[cairo] Performance of the image surface back-end

[Nicholas Allen]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
 
Oops - another mistake. For some reason make didn't recompile my
changes. After making the changes I now get 4868 ms which is getting
better. However, this is still 25 times slower than our code so still
indicates some serious performance problems.

Nick


Nicholas Allen wrote:
> Hi,
>
> I have changed my line drawing code so that if the line is
> horizontal or vertical it will use cairo_rectangle and cairo_fill
> to fill it. This avoids the problem of line caps altogether and is
> guaranteed to be pixel aligned. It seems the line capping code has
> absolutely nothing to do with the performance problems of cairo.
> Using this new approach I achieve almost identical performance as
> previously. It takes 15476 ms to draw the horizontal lines compared
> with 15839 ms when drawing the end caps on a non pixel aligned
> line. This is basically no change. To put it in perspective using
> our simple graphics API it takes 192 ms to to the same thing which
> means we are 81 X faster than Cairo!
>
> The image surface back-end is so slow as to be completely unusable.
>  How can it possibly be this slow?
>
> Nick
>
>
> My line drawing code looks like this:
>
> void ACairoGraphPort::DrawLine(TPoint P0, TPoint P1, TColor Color,
> TInt Thickness) { SetColor(Color);
>
> if (P0 == P1) { cairo_rectangle(mpCairo, P0.X(), P0.Y(), Thickness,
> Thickness); cairo_fill(mpCairo); } else if (P0.X() == P1.X())  //
> Vertical line { if (P0.Y() < P1.Y()) { cairo_rectangle(mpCairo,
> P0.X(), P0.Y(), Thickness, P1.Y() - P0.Y()); } else {
> cairo_rectangle(mpCairo, P1.X(), P1.Y(), Thickness, P0.Y() -
> P1.Y()); }
>
> cairo_fill(mpCairo); } else if (P0.Y() == P1.Y())  // Horizontal
> line { if (P0.X() < P1.X()) { cairo_rectangle(mpCairo, P0.X(),
> P0.Y(), P1.X() - P0.X(), Thickness); } else {
> cairo_rectangle(mpCairo, P1.X(), P1.Y(), P0.X() - P1.X(),
> Thickness); }
>
> cairo_fill(mpCairo); } else  // Diagonal line {
> cairo_set_line_width(mpCairo, Thickness); cairo_move_to(mpCairo,
> P0.X(), P0.Y()); cairo_line_to(mpCairo, P1.X(), P1.Y());
> cairo_stroke(mpCairo); } }
>
> Vladimir Vukicevic wrote:
>> On Aug 15, 2008, at 6:20 AM, Nicholas Allen wrote:
>>> Cairo (image surface back-end version 1.6.4 on Windows):
>>>
>>> ~     DrawHorizLine                 44352 ms
>>> 1000000
> solid
>>> horiz lines from 0.5, 50.5 to 500.5, 50.5
>> This is not actually a pixel-aligned horizontal line, unless you
>> are using the non-default SQUARE endcaps.  The horizontal line
>> would be from 0.0, 50.5 to 501.0, 50.5.  With this line, you are
>> getting essentially the slowest possible path for path
>> rasterization; however, even with that, it seems quite slow.  As
>> others mentioned, seeing the source for your benchmark would be
>> useful!  (If, however, you are explicitly setting SQUARE endcaps,
>> then something is likely going quite wrong.)
>
>> - Vlad
>
>

_______________________________________________
cairo mailing list
cairo at cairographics.org
http://lists.cairographics.org/mailman/listinfo/cairo

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.7 (MingW32)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org
 
iD8DBQFIqWEIbpmWsXfOU58RAm80AKCxt7kPx0NpvhXdIc8eJgMKBtsN6QCbBYQ0
6OxfVAAzflGMqakvHI8CT3o=
=TUHu
-----END PGP SIGNATURE-----