Speed up TopologyComputation

Registered by Garth Wells

From Bugzilla:

It takes quite a while to compute the facet topology for large 3D mesh. This is
needed to load a facet meshfunction for assembly for exterior facet integrals
over subdomains.

A speed up of this procedure would be nice!

See:
<http://www.fenics.org/pipermail/dolfin-dev/2008-June/008220.html>

Johan

Blueprint information

Status:
Complete
Approver:
None
Priority:
Medium
Drafter:
None
Direction:
Needs approval
Assignee:
None
Definition:
Approved
Series goal:
Accepted for 1.0.x
Implementation:
Implemented
Milestone target:
None
Started by
Anders Logg
Completed by
Anders Logg

Related branches

Sprints

Whiteboard

AL: Don't know how to speed this up. I've done my very best to make it fast. Feel free to suggest a better algorithm. (Which is not to say there may not be simple optimizations, inlining etc that can speed up TopologyComputation.cpp.)

See this Blueprint:

    https://blueprints.launchpad.net/dolfin/+spec/mesh-file-connectivity

AL: Update: By optimizing compute_from_intersection in TopologyComputation, the speed has now been improved. The speedup is roughly a factor 5.

(?)

Work Items

This blueprint contains Public information 
Everyone can see this information.

Subscribers

No subscribers.