For the application partners in DEEP-ER, Barcelona was the place to be last week: From May 11th to 13th, application developers from Astron, BSC, Inria and KU Leuven were working intensively on optimising their codes for the DEEP-ER system at Barcelona Supercomputing Center. Together with experts from Intel, BSC and JSC, the DEEP-ER software developers were working intensively on preparing their codes for the upcoming DEEP-ER exascale-enabling machine. After three background presentations on usage and implementation of scalable checkpoint-restart in DEEP-ER, OmpSs and task-based resiliency, and performance optimisation for KNL – the new generation of Intel Xeon Phi – a clear focus was devoted to extensive hands-on sessions.
Overall, the working atmosphere was great and the team managed to make a leap forward. The partners are confident they will be able to demonstrate significant progress on the application side for the next project review in autumn.
"The idea of the training clearly was 'bring experts together'. Thanks to everyone's commitment we have made a significant step forward towards better resiliency and performance. I am looking forward to seeing how this reflects in the final system," Damian Alvarez, Team Lead Applications DEEP-ER.
Thanks to all partners involved and special thanks to BSC for hosting the workshop! It was extremely productive and fun at the same time ;)