‘The Successful Interaction of ChEESE and POP - a cross-collaboration of CoEs’

NAG, the National Algorithms Group, supplies numerical, symbolic, statistical, visualisation and simulation software through the NAG Chest Agreement.

NAG is a proud partner of the EU POP project, and via POP, regularly presents learning webinars around application performance.

Please join NAG for latest offering in the Webinar Series, taking place on Tuesday 12 November - ‘The Successful Interaction of ChEESE and POP - a cross-collaboration of CoEs’. (CoEs are Centres of Excellence)

ChEESE is a new initiative to integrate HPC and data across Solid Earth related disciplines in Europe. One of its main goals is to prepare ten flagship codes for exascale environments. These codes cover fields of study highly relevant for urgent computing, probabilistic hazard assessment and data science, such as computational seismology, magnetohydrodynamics, physical volcanology, and tsunami modeling. POP conducted performance assessments of all these codes at the start of the project and will do so again at the end.

We will look in more detail at one of these, a physical volcanology code called Fall3D and will explain how work conducted within ChEESE, aided by insights provided by the POP assessment, has led to a 10x improvement in scalability in a fairly short period of time.

About the Presenters:

Mauricio Hanzich leads the Software Development group at the Computer Applications for Science and Engineering department of Barcelona Supercomputing Center. His main focus is software research and development for the oil industry, with broad experience in the development of HPC software for the whole stack, from the hardware to the application level.

Claudia Rosas has been a Post Doc in the Computer Applications for Science and Engineering department of Barcelona Supercomputing Center since 2016, with about 10 years’ experience in software engineering and performance analysis and optimization.

Like to learn more?

Over the last year or so the POP Team, collaborators and users have presented some really insightful sessions. All were recorded and can be viewed by clicking on the links below:

• Using OpenMP Tasking
• How to Improve the Performance of Parallel Codes
• Exascale Matrix Factorization: Using Supercomputers and Machine Learning for Drug Discovery
• Software for Linear Algebra Targeting Exascale SLATE project

POP offers a unique free of charge service to EU organisations that helps analyse software and recommend improvements with a focus on HPC and parallelization. For more information see the POP website here.