OpenFOAM is free CFD software. While one can build it from source, there are also prebuilt Ubuntu repositories that I used in my testing.

After installing, it tells you to:

** To use OpenFOAM please add
**
**    . /opt/openfoam5/etc/bashrc
**
** To your ~/.bashrc

Looking on the openfoamwiki.net I find a motorbike benchmark used to compare different systems. Downloading the tar file for this benchmark and it includes a “run.sh” script that fairly easily runs OpenFOAM.

Output from a trial run of the script is below:

Prepare case run_1...
Running surfaceFeatureExtract on /home/mev/openfoam-motorbike/bench_template/run_1
Running blockMesh on /home/mev/openfoam-motorbike/bench_template/run_1
Running decomposePar on /home/mev/openfoam-motorbike/bench_template/run_1
Running snappyHexMesh on /home/mev/openfoam-motorbike/bench_template/run_1

real	25m20.487s
user	25m7.447s
sys	0m11.482s
Prepare case run_2...
Running surfaceFeatureExtract on /home/mev/openfoam-motorbike/bench_template/run_2
Running blockMesh on /home/mev/openfoam-motorbike/bench_template/run_2
Running decomposePar on /home/mev/openfoam-motorbike/bench_template/run_2
Running snappyHexMesh in parallel on /home/mev/openfoam-motorbike/bench_template/run_2 using 2 processes

real	17m21.154s
user	34m26.927s
sys	0m9.875s
Prepare case run_4...
Running surfaceFeatureExtract on /home/mev/openfoam-motorbike/bench_template/run_4
Running blockMesh on /home/mev/openfoam-motorbike/bench_template/run_4
Running decomposePar on /home/mev/openfoam-motorbike/bench_template/run_4
Running snappyHexMesh in parallel on /home/mev/openfoam-motorbike/bench_template/run_4 using 4 processes

real	11m1.560s
user	43m44.388s
sys	0m10.989s
Prepare case run_6...
Running surfaceFeatureExtract on /home/mev/openfoam-motorbike/bench_template/run_6
Running blockMesh on /home/mev/openfoam-motorbike/bench_template/run_6
Running decomposePar on /home/mev/openfoam-motorbike/bench_template/run_6
Running snappyHexMesh in parallel on /home/mev/openfoam-motorbike/bench_template/run_6 using 6 processes

real	12m5.547s
user	61m10.749s
sys	11m4.611s
Prepare case run_8...
Running surfaceFeatureExtract on /home/mev/openfoam-motorbike/bench_template/run_8
Running blockMesh on /home/mev/openfoam-motorbike/bench_template/run_8
Running decomposePar on /home/mev/openfoam-motorbike/bench_template/run_8
Running snappyHexMesh in parallel on /home/mev/openfoam-motorbike/bench_template/run_8 using 8 processes

real	12m10.087s
user	84m39.644s
sys	12m12.322s
Prepare case run_12...
Running surfaceFeatureExtract on /home/mev/openfoam-motorbike/bench_template/run_12
Running blockMesh on /home/mev/openfoam-motorbike/bench_template/run_12
Running decomposePar on /home/mev/openfoam-motorbike/bench_template/run_12
Running snappyHexMesh in parallel on /home/mev/openfoam-motorbike/bench_template/run_12 using 12 processes

real	14m1.820s
user	87m30.173s
sys	24m19.382s
Prepare case run_16...
Running surfaceFeatureExtract on /home/mev/openfoam-motorbike/bench_template/run_16
Running blockMesh on /home/mev/openfoam-motorbike/bench_template/run_16
Running decomposePar on /home/mev/openfoam-motorbike/bench_template/run_16
Running snappyHexMesh in parallel on /home/mev/openfoam-motorbike/bench_template/run_16 using 16 processes

real	13m34.016s
user	83m47.256s
sys	24m21.869s
Prepare case run_20...
Running surfaceFeatureExtract on /home/mev/openfoam-motorbike/bench_template/run_20
Running blockMesh on /home/mev/openfoam-motorbike/bench_template/run_20
Running decomposePar on /home/mev/openfoam-motorbike/bench_template/run_20
Running snappyHexMesh in parallel on /home/mev/openfoam-motorbike/bench_template/run_20 using 20 processes

real	15m49.668s
user	93m33.448s
sys	32m39.249s
Prepare case run_24...
Running surfaceFeatureExtract on /home/mev/openfoam-motorbike/bench_template/run_24
Running blockMesh on /home/mev/openfoam-motorbike/bench_template/run_24
Running decomposePar on /home/mev/openfoam-motorbike/bench_template/run_24
Running snappyHexMesh in parallel on /home/mev/openfoam-motorbike/bench_template/run_24 using 24 processes

real	16m54.700s
user	104m49.869s
sys	30m5.564s

It looks like it runs the same workload but with 1,2,4,6,8,12,16,20 and 24 processes. Also interesting to see the lowest elapsed time is actually 6 processes and not either 4 or 8 for my 4-core, 8-hyperthread processor.