5 ms·
Going to http://ie.microsoft.com/testdrive/ http://ie.microsoft.com/testdrive/ on Chrome (dev-channel) shows a little banner with the following message, "Cool,
by prashantv 15y ago
Going to http://ie.microsoft.com/testdrive/ http://ie.microsoft.com/testdrive/ on Chrome (dev-channel) shows a little banner with the following message,
"Cool, you're using a Chrome 12 nightly build!
Don't forget to enable your partial hardware acceleration in the about:flags thingy..."
Seems like they're actually worried about Chrome, since Firefox and Safari don't have any similar messages.
- sid0 15y agoFirefox already has what they call "full" hardware acceleration, and Safari has none I think, so there's nothing to show there.
- fleitz 15y agoIMHO Safari actually has the best hardware acceleration. Chrome-dev channel is catching up quickly. By hardware acceleration I'm talking about things like -webkit-transform and the support for various 3D transforms. I'm unsure of how it's implemented under the hood but I know Safari kills on 3D transforms and the best part is it works on mobile.
- sid0 15y agoMere support for CSS transforms is actually really limited. What IE9 and Firefox 4 do on Windows 7 is accelerate all drawing and layer compositing. What it means is that as the browsers gain support for CSS animations and such, they get accelerated for free. Stuff like Canvas and SVG also get accelerated. See http://hacks.mozilla.org/2010/09/hardware-acceleration/ http://hacks.mozilla.org/2010/09/hardware-acceleration/.
- kogir 15y agoI think they also want any impromptu benchmarks people run to be as fair as possible. To the best of my knowledge, the latest FF and Safari both have all their supported hardware acceleration turned on by default. I've met some members of the IE tech team (who don't control the marketing). The people I know really want to win without cheating and support as much as possible, but are very wary of releasing anything too early. They're very careful to wait until the standards are pretty precise and stable before releasing an implementation, lest they be accused of trying to "sabotage" something with an unintentionally different (but still within the vague spec) implementation.