Devito Workshop at 2022 Rice Energy High Performance Computing Conference

Devito is a domain-specific Language (DSL) and code generation framework for the design of highly optimised finite difference kernels for use in simulation, inversion methods, FWI and RTM. Devito utilises SymPy to allow the definition of operators from high-level symbolic equations and generates optimised software for a wide range of CPU and GPU processors.

The Devito hackathon will consist of two tracks:

  • The training stream will lead participants new to Devito from the development of finite-difference solvers within Devito through to the development of a basic FWI framework that can run on the Cloud.
  • The second track, led by a team from NVidia, will show how to leverage the Devito JIT-backdoor to enable developers to build, debug, profile, and develop class-leading and cutting-edge software that utilizes the latest visual computing hardware from Nvidia. This session profiles the OpenACC code generated by Devito using Nvidia Nsight Compute. It provides detailed performance metrics and API debugging via a command-line tool that is used to target, create, and optimize CUDA kernels for Devito.

The training track will use Jupyter notebook, which can be either installed on the participants’ laptops or run directly on the Azure Cloud instances provided for the workshop. Hackathon participants will have access to a range of Azure CPU and GPU instances for benchmarking and technical support from the Devito team.

The workshop will be concluded by a series of Lightning talks that illustrate the potential of Devito from different user perspectives ranging from end users (Chevron, Schlumberger), cloud providers (Microsoft), hardware providers (Nvidia) and academia (Georgia Tech).

Date:
Haut-parleurs:
John Washbourne, Luke Decker, Christopher Leader, Philipp Witte, Ken Hester, Mathias Louboutin, Gerard Gorman
Affiliation:
Chevron, Schlumberger, Microsoft, Nvidia, Georgia Institute of Technology, Devito Codes

Regardez suivant