[Liboil] 2D convolution with Orc

[jcupitt at gmail.com]

Last mail, I tried the morphological operations in Orc as well. I get
about a 5x speedup:

        $ time vips --vips-novector im_dilate wtcmono.v wtc2.v 4way.mor
        real    0m0.727s
        user    0m0.730s
        sys     0m0.290s

        $ time vips im_dilate wtcmono.v wtc2.v 4way.mor
        real    0m0.453s
        user    0m0.150s
        sys     0m0.260s

This is for a 4-connected dilate of a 10,000 x 10,000 8-bit mono image
running on a 2 x Opteron 254 machine at 2.7 Ghz.

The morphological operators work really well with array tiling, the 5x
speedup stays even with rather large matrices. I've merged my Orc
hacking to vips trunk, so the next vips stable should include this
stuff.

Code here if anyone is curious:

  http://vips.svn.sourceforge.net/viewvc/vips/vips7/trunk/libvips/morphology/morphology.c?view=markup

John