Re: [spam score 3.32/10.0 -pobox] [gtk-list] gdk_draw_* and friends
- From: Havoc Pennington <hp redhat com>
- To: gtk-list redhat com
- Cc: otaylor redhat com, Nicholas Lewin-Koh <kohnicho comp nus edu sg>, Deborah Swayne <dfs research att com>
- Subject: Re: [spam score 3.32/10.0 -pobox] [gtk-list] gdk_draw_* and friends
- Date: 07 Apr 2000 11:43:28 -0400
Deborah Swayne <dfs@research.att.com> writes:
> I've been interested all discussion about plural gdk_draw
> routines, since I'm used to using those in X, and I do a <lot> of
> drawing -- scatterplots of up to 100,000 points, which may be in
> motion. I was delighted to be convinced that they don't improve
> performance, because it simplifies my code to get rid of them.
>
BTW, you may want to consider client-side drawing into an RGB buffer
for this; it works well for Guppi (www.gnome.org/guppi).
Basically you can do all your drawing directly into an RGB buffer,
then ship the RGB buffer to the server. It can be faster than sending
tons of X protocol over the wire, depending on what you're doing.
(In general, if your end result can be expressed more concisely as a
grid of pixels than as a series of primitive shape-drawing requests,
then it may be faster to draw on the client side. The extreme is
rendering an image by gdk_draw_point() of each pixel, where an RGB
buffer definitely means a huge speedup. The other extreme is that
shipping an entire RGB buffer over the wire to draw a line, instead of
shipping the two line endpoints over the wire, means the RGB buffer is
a huge slowdown. Of course there's a big continuum in between the
extremes.)
Havoc
[
Date Prev][
Date Next] [
Thread Prev][
Thread Next]
[
Thread Index]
[
Date Index]
[
Author Index]