I use these with caution on animated pages or large elements; I don't have links to hand, but have lots of anecdotal stories of performance degradation with CSS drop shadows.
Just a friendly word of (unbacked) warning to frontend devs :)
Games do pretty simple things on the GPU. Download textures, then transform them. They do a _lot_ of it, but conceptually the operations are not actually all that complicated or varied, by and large.
Browsers have this little thing they have to deal with called "text". Turns out that stuffing reasonable text into a texture and then scrolling it, say, is not something GPUs can do a good job of. This is why games just bitmap all their text and it looks like crap, by the way. I doubt you'd want text in a browser to look like it does in most games.
Browsers are working on making this faster, by the way. But it also turns out that not only are GPUs designed around stuff that games do (which is not the same set of stuff that browsers need done), but so are graphics drivers. The parts of them that Firefox 4 has been trying to use are clearly undertested, because they're ... rather buggy. Which is why there's now a driver blacklist (whitelist on Linux) and so forth...
Tesselation, volumetric shadows, depth of field, etc. Can you really say with a straight face that it's not doing anything complicated? And here we are struggling with drop-shadows. I know the problem domains are different, but I can't help but feel completely underwhelmed at the current state of the web when it comes to graphics. Somehow I think there's more to it than "fonts are hard."
It's not only that "fonts are hard" but that "the transformations people do to web pages are very different from the ones the GPU is good at", as I understand.
At which point we've reached the end of my second-hand graphics knowledge, unfortunately. I wish I could give you a better technical explanation here...
At palm, I didn't work on the 3d acceleration bits, but I worked with those that did. This explanation is pretty much spot on. In an environment where you don't already know your textures, 3d acceleration is a very difficult task. That's why only a select number of CSS transforms have been hardware accelerated to date.
I doubt that text is a difficult problem for the GPU. Aside from the software stack and font format problem, you should be able to render text as vectors and put some antialiasing on it.
One other thing. The expensive thing with shadows is not just the compositing, but recomputing the shadow itself. Any time something changes about the thing that's casting the shadow you have to rebuild the shadow. If the change is of a kind the GPU knows about (polygon moved or rotated or whatnot), it might be doable on the GPU. If it's something like "the width changed and the text layout has to be redone", you lose.
Because these calculations are done on software, not GPU. GPU programming is hard. Using the GPU requires extra-integration and extra-work that most developers are not willing to do.
They actually completely stopped development on IE between version 6 and 7. There was no IE dev team, beyond people patching bugs.
Obviously this did not prevent other people doing GPU-acceleration, but the leading player deliberately halting development will and did cause the market to stagnate.
That's not really relevant to the issue at hand, though. Bringing up Microsoft's misbehavior with IE 6 in this context is like bringing up the time a kid cut in front of you for lunch in third grade when you meet him in college.
Indeed, simply scrolling this demo page in Chrome 9 feels very laggy. Edit: OS X, late 2009 MBP, BUT with the BetterHN tab/iframe on top (other pages are smooth though).
Also in Safari scroll is laggy. Apparently webkit is redrawing the shadows every time. I actually made shadows very similar to these for one of my web apps but didn't end up using them because it made scrolling laggy.
Yep, Chrome is far better on Windows than OS X. So long as I was using OS X on my MBP I couldn't use Chrome and kept going back to FF - but since switching the machine to Win 7 I can't not use Chrome.
It may be specific to my MBP model (17" Core i7) which has the weirdo combination of Intel and Nvidia GPUs that are switched flakily in software at OSX's whim and Chrome has been known to trip the GPU switching - http://www.google.com/search?sourceid=chrome&ie=UTF-8... .
Scrolling was very jittery in general and page rendering/painting was also slower. Not like molasses slow but definitely not acceptable for a Core i7 machine.
With Win 7 which uses only one GPU, the Nvidia one, it is very smooth and fast as I said - definitely wouldn't want to go back.
I am running the 13" mbp and have not noticed anything out of the ordinary though now I am going to start using firefox for a few weeks and see how I feel.
I'm seeing an interesting effect on OS X Leopard with Safari 5. If I bring another program to the foreground on top of the Safari window, and then switch back to Safari, only the shadows and background color appear. It takes a second or so for the white boxes and text to appear, as well as the chrome around the top portion of the window.
This was repeatable several times at first, but now it's become intermittent. It takes me back to the days of tweaking settings in X-Windows, when there was a "saveunder" setting that would specify whether to retain the contents of a window in memory when it became obscured by another window.
This is true in my experience. CSS shadows are awesome, but use with care when you're animating! I see a huge drop in framerate in my CSS animation tool when I have a big fat drop shadow on an element.
I can attest to this. I had a page with four dropped shadows and one div that is animated via jQuery's slideDown/slideUp methods. Sliding up and down was painfully slow and ragged.
Just a friendly word of (unbacked) warning to frontend devs :)