-----Forwarded Message-----
From: James Chaldecott james@swythan.co.uk To: m.hearn@signal.dra.hmg.gb Subject: Re: GDI question Date: 25 Feb 2003 09:32:48 +0000
Mike,
I'm a Win32 developer, and I take a look at WineHQ every so often and I noticed this thread. I don't know what Wine performance is like, but I thought you might be interested to know that some Windows graphics drivers (particularly Nvidia) are also very slow doing off-screen BitBlt operations under some circumstances.
BitBlt'ing between two memory DC's seems to get very slow if the "acceleration level" is set to "Full" (the default) in the driver. I'm guessing they are, in fact, trying to overoptimize and they assume that at least one of the DC's is a real device (i.e. already in video memory), although I don't really know. We have the problem with our software and found the easy way to overcome it was to drop the driver acceleration factor a notch! I also spotted a bug in Mozilla's Bugzilla DB, and did some profiling for them.
http://bugzilla.mozilla.org/show_bug.cgi?id=170272
Anyway, I thought you might be interested. I suppose it means that lots of the BitBlt's being done *are* being accelerated by the hardware, and this is a corner case were performance degrades.
James Chaldecott
P.S. Feel free to post this to the list if it's interesting, but I don't subscribe, ATM.
Well, I just tried a simple comparison of the Adobe SVG plugin under Wine and IE6 on Windows XP. Wine _is_ slower by a visible amount when doing repeated bitblts such as the plugin does on an animation I have, but it's certainly not something that'd actually affect real world usage. I don't think GDI performance is really an issue at present, most apps don't do these kind of repeated double-buffer switches (i assume flash doesn't actually use the gdi to render into its backbuffer?) so it only affects apps like photoshop and director/flash.
In particular when I run the SVG plugin, I can see a FIXME talking about a potential optimization, so really the issue is just a case of elbow grease for optimizations (esp in bitblt) rather than a rewrite I'd guess.