Videos from JuliaCon are now available online
2018
Previous editions: 2017 | 2016 | 2015 | 2014
Przemyslaw Szufel

Warsaw School of Economics



Performance of a distributed Julia simulation on an AWS Spot Fleet vs a Cray supercomputer

In this presentation performance of Julia in-built parallelism will be demonstrated on Cray supercomputer (Okeanos - https://www.top500.org/system/178753) vs a cluster of AWS Spot Fleet instances. For illustrative purposes a massively parallelized version of the classical Schelling (1974) segregation social model will be used. Scaling of the parallelized simulation from 400 up to 4096 cores will be discussed and compared. I will also share my experience about those small details that you should know before trying to setup a Julia cluster on AWS and a Cray.

Speaker's bio

Przemyslaw Szufel is an Assistant Professor in the Decision Support and Analysis Unit at the Warsaw School of Economics, Poland. His current research focuses on distributed systems and methods for execution of large-scale simulations for numerical experiments and optimization. Dr. Szufel is currently working on asynchronous algorithms for the parallel execution of large-scale computations in the cloud and distributed computational environments.