Re: Data-dependent slowdown of NSBezierPath
Re: Data-dependent slowdown of NSBezierPath
- Subject: Re: Data-dependent slowdown of NSBezierPath
- From: Lance Bland <email@hidden>
- Date: Mon, 20 May 2002 15:39:11 -0400
On Monday, May 20, 2002, at 03:16 PM, John C. Randolph wrote:
>
> For reference, Display Postscript on Openstep on a 1.8 GHz Pentium IV
>
> is 4
>
> to 10 times faster than Quartz on a 450 MHz G4
1.8 GHz / 0.45 Ghz == 4. So, no wonder!
>
Oh, for crying out loud! Yes, the Bezier path rendering is slow when
>
you have *thousands* of intersections. This is because correct
>
rendering of a path with self-intersections using a translucent (or
>
pattern) color requires that you figure out where all the intersections
>
are before you rasterize them.
>
>
DPS didn't anti-alias, and it only dealt with alpha when compositing.
Our testing shows that Quartz is about the same or faster than DPS on
OpenStep when CPU frequency is factored in, and if the right performance
algorithms are used.
(and for anyone that cares ... Quartz is several times faster than
OpenStep DPS on Solaris).
>
I would also point out that DPS had a hard limit on the number of
>
segments a path could have, which Quartz 2-d doesn't.
Do you mean for user path construction only? or was the stack path limit
hard coded also?
So, in the balance I would say Quartz is good stuff. With that said, I
can't wait to see Quartz Extreme.
-lance
------------------------------------
Lance Bland
mailto:email@hidden
web charts at
http://www.vvi.com/products/chart
_______________________________________________
cocoa-dev mailing list | email@hidden
Help/Unsubscribe/Archives:
http://www.lists.apple.com/mailman/listinfo/cocoa-dev
Do not post admin requests to the list. They will be ignored.