Example Applications Built with the

Charm++ ParFUM Framework

1) Parallel Spacetime Discontinuous Galerkin

2) Rocrem: Remeshing a Rocket

Isaac Dooley

Parallel Programming Lab

University Illinois Urbana-Champaign

http://charm.cs.uiuc.edu

Parallel Spacetime Discontinuous Galerkin

LACSI 2005 3

Overview• My background in parallel programming• How the Spacetime Discontinuous Galerkin

Method utilizes unstructured meshes.• Requirements to parallelize SDG• Existing functionality in ParFUM which satisfies

some SDG requirements• New functionality which has been added to

ParFUM to support the rest of the SDG requirements

LACSI 2005 4

Parallel Programming Lab• Our focus is parallel programming, especially in

frameworks and dynamic or adaptive applications• We are not Computational Geometers, nor

Mathematicians.• We try to build general purpose reusable high

performance frameworks• Charm++ and AMPI• Focus on Migratable Objects and Virtualization• Multiple Platforms (Clusters, SMPs, BlueGene/L)

LACSI 2005 5

Spacetime Discontinuous Galerkin• Collaboration with:

• Bob Haber, Jeff Erickson, Mike Garland, …• NSF funded center

• SDG Motivation: Spatial adaptivity is needed in structural dynamics applications. Why shouldn’t we also adapt in the time dimension?

LACSI 2005 6

Spacetime Discontinuous Galerkin

• Mesh generation is an advancing front algorithm called Tent Pitcher.

• Adds a set of new elements called patches to the mesh, then solves them, thus advancing the front.

• Each patch depends only on inflow elements.

LACSI 2005 7

Unsolved Patches

1-d Mesh Generation

Tent Pole Tim

e

Space

LACSI 2005 8Solved Patches

Unsolved Patches

1-d Mesh Generation

Tim

e

LACSI 2005 9Solved Patches

Unsolved Patches

Refinement

1-d Mesh Generation

Tim

e

LACSI 2005 10

Adaptive SDG

• Method described in• Abedi, Zhou, et. al.; Spacetime meshing with

adaptive refinement and coarsening; 2004

• Tent poles are not just pitched above existing space nodes

• Entire space-time mesh or frontier is built as a mesh. Non-adaptive SDG can store patches as attributes of nodes in original mesh.

LACSI 2005 11

2-d Adaptive Mesh Generation

LACSI 2005 12

2-d Adaptive Mesh Generation

LACSI 2005 13

2-d Adaptive Mesh Generation

LACSI 2005 14

2-d Adaptive Mesh Generation

LACSI 2005 15

2-d Adaptive Mesh Generation

LACSI 2005 16Courtesy: Shuo-Heng and Michael Garland

QuickTime™ and aMPEG-4 Video decompressor

are needed to see this picture.

LACSI 2005 19

Master/Slave Parallelization of SDG

• The first parallelization of the SDG method was based on the observation that each patch could be solved independently.

• Thus the space mesh is not partitioned, but maintained on one master processor.

• Workers request patches to solve from the master processor.

• This method resulted in a bottleneck at the processor holding the entire space mesh.

LACSI 2005 20

Must Parallelize the Geometry!

• Do not want a single processor bottleneck.• We have an initial mesh, we’ll partition the geometric

space mesh.• We need consistent ghost element layer.• We need a locking mechanism for updating ghost values at

appropriate times to ensure we have a consistent mesh.• We will need the ability to incrementally add/remove

elements to/from the mesh, maintaining consistency across all processors.

LACSI 2005 25

ParFUM - New FeaturesRequired for non-adaptive space-time meshing

•Incremental updates to ghost layers of adjacent processors•Locking of individual elements or nodes. •No global synchronization.•New adjacency data structures

•Element to element•Node to node•Node to element

0

20

40

60

80

100

120

140

0 10 20 30 40 50 60 70

Processors

Patches Per Second

Non-adaptive SDG Program Initial Results

LACSI 2005 28

ParFUM Support For Adaptivity

• Load balancing is required in any efficient framework for adaptive SDG, since mesh partitions can differ in size by orders of magnitude.

• We have already extended ParFUM to provide parallel incremental mesh modification primitives.

• The primitives allow simple coding of incremental mesh modification, refinement, coarsening, and repair routines.

Rocrem: Remeshing a Rocket

LACSI 2005 30

Rocrem: Remeshing a Rocket• Old mesh distorted by burning and pressure• Create new mesh from old by remeshing

• Using Simmetrix for sequential tetrahedral meshes• Performs variable mesh sizing

• Repartition new mesh• Transfer solution data in parallel from deformed

mesh to new mesh• Uses accurate conservative transfer (volume-weighted

averaging) for cell-centered solution data of Rocflu fluids mesh

• Uses linear interpolation for node-centered solution data

• Restart simulation using new mesh and solution data

LACSI 2005 31

Serial Remeshing• ParFUM: stitch together partitioned surface mesh to

form a serial mesh• Use Simmetrix to remesh surface and create volume

• Uses regional sizing

• Internally coarse volume must be refined

LACSI 2005 32

Parallel Solution Transfer• ParFUM: partition mesh for parallel transfer

• Collision Detection Library: match up old mesh with overlaid new mesh

• Data Transfer Library: conservative, accurate volume-to-volume transfer

LACSI 2005 33

Special Features• Scalable parallel solution

transfer

• Supports both node- and cell-centered solution data

• Supports node- and face-centered boundary conditions

• Preserved through remeshing and data transfer process

• Boundary extrusion

• Complete HDF to HDF conversion with restart files

LACSI 2005 34

Boundary Extrusion• After remeshing, new mesh

boundary does not perfectly coincide with old mesh boundary

• Curved surface yields points on new surface beyond surface of old mesh

• Solution: extrude boundary faces of old mesh to cover new mesh

• All volumes of new mesh now covered by old mesh

• Proceed with solution transfer

LACSI 2005 35

Thank-you for listening to my talk!

Questions or Comments?