ffc team mailing list archive

Thread
Date

[HG] Modify implementation of vertexeval() to only compute an array

To: ffc-dev@xxxxxxxxxx
From: FFC <ffc@xxxxxxxxxx>
Date: Tue, 30 May 2006 09:43:36 +0200

One or more new changesets pushed to the primary FFC repository.
A short summary of the last three changesets is included below.

changeset:   392:69d4b4f1a0cdd0a89f6d7942ac0e338a4efc32ce
tag:         tip
user:        "Anders Logg <logg@xxxxxxxxx>"
date:        Tue May 30 09:43:25 2006 +0200
files:       src/ffc/compiler/vertexeval.py src/ffc/format/dolfin.py src/ffc/format/dolfinswig.py
description:
Modify implementation of vertexeval() to only compute an array
of vertex nodes, not actually picking the values from the vector.


changeset:   391:38b7d176311f91e48a94d83a642f15c0cb07f367
user:        "Anders Logg <logg@xxxxxxxxx>"
date:        Thu May 11 13:00:10 2006 +0200
files:       src/bench/bench.log src/ffc/compiler/elementtensor.py
description:
Further optimization of the code generation. More optimizations are
possible but would require some substantial effort.


changeset:   390:abe26644b7fe8b67df6526aee407b8c060bebab9
user:        "Anders Logg <logg@xxxxxxxxx>"
date:        Thu May 11 10:39:06 2006 +0200
files:       src/bench/bench.log src/ffc/compiler/elementtensor.py
description:
Optimize code generation (patch from Garth).

Time to compile the new benchmark problem (Stabilization.form)
reduced from 68.606 seconds to 46.744 seconds (about 30%).


----------------------------------------------------
For more details, visit http://www.fenics.org/hg/ffc