In this lecture, we will cover the basics of programming NVIDIA GPUs using CUDA and OpenACC to enable high-performance computing. By the end, participants will have a solid understanding of GPU hardware and how to leverage parallel processing for applications involving intensive data processing. The lecture will address memory management, kernel execution, and optimization techniques in both the CUDA and OpenACC environments. We will compare and contrast these two powerful programming models, highlighting their unique features, advantages, and use cases.
Speaker
Nitin Shukla – CINECA
Nitin Shukla received his Master’s in Physics Engineering and PhD in Computational plasma physics from Instituto Superior Técnico (IST), Department of Physics, (Lisbon, Portugal). After his PhD, he joined the supercomputing centre CINECA in Bologna, Italy, as a High-Performance Computing (HPC) analyst, where he is involved in developing and deploying parallel codes on modern HPC architectures with a variety of programming languages and paradigms. Nitin Shukla is the coordinator of SPACE center of execellence EU project of CINECA and PI of the project try21 to port the plasma code ECsim to GPU using OpenACC, co-developer of the CUDA version of the XShell code (H2020 ChEESE project). He is also DevTeam member of the project DESTINE for optimisating, performing analysis of the code RAPS20.
He is also a technical referee for national and international HPC calls and convener of courses on JuliaLang, OpenMP, OpenACC, and CUDA.
Event Timeslots (2)
Thu 19 – Accelerators
-
N. Shukla (CINECA)
Thu 19 – Accelerators
-
N. Shukla (CINECA)
