Seminar by Sascha Hunold : “Reproducibility in HPC: From Challenge to Reality” and Seminar by Ruben Laso: “Performance Portability in C++: standard is (maybe) the way”Seminar by Sascha Hunold : “Reproducibility in HPC: From Challenge to Reality” and

Title: Reproducibility in HPC: From Challenge to Reality

Abstract: In this talk, we will explore the challenges of reproducibility
research, with a particular focus on high-performance computing.

We will begin by reviewing the reproducibility initiatives undertaken at SC24 to enhance the reliability of scientific papers. Next, we will present a concrete example of implementing reproducible research using experimental testbeds. Specifically, we will examine the capabilities of the Chameleon Cloud platform, which provides essential tools and infrastructure for achieving truly reproducible scientific papers.

We will demonstrate, through a case study, how to achieve 'one-click' reproducible papers, making scientific research more transparent and easily verifiable.

Bio:

Sascha Hunold is an associate professor at the Technical University of Vienna, Austria.

He received the Ph.D. degree in Computer Science from the University of Bayreuth, Germany, and the M.Sc. degree in Computer Science from the University of Halle-Wittenberg, Germany.
Afterwards, he held postdoctoral positions in Berkeley, Grenoble, and Heidelberg.

His research focuses on MPI, OpenMP, scheduling algorithms, and reproducibility in the context of parallel computing.

Seminar by Ruben Laso: Performance Portability in C++: standard is (maybe) the way

Abstract:
In this talk, we will discuss performance portability in C++, introducing contenders like Kokkos and SYCL, and focusing on the possibilities within ISO C++ and the standard library (STL). We will also present pSTL, a benchmark designed to measure how different compilers, backends, and algorithms perform across various systems. Finally, we will share results from tests on both multicore CPUs and GPUs.

Bio:

Ruben Laso is a postdoctoral researcher at the University of Vienna. He earned his PhD in 2023 from the Universidade de Santiago de Compostela, specialising in High-Performance and Parallel Computing. He also holds an MSc in Industrial Mathematics (2019) and a BSc in Computer Science (2017) from the same institution. His research focuses on parallel computing, with a particular emphasis on shared-memory systems and accelerators.