ffc team mailing list archive

[ffc@xxxxxxxxxx]

Commit from logg (2005-12-09 22:19 CET)
----------------

Further speedup computation of reference tensor.

Using Numeric ufunc add(A0, B, A0) with third argument for storage
shaves another factor 2 or so off the computation of the reference
tensor.

  ffc  ChangeLog                                1.107
  ffc  src/ffc/compiler/monomialintegration.py  1.6
  ffc  src/ffc/compiler/referencetensor.py      1.21
  ffc  src/reference/Elasticity.h               1.34
  ffc  src/reference/EnergyNorm.h               1.12
  ffc  src/reference/Mass.h                     1.20
  ffc  src/reference/NavierStokes.h             1.32
  ffc  src/reference/Projection.h               1.9
  ffc  src/reference/Stokes.h                   1.17