GPU Computing Workshops

To help the NCAR and university communities prepare to use the extensive GPU-computing capabilities of the new Derecho system, CISL is launching a series of workshops and tutorials on Thursday, February 3, at 10 a.m. Mountain time for scientists, software engineers, and students in the Earth system sciences.

Storm clouds, image from Pixabay

The GPU series is for for scientists, software engineers, and students in Earth system science.

The series of one-hour sessions will be held every two weeks through August using the Zoom meeting platform. They are intended for both new and more experienced GPU programmers and require familiarity with the UNIX shell; Fortran, C/C++, and/or Python; and typical program compilation workflows.

Topics for individual sessions are listed below, with links to pages that provide more detail about topics to be covered and about registration. Sessions will include dedicated Q&A time and some will involve interactive coding with participants. Participants are not required to attend all sessions.

By registering and attending, you agree to abide by the UCAR Code of Conduct.

Feb 3 Introduction to Parallel Programming 
Please register by January 28.
Feb 17 Why Use GPU Accelerators?
Mar 3 Introduction to GPU and Accelerator Architectures
Mar 17 Software Infrastructure and Make Systems
Mar 31 Directive Based Programming with OpenACC (1 of 2)
Apr 14 Directive Based Programming with OpenACC (2 of 2)
Apr 28 Hands-On Session Using OpenACC in MPAS-A
May 19 Multi-GPU Programming (1 of 2)
Jun 2 Multi-GPU Programming (2 of 2)
Jun 16 IDEs, Debugging, and Optimization Tools for GPU Computing
Jun 30 Hands-On Session with Nsight Systems and Compute
Jul 14 Co-Design Process Training for Scientists and Project Leads
Jul 28 CuPy and NumPy/SciPy via Legate/cuNumerics
Aug 11 Numba with CUDA, CuPy, and Dask
Aug 25 Optimizing AI/ML workflows in Python for GPUs