Technical Programme >Dedicated Sessions >Advances in HPC for Geophysical Applications

Dedicated Sessions

Forms

Add Event To

Search


Member of  
 
 

Advances in High Performance Computing for Geophysical Applications

Convenors M. Hanzich (Barcelona Supercomputing Center)
  A. St-Cyr (Shell Global Solutions International BV)
  P. Thierry (Intel Corporation)


In order to keep the pace of our ever-growing need for more energy resources, new locations around the globe need to be explored. However, those new locations also impose new challenges to the exploration enterprise.
High Performance Computing (HPC) is one of the main tools in current exploration geophysics for enhancing the opportunities for finding new resources in such challenging scenarios. This in turn, implies a greater need in HPC resources from both, the hardware and software fronts.
For being able to provide the proper HPC answers for different situations and challenges, HPC also need to evolve and grow to provide the needed alternatives for making the exploration viable.
It is then the aim of this session to review the current and future state of the HPC solutions for many different problems such as modeling, migration or inversion. Of course, it is expected from those solutions to be aware of the characteristics of modern hardware architectures and latest advances in algorithms.
As many of the geophysical tools used for exploration are based on a internal modeling kernel, either based on waves (acoustic, elastic, visco-elastic, electromagnetic, …) or rays, contributions on those aspects will be shown.
Also, as geophysical systems tend to use more and more HPC resources and for a long period of time, aspects such as resiliency to hardware and software failures are also a main aim of the session.
As the difference in latencies and bandwidth when considering memory access, disk I/O or networks might be some orders of magnitude in current hardware architectures, there are always new challenges for keeping bottlenecks at bay. Hence, solutions for tackling such bottleneck problems are also an important issue for the current session.
Finally, as details tend to hinder efficiency most of the time in HPC-related problems, those original solutions to specifics of challenging problems will be considered and included for this session.

To submit an abstract for this dedicated session, please select topic "6.07" during the submission process and keep the template instructions in mind.
Please note that this dedicated session is a follow-up on the dedicated session on the subject, which took place in Amsterdam in 2014. Please click here to see the programme of the dedicated session in 2014.