Branch: master
Find file History
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Type Name Latest commit message Commit time
..
Failed to load latest commit information.
HPCG-Benchmark-3.0_2016.02.27.05.57.04.yaml
INSTALL
Make.ICPC
Make.MPI
Make.MPI_OPENMP
README.md

README.md

hpcg

Run

mpirun -n 24 -f /root/nfs/hosts ./xhpcg --nx=16 --rt=3600

Cluster Specs

Cluster CPU Cores Accelerator
6 x86 nodes Intel(R) Xeon (R) CPU 5140 @ 2.33 GHz 4 None

Version

Cluster HPCG Version MPICH Version Intel MPI
6 x86 nodes 3.0 Source 3.2 5.1.3
# Download and extract instructions
cd /nfs
wget http://www.hpcg-benchmark.org/downloads/hpcg-3.0.tar.gz
tar -xvf hpcg-3.0.tar.gz

Makefile

MPI

We based our makefile off Make.Linux_MPI from the hpcg-3.0 source. We copied it as Make.MPI. To customize, we linked the MPI libraries properly as they were configured on the 6 node cluster.

Install/Setup Instructions

We first downloaded the tar source into our NFS folder on our MPI enabled 6 node cluster. After extracting the source, we used the following commands for setup:

cd hpcg-3.0/
mkdir build
cp ../hpc/hpcg/Make.MPI setup/
cd build/
/nfs/hpcg-3.0/configure MPI
make

Benchmarking Performance - 60 seconds

Mesh sizes must be divisible by 8.

Cluster Version Mesh Size Cores GFLOP/s Type
6 x86 nodes 3.0 Source 104 6 1.31751 MPI
6 x86 nodes 3.0 Source 16 4 1.17194 MPI
6 x86 nodes 3.0 Source 16 6 2.3234 MPI
6 x86 nodes 3.0 Source 16 24 3.44749 MPI
6 x86 nodes 3.0 Source 16 24 3.39781 MPI_OPENMPI
6 x86 nodes 3.0 Source 24 24 3.10402 MPI_OPENMPI
6 x86 nodes 3.0 Source 32 24 3.02452 MPI_OPENMPI
6 x86 nodes 3.0 Source 48 24 2.31746 MPI_OPENMPI
6 x86 nodes 3.0 Source 112 24 2.815 MPI_OPENMPI
5 x86 nodes 3.0 Source 16 20 3.11515 ICPC
5 x86 nodes 3.0 Source 24 20 2.61601 ICPC
5 x86 nodes 3.0 Source 104 20 2.34242 ICPC

Benchmarking Performance - 1 hour

Cluster Version Mesh Size Cores GFLOP/s Type
6 x86 nodes 3.0 Source 16 24 3.48752 MPI_OPENMPI