It remains incredibly disappointing that Android still doesn't support SVG (including in 2.3), however I discount this article immediately as it comes from AppleInsider. It's like listening to Gruber talk about Android.
Though what's the deal with the caption on the SunSpider benchmark (the one where they don't note that smaller=better): They say that "iOS takes advantage of the GPU as well, so scores here do not reflect actual performance".
What? The GPU has zero relevance to base JavaScript performance, and that aside is idiocy.
The review is by Javascript library company Sencha. Who can talk about cross browser js compat than the company that has handful of JS libraries that are expected to work across browsers and devices?
The website on which the news was posted may have an Apple bias, but the reviewer doesn't.
Yeah, Sencha did this to get their name out there. Not really sure why they are necessary, though, given that ACID scores, HTML5 test results, and performance numbers are known by everyone.
I think they meant to say real-world performance, not actual performance. GPU acceleration affects real-world HTML5 animation techniques, especially on mobile devices. It depends on what's being measured. iOS has GPU acceleration support in animation effects, while Android has a JIT enabled JavaScript VM. (iOS restricts code execution which prevents JITs from running, at least in 4.x). The importance of these features greatly depends on the application being executed.
Software and hardware acceleration frameworks matter far more in mobile than on the desktop, where battery life, memory, and CPU tuning are crucial and painfully clear when you care about speed.
The bit about iOS barring JIT is technically incorrect. I've shot that one down before so I won't bother now, but when Apple gets around to adding a JIT, miraculously that overarching constraint will disappear.
However yes the GPU absolutely matters in some cases. They should print those benchmarks if their goal is making lemonade, however. They shouldn't just try to manufacture reality.
You've misunderstood the caption. They are not claiming that the GPU would alter the result of the benchmark, but rather that the fact that it doesn't limits it's usefulness as a comparison.
It's poor writing, but it's not such poor writing that something that begins "SunSpider JavaScript benchmarks only test the CPU..." can be interpreted to mean exactly the opposite without a pretty aggressive presumption of bias on the part of the reader.
If they have a problem with the benchmark (personally I think SunSpider is a garbage benchmark) they shouldn't use it.
They definitely shouldn't include it and then include a completely unsupported (and technically irrelevant) trump statement. That single caption says all that needs to be said about that article.
Though what's the deal with the caption on the SunSpider benchmark (the one where they don't note that smaller=better): They say that "iOS takes advantage of the GPU as well, so scores here do not reflect actual performance".
What? The GPU has zero relevance to base JavaScript performance, and that aside is idiocy.