|Re: [eigen] generic unrollers|
[ Thread Index | Date Index | More lists.tuxfamily.org/eigen Archives ]
Meh. The various unrollers are a bit too different to unify. Some implement binary ops, some implement unary ops. Some need to know the row,col at each step (like visitor) while the other don't. On top of that there is another problem: currently a large part of the debug size is generated by the unrollers, so if we add another layer of abstraction, we basically double that size. So instead of doing a generic unroller I'll just rationalize the existing ones i.e. I'll start with Assign.h, adding partial unrolling (only rows, only cols) and rationalizing it; then I'll reuse that as a starting point for the other unrollers. On Monday 09 June 2008 08:50:47 Benoît Jacob wrote: > Since this is making the unrollers even more complicated, and since we now > have several different places in eigen with such unrollers (Assign.h, > Redux.h, Part.h, Visitor.h) I think it's now more than time to move to > generic unrollers. These would take as template argument a struct providing > the method to apply to each coeff, etc.
Description: This is a digitally signed message part.
|Mail converted by MHonArc 2.6.19+||http://listengine.tuxfamily.org/|