ACES: Introduction to PyFR, a Scalable Open-source CFD Flow Solver


Instructor: Sambit Mishra

Time: Tuesday, March 26, 2024 — 10:00AM-12:30PM CT

Location: online using Zoom

Prerequisites: Current ACCESS ID, basic Linux/Unix skills, basic Python skills.

This one-hour course offers a beginner-level introduction to running PyFR simulations on clusters, covering PyFR's advantages over traditional solvers, practical setup and execution of simulations using various backends, and orchestrating simulations across multiple nodes and accelerators, with all necessary scripts provided for hands-on learning. This is a research-level course aimed for graduate level students who are working with fluid flow solvers and software, providing them with a more versatile, efficient and scalable alternative to perform fluid flow simulations on the clusters. Learning Outcomes: Understanding of PyFR's advantages and hands-on experience in simulation configuration and execution.

Course Materials

  • Introduction to PyFR (Spring 2024): PDF


Attendees will log in to the cluster during the training session in order to follow examples and complete exercises with the instructor's guidance.

All necessary scripts for setup and execution provided before the course.


This course will be organized as follows:

  • Introduction to PyFR (15 min): Overview of PyFR
    • Comparison with OpenFOAM and StarCCM+, citing research papers.
    • Emphasis on PyFR's scalability and flexibility across CPUs and GPUs.
  • Simulation Setup and Execution (1 hour): Explanation of configuration, mesh files, and scripts.
    • Step-by-step guide to run a 2D incompressible flow past cylinder simulation using a CUDA backend.
    • Demonstration of data post-processing and visualization in Paraview.
  • Scaling to many accelerators (30 min): Setup and execution of a larger case using slurm scripts.
    • Showcasing ease of switching between PyFR's OpenMP, CUDA, and OpenCL backends on the ACES platform.
    • Guidance on orchestrating simulations across multiple nodes and accelerators.

Note: During the class sessions many aspects of the material will be illustrated live via a login to a training system. Attendees will follow along and complete the exercises on their own laptops.

We gratefully acknowledge support from NSF award #1925764, CC* Team: SWEETER -- SouthWest Expertise in Expanding, Training, Education and Research, Texas A&M's High Performance Research Computing, and Texas A&M's Laboratory for Molecular Simulation.