With increasing heterogeneity in both computational devices and memory organization in high performance computing platforms, movement of data between devices and layers of memory has become a critical challenge for science and engineering users of these platforms. Imminent arrival of exascale platforms has brought these concerns to the forefront.
The objective of this workshop is to understand the impact of abstractions and tools that have been around for some time and have seen some adoption.
COLOC is a joint workshop of EuroPar 2021.

2:20pm CEST/ 7:20am CDT – Introduction

Keynote 1 : Torsten Hoefler, Professor, Department of Computer Science, ETH Zurich

2:30pm CEST/7:30am CDT –  Data-Centric Python – Productivity, portability and all with high performance!


3:00pm CEST/8:00am CDT –  Locality-Aware Scheduling of Independent Tasks for Runtime System

3:30pm CEST/8:30am CDT – High Performance Computing with Java Streams

4:00pm CEST/9:00 am CDT – Break

4:15pm CEST/9:15am CDT – Exploring Strategies to Improve Locality Across Many-core Affinities

4:40pm CEST/9:40am CDT – Monitoring Collective Communication Among GPUs


Keynote 2 : William E. Hart — Data Anal. and Opt. Lead, The Exascale Computing Project 5:05pm CEST/10:00am CDT – ECP: Data Analytics and Optimization Applications On Accelerator-Based Systems


5:40pm CEST/10:40am CDT – Panel



Comments are closed.