Join our global community on Discord and let's shape the next wave of simulation and scientific computing Let's shape the future of simulation together

Join us on Discord

AMR-Wind Benchmark - Stable Atmospheric Boundary Layer Scaling on the Cloud

This benchmark evaluates the performance of the AMR-Wind simulator by running a stable atmospheric boundary layer simulation. The simulation uses an uniform mesh resolution and directly follows well-established reference studies in the literature.

Last Updated on: 2025-09-18

The original input files can be found here.

🔬 Simulation Details

Category	Parameter	Value
Domain	Size (x,y,z)	400 × 400 × 400 m³
	Grid resolution	512 × 512 × 512
Time Control	Max simulated time	36000 s
	Fixed timestep	0.0625 s
	CFL factor	0.95

Software Versions

The software versions used in this benchmark were the following:

Component	Version
AMR-Wind	3.7.0
MPI	OpenMPI v4.1.2
Operating System	Ubuntu 22.04.5 LTS
kernel	6.8.0-83-generic

Computational Resources Tested

The benchmark was conducted on the following machine series:

Intel-based

C2
Intel Xeon Scalable (Cascade Lake)
C3
Intel Xeon Scalable (Sapphire Rapids)
C4
Intel Xeon Scalable (Granite Rapids) — latest generation

AMD-based

C2D
AMD EPYC (Milan)
C3D
AMD EPYC (Genoa)
C4D
AMD EPYC (Turin) — latest generation

Notes on Machine Selection

This simulation requires approximately 150 GB of memory (with actual usage varying depending on the number of cores).
As a result, some machine configurations were excluded due to insufficient resources, including:
- c3d-highcpu-60
- c4d-highcpu-64
- Local test machines
- Other similar or lower end setups
Within the C2 series, only the standard configuration was evaluated, since highcpu variants are not available.

See the results of the benchmark on the following pages:

FAQ

Find answers to commonly asked questions about AMR-Wind.

Execution Times Comparison

This plot shows how the total runtime of the simulation varies across different configurations.