CEED Miniapps
CEED is developing a variety of miniapps encapsulating key physics and numerical kernels of high-order applications. The miniapps are designed to be used in a variety of co-design activities with ECP vendors, software technologies projects as well as external partners.
CEED's current miniapps are documented below and can be found on GitHub. Please contact the CEED team if you have any questions.
Nekbone
Nekbone solves a standard Poisson equation using a conjugate gradient iteration with a simple or spectral element multigrid preconditioner on a block or linear geometry.
It exposes the principal computational kernel to reveal the essential elements of the algorithmic-architectural coupling that is pertinent to Nek5000, relevant to large eddy simulation (LES) and direct numerical simulation (DNS) of turbulence in complex domains.
Nekbone-3.1 was released in Oct 2013 as part of the CESAR co-design center.
Laghos New
Laghos solves the time-dependent Euler equation of compressible gas dynamics in a moving Lagrangian frame using high-order finite element spatial discretization and explicit high-order time-stepping.
It captures the basic structure of many other compressible shock hydrocodes, including the BLAST code at Lawrence Livermore National Laboratory. The miniapp is build on top of a general discretization library, MFEM, thus separating the pointwise physics from finite element and meshing concerns.
Computational motives in Laghos include the following:
- Support for moving (high-order) unstructured 2D and 3D meshes.
- Continuous and discontinuous high-order finite element discretization spaces of runtime-specified order.
- Explicit time-stepping loop with a variety of time integrator options. Laghos supports Runge-Kutta ODE solvers of orders 1, 2, 3, 4 and 6.
- Full and partial assembly options for deriving and solving the resulting ODE systems.
- Point-wise definition of mesh size, time-step estimate and artificial viscosity coefficient.
- Domain-decomposed MPI parallelism.
- Optional in-situ visualization with GLVis and data output for visualization / data analysis with VisIt.
NekCEM CEEDling New
NekCEM CEEDling is a NekCEM miniapp, solving the time-domain Maxwell equations for electromagnetic systems. It allows software and hardware developers to understand the basic structure and computational costs of NekCEM over a broad spectrum of architectures ranging from software-based simulators running at one ten-thousandth the speed of current processors to exascale platforms running millions of times faster than single-core platforms.
NekCEM provides flexibility to adapt new programming approaches for scalability and performance studies on a variety of platforms without having to understand all the features of NekCEM, currently supporting CPU/GPU runs.
Holmes New
Holmes is an experimental testbed for multi-level parallel implementations of high-order finite element computations.
Its features include:
- Elements
- Spectral element quadrilateral and hexahedral elements
- Warp & blend high-order nodal triangular and tetrahedral elements
- Discretizations
- Elliptic: continuous and discontinuous Galerkin discretization
- Hyperbolic: discontinuous Galerkin time-domain
- Solvers
- Two level overlapping additive Schwartz elliptic problem preconditioning
- Fast approximate block preconditioning
- Algebraic multigrid coarse solver with heterogeneous acceleration
- Parallel direct solver XXT for the coarsest level solve
- Full algebraic multigrid preconditioning
- Two level overlapping additive Schwartz elliptic problem preconditioning
- Implementation
- Portable on-node parallelism via the Open Concurrent Compute Abstraction, OCCA
- MPI communications for distributed computations
- Dependencies
- The XXT parallel direct solver for the coarsest AMG solve
- GSLIB for optimized gather-scatter operations
- Example CEEDling apps
- Incompressible flow solvers
- Elliptic solvers
- Boltzmann flow solver
- Acoustics time-domain solver
Holmes is being developed in Tim Warburton's group at Virginia Tech.
The project GitHub repository will be made public in 2017. Stay tuned for more details...
HPGMG-FE
The High-Performance Geometric Multigrid Finite Element benchmark is a Full Multigrid solver for a third order accurate finite element discretization of an elliptic operator on mapped grids. The benchmark is designed to give a picture of the "performance spectrum" across a range of problem sizes and thus solution time, thereby giving a clear picture of both weak and strong scaling concerns.
For details, see the HPGMG website.
MFEM Example Codes
The MFEM distribution includes a large number of simple example codes that can be viewed as simple miniapps for model high-order physics:
- Example 1: nodal H1 FEM for the Laplace problem,
- Example 2: vector FEM for linear elasticity,
- Example 3: Nedelec H(curl) FEM for the definite Maxwell problem,
- Example 4: Raviart-Thomas H(div) FEM for the grad-div problem,
- Example 5: mixed pressure-velocity FEM for the Darcy problem,
- Example 6: non-conforming adaptive mesh refinement (AMR) for the Laplace problem,
- Example 7: Laplace problem on a surface (the unit sphere),
- Example 8: Discontinuous Petrov-Galerkin (DPG) for the Laplace problem,
- Example 9: Discontinuous Galerkin (DG) time-dependent advection,
- Example 10: time-dependent implicit nonlinear elasticity,
- Example 11: parallel Laplace eigensolver,
- Example 12: parallel linear elasticity eigensolver,
- Example 13: parallel Maxwell eigensolver,
- Example 14: Discontinuous Galerkin (DG) for the Laplace problem,
- Example 15: dynamic AMR for Laplace with prescribed time-dependent source,
- Example 16: time-dependent nonlinear heat equation,
- Example 17: Discontinuous Galerkin (DG) for linear elasticity,
- Example 18: Discontinuous Galerkin (DG) for the Euler equations,
- Example 19: incompressible nonlinear elasticity.
Most of the examples have a serial and a parallel version, illustrating the ease of transition and the minimal code changes between the two.
Of particular relevance to CEED is the HPC versions of the example codes that use a set of templated classes to efficiently implement high-order operator evaluation.