HPC2002 Result: Dell PowerEdge 1750 cluster (tested by Purdue University)

Benchmark	Reference Time	Runtime	Ratio
351.seis_m	86400	534	162
SPECseisM2002	162

Benchmark

Reference
Time

Runtime

Ratio

351.seis_m

86400

534

162

SPECseisM2002

162

Hardware

Hardware Vendor:

Dell

Model Name:

PowerEdge 1750 cluster

CPU:

Intel Pentium4 Xeon

CPU MHz:

3200

FPU:

Integrated

CPU(s) enabled:

16 cores, 16 chips, 1 core/chip (Hyper-Threading Technology enabled)

CPU(s) orderable:

1 or 2 per node

Primary Cache:

12KB (I) micro-ops (trace) + 8KB (D) on chip

Secondary Cache:

512KB on chip

L3 Cache:

1 MB on chip

Other Cache:

None

Memory:

2 GB DDR PC2100 CL2.5 ECC Registered per node

Disk Subsystem:

1x36 GB SCSI per node

Other Hardware:

see notes

Software

Parallel:

MPI

Processes-Threads:

MPI Processes:

OpenMP Threads:

Operating System:

RedHat Enterprise Linux, Advanced Server version 3 (4)

Compiler:

Intel C++ Compiler- icc, Version 8.1
Build 20050309Z for Linux
Intel Fortran Compiler- ifort, Version 8.1
Build 20050309Z for Linux

File System:

NFS shared file system

System State:

Multi-user

Other Software:

mpich-vapi (see notes for configuration)

Notes / Tuning Information
Tested by Purdue University Approved Alternate Source: approved alternate source, "purdueintel", was used to fix Fortran standard violations Flags (Fortran & C): CPP Flags: -I. -C -P -traditional -Dmpi -DSPEC_HPG_MPI COPTIMIZE = -O3 -static -march=pentium4 -mcpu=pentium4 -Dmpi \ -DSPEC_HPG_MPI -DSPECDONOTNEEDARG -DFORTRAN_UNDERSCORE FOPTIMIZE = -O3 -mp -static -fp_port -xW -axW -tpp7 LDOPTIONS = -O3 -mp -Bstatic Submit command to run MPI application: PBS Version: PBSPro 5.4.1.41640 PBS Command to get resources (for cyclic allocation of processes): qsub -I -q preemptx -lnodes=8:IB:ppn=2,walltime=1:00:00 use_submit_for_speed=1 submit=mpirun -np 16 -hostfile \$PBS_NODEFILE $command Cluster config: Nodes and file server use NFS shared file system Two CPUs per node, Hyper-Threading Technology enabled File server: 2 x 3.06 GHz Intel Xeon processors 4 GB DDR PC2100 CL2.5 ECC Registered Memory 5 x 72 GB 10K RPM SCSI Drives Hardware RAID-5 (Dell PERC/3Di option) Debian Linux, 3.1 "sarge" ext3 local file system NFS shared file system Network (for computation and file server): Infiniband: Topspin HBAs, Topspin 120 switches MPICH-1.2.5 Configuration CC=/opt/intel_cc_81/bin/icc CXX=/opt/intel_cc_81/bin/icc FC=/opt/intel_fc_81/bin/ifort F90=/opt/intel_fc_81/bin/ifort ./configure --with-device=ch_vapi --with-arch=LINUX --without-mpe \ --without-romio --disable-sharedlib \ -cflags="-D_SMP_ -DUSE_STDARG -DLAZY_MEM_UNREGISTER -DUSE_INLINE \ -DRDMA_FAST_PATH -DEARLY_SEND_COMPLETION -D_REENTRANT -O2 \ -DVIADEV_RPUT_SUPPORT -MD -fpic -DMT_LITTLE_ENDIAN -D__LINUX__ \ -DVAPI -DTS_HOST_DRIVER -DCPU_X86 \ -I/usr/local/topspin/include -I/usr/local/topspin/include/vapi \ -L/usr/local/topspin/lib " \ -lib="-lvapi -lmtl_common -lmpga -lmosal -lpthread" All BIOS parameters left with factory defaults. For a description of Intel compiler flags,portability flags, and system parameters used to generate this result,please refer to PURDUE-20050329-INTEL-LINUX-XEON.txt in the flags directory

Notes / Tuning Information

Tested by Purdue University
Approved Alternate Source: approved alternate source, "purdueintel",
was used to fix Fortran standard violations
Flags (Fortran & C): 
CPP Flags: -I. -C -P -traditional -Dmpi -DSPEC_HPG_MPI
COPTIMIZE = -O3 -static -march=pentium4 -mcpu=pentium4 -Dmpi \
            -DSPEC_HPG_MPI -DSPECDONOTNEEDARG -DFORTRAN_UNDERSCORE
FOPTIMIZE = -O3 -mp -static -fp_port -xW -axW -tpp7 
LDOPTIONS = -O3 -mp -Bstatic

Submit command to run MPI application:
PBS Version: PBSPro 5.4.1.41640
PBS Command to get resources (for cyclic allocation of processes):
qsub -I -q preemptx -lnodes=8:IB:ppn=2,walltime=1:00:00
use_submit_for_speed=1
submit=mpirun -np 16 -hostfile \$PBS_NODEFILE $command
Cluster config:
Nodes and file server use NFS shared file system
Two CPUs per node, Hyper-Threading Technology enabled
File server:
2 x 3.06 GHz Intel Xeon processors
4 GB DDR PC2100 CL2.5 ECC Registered Memory
5 x 72 GB 10K RPM SCSI Drives
Hardware RAID-5 (Dell PERC/3Di option)
Debian Linux, 3.1 "sarge"
ext3 local file system
NFS shared file system
Network (for computation and file server):
Infiniband: Topspin HBAs, Topspin 120 switches

MPICH-1.2.5 Configuration
CC=/opt/intel_cc_81/bin/icc
CXX=/opt/intel_cc_81/bin/icc
FC=/opt/intel_fc_81/bin/ifort
F90=/opt/intel_fc_81/bin/ifort
./configure --with-device=ch_vapi --with-arch=LINUX --without-mpe \
--without-romio --disable-sharedlib \
-cflags="-D_SMP_ -DUSE_STDARG  -DLAZY_MEM_UNREGISTER -DUSE_INLINE \
-DRDMA_FAST_PATH -DEARLY_SEND_COMPLETION -D_REENTRANT -O2  \
-DVIADEV_RPUT_SUPPORT -MD -fpic -DMT_LITTLE_ENDIAN  -D__LINUX__ \
-DVAPI -DTS_HOST_DRIVER -DCPU_X86 \
-I/usr/local/topspin/include -I/usr/local/topspin/include/vapi \
-L/usr/local/topspin/lib " \
-lib="-lvapi -lmtl_common -lmpga -lmosal -lpthread"
All BIOS parameters left with factory defaults.
For a description of Intel compiler flags,portability flags,
and system parameters used to generate this result,please refer
to PURDUE-20050329-INTEL-LINUX-XEON.txt in the flags directory

First published at SPEC.org on 02-Jun-2005

Generated on Thu Jul 28 18:21:54 2005 by SPEC HPC2002 HTML formatter v1.01