ffc team mailing list archive

[Anders Logg <logg@xxxxxxxxx>]

ok. I'll fix it. The problem is in xml.py. It now needs to get the
size of the block (size_a) from somewhere else, since the block values
are no longer generated for the XML format.

/Anders

On Sat, Oct 08, 2005 at 10:01:24AM +0200, Garth N. Wells wrote:
> I tried using the BLAS format, but changes in src/ffc/compiler/elementtensor.py
> have broken the .xml files. "size_a" is always zero, which leads to a DOLFIN
> error. The problems appears to be on line 117 of src/ffc/compiler/elementtensor.py. 
> 
> Garth
> 
> Quoting ffc@xxxxxxxxxx:
> 
> > Commit from logg (2005-10-07 22:58 CEST)
> > ----------------
> > 
> > A bunch of optimizations:
> > 
> >   - Only precompute declarations a0, aK, gK when needed as decided by
> >     the current format. (Formats should return None when not needed.)
> > 
> >   - Write BLAS XML data directly to file instead of caching in a string.
> >     Very little difference in speed and much cheaper in memory for large
> >     files.
> > 
> > These optimizations considerably improve compile-time and memory usage
> > for complicated forms. Test results for Cahn-Hilliard will be reported
> > in a separate post.
> > 
> >   ffc  TODO                               1.49
> >   ffc  src/bin/ffc                        1.28
> >   ffc  src/ffc/compiler/elementtensor.py  1.17
> >   ffc  src/ffc/format/ase.py              1.9
> >   ffc  src/ffc/format/dolfin.py           1.48
> >   ffc  src/ffc/format/latex.py            1.17
> >   ffc  src/ffc/format/raw.py              1.13
> >   ffc  src/ffc/format/xml.py              1.5
> > 
> > _______________________________________________
> > FFC-dev mailing list
> > FFC-dev@xxxxxxxxxx
> > http://www.fenics.org/cgi-bin/mailman/listinfo/ffc-dev
> > 
> 

-- 
Anders Logg
Research Assistant Professor
Toyota Technological Institute at Chicago
http://www.tti-c.org/logg/