Change particle affector to work on N particles
This should severely cut down on the virtual call overhead when running
affectors. It also switches to model of "for each affector, do each
particle", rather than "for each particle, do each affector". This is
going to be an I$ improvement, and (if we ever care), make it possible
to do wide-mode threading of the affectors on disjoint blocks of
particles.
Bug: skia:
Change-Id: Ia4e8e8d461dba4e303c45766ea2b4ece1bca9196
Reviewed-on: https://skia-review.googlesource.com/c/195123
Commit-Queue: Brian Osman <brianosman@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
3 files changed