dolfin team mailing list archive

[Robert C Kirby <kirby@xxxxxxxxxxxx>]

> > 3. It actually matters that the default mode of FFC removes any
> > products with zeros in the tensor product. BLAS does not know about
> > zeros.
>
> This seems to be a severe limitation. For PDE systems with Lagrange
> elements (the typical case I guess), there will be lots of zeros. From
> previous discussions about FFC/Ferari, the conclusion was that
> skipping the zero elements was the dominant optimization for computing
> the element matrix.


This is exactly why I've been talking about inferring block structure 
inside the compiler and building a block 3x3 local stiffness matrix 
(each block might be different).  For the vector Laplacian, it's block 
diagonal, etc.  You can combine doing the set of all different blocks 
with level 3 BLAS (on each element) with getting to throw away all the 
coarse-grained zeros.

Rob