Hybrid computing platform to enable complex multi-physics DNS/LES from desktop to exascale

Abstract

Abstract Publically ReleasableWe propose to develop the hybrid CPUGPU paradigm for unstructured overset grid largeeddysimulation,, that promises transformational speedups in computing time. Our goal is to enableexascale simulations on the largest platforms as w,ell as highly affordable simulations on desktopmachines. We will develop the capability for our inhouse solver MPCUGLES and demonst,rate itsimpact on our current ONR projects. The speedup enabled by the proposed research willsignificantly impact the utilization of, LES for practical problems. Developing a GPU extension fora code as complex as MPCUGLES is not straightforward and involves signifi,cant algorithm anddatahandling development, and hardwareaware programming. The proposed machine isessential to (i) develop the hyb,rid accelerated methodology, (ii) evaluate it for relevant problemslike crashback, cavitation and hull maneuvering, (iii) explore th,e potential in detail in terms ofwhat is feasible on machines ranging from desktop machines to leadership class machines. If theprom,ise of the capability being developed is realized, it will enable LES of practical Navy relevantproblems to be performed on desktops,.

Document Details

Document Type
DoD Grant Award
Publication Date
Sep 03, 2022
Source ID
N000142212520

Entities

People

  • Krishnan Mahesh

Organizations

  • Office of Naval Research
  • Regents of the University of Minnesota
  • United States Navy

Tags

Readers

  • Computational Fluid Dynamics (CFD)
  • Distributed Systems and Data Platform Development