Skip to main content

Vortex-resolving modeling of gas dynamics problems on hybrid supercomputers

The meeting of the seminar

«Supercomputer technologies

in science, education and industry»

Chairman of the organizing committee
В.А. Садовничий V.A. Sadovnichy
academician, rector of Lomonosov Moscow State University

Co-Chairs of the organizing committee
Вл.В. Воеводин Vl.V. Voevodin
Corr. RAS, Director of Research Computing Center of Moscow State University
И.А. Соколов I.A. Sokolov
academician, dean of the faculty of Computational Mathematics and Cybernetics of Lomonosov Moscow State University

Scientific Secretary of the seminar
  E.V. Mortikov
Ph.D., Research Computing Center of Moscow State University

 

The seminar was organized on the basis of the Scientific and Educational Center «Supercomputer Technologies». The reports cover all aspects of the use of supercomputers, parallel computing systems and distributed data processing methods for solving large computational problems. The seminar has a pronounced interdisciplinary nature, however, the various nuances of using supercomputer technologies are of interest to researchers from various fields.

The seminar is supported by the Moscow Center of Fundamental and Applied Mathematics.

 


PROGRAM
16:30
Gorobets A.V. Doctor of Physical and Mathematical Sciences, M.V. Keldysh Institute of Applied Mathematics RAS

Vortex-resolving modeling of gas dynamics problems on hybrid supercomputers

The report is devoted to resource-intensive supercomputer modeling of gas dynamics problems on hybrid computing systems. Heterogeneous parallel NOISETTE code for vortex-resolving calculations of aerodynamics and aeroacoustics problems will be presented, which can simultaneously use central and graphic processors. The numerical algorithm is based on edge-oriented schemes of increased accuracy on unstructured mixed meshes. Implicit time schemes, hybrid RANS-LES turbulence models are used. Parallel implementation allows you to use a large number of computing devices of various types: multi-core CPUs, GPUs from various manufacturers, including NVIDIA, AMD, Intel, Intel Xeon Phi processors and accelerators. The report will talk about the adaptation of a numerical algorithm to reduce memory consumption, adaptation of parallelization to the stream processing paradigm, implementation of computations using the open standard OpenCL, organization of data exchange, implementation of a heterogeneous mode with simultaneous use of CPU and GPU. Parallel efficiency in calculations on various supercomputers will be shown.

 


The seminar will be held in the form of a webinar on the Zoom platform.

To participate in the webinar, we ask you to enter your data in the Google spreadsheet:

https://docs.google.com/spreadsheets/d/1f73UBZoglxwRIpm7rSjbmHJDZ__FfjaydyRco43chJs/edit?usp=sharing

A link to the Zoom conference will subsequently be sent to your email address.