1v2 icc icpc ifort
Absoft 10.1 absoft NA NA f90/f95/f77

 

 

 

Libraries

Name Location
AMD Core Math Library /share/apps/acml4.0.1
Hierarchical Data Format /share/apps/absoft10.1/extras/hdf(4,5)
fftw v2.1.5 /share/apps/fftw2
fftw v3 /share/apps/fftw3
cmkl - Intel's Cluster Math Kernel Library /share/apps/intel/cmkl/10.0.1.014,/share/apps/intel/cmkl/10.0.3.020
ATLAS /share/apps/absoft10.1/extras/ATLAS

 

 

 

MPI environments (that use Infiniband)

Name Version Module
openmpi 1.2.4,1.2.5,1.2.8(compiled using intel v10.1.015) openmpi-1.2.4, openmpi-1.2.5, openmpi-1.2.8_intel
mvapich2 1.0.1, 1.0.2p2 mvapich2-1.0.1, mvapich2-1.0.2
mvapich1 1.0p1 (compiled for pgi and intel) mvapich-1.0-intel, mvapich-1.0-pgi

 

 

 

Computational software

Name Version Module Notes
ABINIT 5.4.4, 5.4.4p, 5.5.4, 5.6.4 abinit/abinit-544_mvapich2, espresso/espresso-402_openmpi125, abinit/abinit-544_openmpi125, abinit/abinit-544p_openmpi125, abinit/abinit-554_openmpi125, abinit/abinit-564_openmpi125 The kpt/spin parallel version is named abinip, the band/fft parallel version is named abinip_bfft for versions earlier than 5.6.4
Quantum Espresso CVS from 02/27/08, v3.2.3, v4.0.2 espresso/espresso-CVS_openmpi125, espresso/espresso-323_mvapich2, espresso/espresso-402_openmpi125 Parallel versions of the codes d3.x, phcg.x, ph.x and pw.x are named p_d3.x, p_phcg.x, p_ph.x, p_pwcond.x, p_pw.x
Dacapo SVN from 03/31/08 dacapo/dacapo_SVN033108
Siesta v2.0.1 siesta/siesta-2.0.1

 

 

 

©2009 The College of William and Mary

Installed software

The following lists the compilers, libraries, mpi environments and computational packages that are currently installed on the cluster. Most packages have an environment module in place for users to load. Please send email to Eric J. Walter if you need additional software installed.

Compilers

Command
Description Version(s) Module C C++ Fortran
Portland Group / PGI 7.1, 7.1-3, 7.2-3 pgi/7.1, pgi/7.1-3, pgi/7.2, pgi/7.2-3 pgcc pgCC pgf90/pgf95/pgf77
Pathscale 3.0 pathscale pathcc pathCC pathf90/pathf95/pathf77
Intel 9.1.036, 10.1.012, 10.1.015 intel/9.1, intel/10.1, intel/10.1v2 icc icpc ifort
Absoft 10.1 absoft NA NA f90/f95/f77

 

 

 

Libraries

Name Location
AMD Core Math Library /share/apps/acml4.0.1
Hierarchical Data Format /share/apps/absoft10.1/extras/hdf(4,5)
fftw v2.1.5 /share/apps/fftw2
fftw v3 /share/apps/fftw3
cmkl - Intel's Cluster Math Kernel Library /share/apps/intel/cmkl/10.0.1.014,/share/apps/intel/cmkl/10.0.3.020
ATLAS /share/apps/absoft10.1/extras/ATLAS

 

 

 

MPI environments (that use Infiniband)

Name Version Module
openmpi 1.2.4,1.2.5,1.2.8(compiled using intel v10.1.015) openmpi-1.2.4, openmpi-1.2.5, openmpi-1.2.8_intel
mvapich2 1.0.1, 1.0.2p2 mvapich2-1.0.1, mvapich2-1.0.2
mvapich1 1.0p1 (compiled for pgi and intel) mvapich-1.0-intel, mvapich-1.0-pgi

 

 

 

Computational software

Name Version Module Notes
ABINIT 5.4.4, 5.4.4p, 5.5.4, 5.6.4 abinit/abinit-544_mvapich2, espresso/espresso-402_openmpi125, abinit/abinit-544_openmpi125, abinit/abinit-544p_openmpi125, abinit/abinit-554_openmpi125, abinit/abinit-564_openmpi125 The kpt/spin parallel version is named abinip, the band/fft parallel version is named abinip_bfft for versions earlier than 5.6.4
Quantum Espresso CVS from 02/27/08, v3.2.3, v4.0.2 espresso/espresso-CVS_openmpi125, espresso/espresso-323_mvapich2, espresso/espresso-402_openmpi125 Parallel versions of the codes d3.x, phcg.x, ph.x and pw.x are named p_d3.x, p_phcg.x, p_ph.x, p_pwcond.x, p_pw.x
Dacapo SVN from 03/31/08 dacapo/dacapo_SVN033108
Siesta v2.0.1 siesta/siesta-2.0.1

 

 

 



©2009 The College of William and Mary