Skip to content

GitLab

  • Menu
Projects Groups Snippets
    • Loading...
  • Help
    • Help
    • Support
    • Community forum
    • Submit feedback
    • Contribute to GitLab
  • Sign in
  • T Trilinos
  • Project information
    • Project information
    • Activity
    • Labels
    • Planning hierarchy
    • Members
  • Repository
    • Repository
    • Files
    • Commits
    • Branches
    • Tags
    • Contributors
    • Graph
    • Compare
  • Issues 936
    • Issues 936
    • List
    • Boards
    • Service Desk
    • Milestones
  • Merge requests 22
    • Merge requests 22
  • CI/CD
    • CI/CD
    • Pipelines
    • Jobs
    • Schedules
  • Deployments
    • Deployments
    • Environments
    • Releases
  • Monitor
    • Monitor
    • Incidents
  • Analytics
    • Analytics
    • Value stream
    • CI/CD
    • Repository
  • Wiki
    • Wiki
  • Snippets
    • Snippets
  • Activity
  • Graph
  • Create a new issue
  • Jobs
  • Commits
  • Issue Boards
Collapse sidebar
  • James Willenbring
  • Trilinos
  • Issues
  • #4581

Closed
Open
Created Mar 08, 2019 by James Willenbring@jmwilleMaintainer

MueLu: create GeometricInterpolationPFactory_kokkos

Created by: lucbv

@trilinos/muelu

Expectations

The aim is to create refactored factory and to have it perform well on manycore architecture and on GPUs.

Current Behavior

Currently the GeometricInterpolation is performed serially in MueLu, although the use of Tpetra makes it already reasonably performant it can be further optimized.

Motivation and Context

This work is part of a general effort to get structured aggregation and geometric multigrid perform well on NGPs

Definition of Done

Well have this factory scale better than it already does. This is fairly vague but the goal is to optimize so being done is hard to measure.

Possible Solution

Right now I would like to modify the way the coarse coordinates are computed to bypass the construction of an importer. It would also be good to improve the performance of the linear interpolation calculation in the case of piece-wise linear interpolation. This can be done using calls to a batched LU.

Related Issues

  • Blocks
  • Is blocked by
  • Follows #2230
  • Precedes
  • Related to
  • Part of
  • Composed of
Assignee
Assign to
Time tracking