Author: Michael Bareford, EPCC

 

Nektar++ [1] is an open-source MPI-based spectral element code that combines the accuracy of spectral methods with the geometric flexibility of finite elements, specifically, hp-version FEM. Nektar++ was initially developed by Imperial College London and is one of the ExaFLOW co-design applications being actively developed by the consortium. It supports several scalable solvers for many sets of partial differential equations, from (in)compressible Navier-Stokes to the bidomain model of cardiac electrophysiology. The test case named in the title is a simulation of the blood flow through an aorta using the unsteady diffusion equations with a continuous Galerkin projection [2]. This is a small and well-understood problem used as a benchmark to enable understanding of the I/O performance of this code. The results of this work lead to improved I/O efficiency for the ExaFLOW use cases.

The aorta dataset is a mesh of a pair of intercostal arterial branches in the descending aorta, as described by Cantwell et al. [1], see supplementary material S6 therein. The original aortic mesh contained approximately sixty thousand elements, prisms and tetrahedra. However, the tests discussed in this report use a more refined version of this dataset, one that features curved elements of aorta. The test case itself is run using the advection-diffusion-reaction solver (ADRSolver) to simulate mass transport. We executed the test case for a range of node counts, 2n, where n is in the range 1 - 8 on ARCHER [3] in order to generate the various checkpoint files that could then be used by a specially written IO benchmarker.

Author: Allan Nielsen, EPFL

 

This year, the 25th ACM International Symposium on High Performance Parallel and Distributed Computing was held in Kyoto, May 31st to June 4th. ExaFLOW was represented by EPFL with a contribution to the Fault Tolerance for HPC at eXtreme Scale Workshop. At the workshop, Fumiyoshi Shoji, Director of the Operations and Computer Technologies Division at RIKEN AIC, delivered an exciting keynote talk on their K-computer and its failures. EPFL and ExaFLOW contributed to the workshop with a talk on Fault Tolerance in the Parareal Method.

 

Author: Anna Palaiologk

 

The Exaflow project was presented by Philipp Schlatter and Erwin Laure at ISC 2016 in Frankfurt, in a workshop entitled: "Form Follows Function - Do algorithms and applications challenge or drag behind the hardware evolution?". The workshoplasted several hours during the afternoon, and drew an audience of about 25 people. Among the speakers were Jack Dongarra, Mark Parsons, Peter Messmer and Marie-Christine Sawley, representing both academia and industry (Intel, Bull-ATOS, Nvidia). Interesting discussion on different approaches about algorithms and hardware developments took place, in line with some of the basic aspects of the Exaflow project.

You can download here the Exaflow presentation slides

 

Author: Anna Palaiologk

 

During the 15th and 16th of September the third all partners project meeting took place in London. The progress of each work package was discussed as well as plans for the upcoming months. Moreover, valuable feedback was received from the Scientific Advisory Board who participated in all sessions of the meeting. After the meeting the partners visited the McLaren Technology Centre. 

As part of the initial dissemination activities of ExaFLOW the Automotive Simulation Center Stuttgart organized on the 23rd of February the exclusive workshop “Simulation Driven Design for Computational Fluid Dynamics” in Leinfelden-Echterdingen (near Stuttgart) in Germany.  The workshop was focusing the trend of embedding CFD software into the automotive design process. Amongst other questions it adressed the following:

  • Is this trend only a part of the long-lasting frontloading process or is it a paradigm shift? Who are the key roles to enable the change? 
  • How much methodological knowledge needs an engineer?
  • How do we handle the conflict of objectives: the need of less system understanding by using a black box software or solving advanced problems by detailed understanding of solver algorithms?
  • Which are important aspects for industrial usage: real-time, automated, easy-to-use, CAD-embedded, bi-directional, multi-physics enabled?