[disclaimer]


This is a personal blog. The opinions expressed here represent my own and not those of any of my employers or customers.

Except if stated otherwise, all the code shared is reusable under a MIT/X11 licence. If a picture is missing a copyright notice, it's probably because I'm owning it.

Wednesday, January 13, 2010

Talk Teaser: Image processing with Mono.Simd

The facts:

Processing time using gdk_pixbuf: 431ms
Same method ported to Mono.Simd: 66ms
That means roughly 6.5 times faster !

Some explanations:
  • The gdk-pixbuf is an unoptimized standard gdk operation (gtk+ 2.18.1), but I don't think a lot of them are optimized either using mmx or SSEx for this platform (x86-64). Feel free to prove me wrong here.
  • Times are averaged.
  • Loading and saving times aren't taken into account here, but both are using gdk_pixbuf operations.
  • The Mono.Simd method acts on vanilla pixbufs, and results are plain old usable pixbufs, not some kind of memory buffer or whatnot.
  • The image attached to this post is not the result of the processing.
This is only a teaser for the short talk I'll be giving at FOSDEM on Sunday Feb 7. Be there if you want to learn more, see the code, or question my sanity for doing this in Mono and not directly as a gdk-pixbuf patch.
By at
Labels: , , , , ,

8 comments:

  1. AnonymousJanuary 13, 2010 at 4:33 PM

    Interesting.

    Do they record that stuff, as I won't be able to attend?

    ReplyDelete
  2. AnonymousJanuary 13, 2010 at 8:28 PM

    A record would be very nice.

    ReplyDelete
  3. AnonymousJanuary 15, 2010 at 12:01 AM

    Very interesting topic. Is there a chance that we can get a recording of your talk ?

    ReplyDelete
  4. StephaneJanuary 15, 2010 at 8:54 AM

    I really hope there won't be any recordings. But I'll publish code and notes afterwards.

    ReplyDelete
  5. UnknownFebruary 12, 2010 at 11:34 PM

    Hi Steph. Saw your talk at FOSDEM, was good. I've previously worked using a combo of c# and matlab for image processing during my PhD and was looking for good ways of correctly typing a byte array for an image. I've a friend who's using pointers for performing image processing operations so he can sequentially perform some local operations in an image.

    He's having a lot of trouble with correctly typing the image data. Do you have any recommendations? The source code would be seriously useful. :)

    ReplyDelete
  6. AnonymousFebruary 17, 2010 at 1:11 AM

    Any chance to see the examples?

    ReplyDelete
  7. Jai PillaiJuly 23, 2010 at 6:01 AM

    Hey
    Will the talk be available online somewhere? Please post the link, if you can get it.
    Thanks

    ReplyDelete
  8. IgorAugust 30, 2010 at 2:27 AM

    Looks like interesting... where can one read more about this? What was the idea behind?

    ReplyDelete

Subscribe to: Post Comments (Atom)