Talking CFD, GPU acceleration, and Fortran+Python with Marco Rosenzwieg

0:00 How Marco got into CFD and high performance computing 10:56 Journey onto GPU accelerated platforms, from Nvidia to AMD 12:39 Transitioning to AMD GPUs from Nvidia GPUS 17:04 Example walkthrough demonstrating how to combing OpenACC with HIPBLAS in a Fortran application 26:48 Overview of the Flow Analysis Suite for Turbulence and Engineering (FASTEN) for post-processing CFD output 30:28 Deep dive into how to expose Fortran+MPI subroutines to Python with f2py 38:42 How to control the name of the python library created by f2py 42:04 Discussion about pros of Python+MPI+Fortran for a post-processing toolkit 44:21 The portability problem for Python+MPI+Fortran 48:46 Geniuses from Pawsey Supercomputing Centre to the rescue 50:02 A demo of launching Singularity/Apptainer container with FASTEN at Pawsey Supercomputing Centre In a fascinating discussion with Marco Rosenzweig, a PhD candidate at the University of Melbourne, we explored cutting-edge developments in computational fluid dynamics (CFD) and post-processing tools for turbulence analysis. This conversation shed light on the pivotal role of programming languages like Fortran and Python, the integration of GPU acceleration, and the innovative ways researchers are addressing challenges in aerospace engineering and combustion research. Learn about how Marco and his team use OpenACC and HIPBLAS to accelerate their CFD code. Additionally, we talk about how to use python, Fortran libraries, and f2py with MPI to very simply and quickly post-process large amounts of CFD data on supercomputers, like Pawsey's Setonix.