Android 2.2 demolishes iOS4 in JavaScript benchmarks
feeds.arstechnica.com
feeds.arstechnica.com
People trying to beat them on spec never will because it's not the game they're playing (at least not right now). They're building and selling user experience; it's the sum of the parts that makes them great, not the quality or performance of individual parts.
Disclosure: Googler, huge Apple fanboy.
1. http://en.wikipedia.org/wiki/Lars_Bak_(computer_programmer)
This is incredibly untrue. The iPhone has dominated virtually every benchmark until very, very recently, owing to a more efficient CPU, and a vastly superior GPU (the Samsung Galaxy S is pretty much the only device that beats even the 3GS).
The iPhone is always talked up on the merits of its spec sheet, whether it's screen technology, case technology, GPU speed, browser speed, and on and on. This has always been its edge.
Whether or not Apple wins or loses on specs (cf. the "I don't care" bear), they're winning big on overall experience, mindshare and ecosystem. Arguments about who has the best GPU misses the point when you can only play the newest, hottest games (or whatever) on the iPhone.
It's sort of like the argument about the Wii vs. the X360 and PS3 or the DS vs. the PSP. Who cares what the specs? The apps are driving the platform.
Also, Apple has no qualms trumpeting better specs, but they frequently don't. The GPU, for instance, has been completely absent from iDevice discussion since the iPad. (It's part of the A4, but Apple doesn't tell you that.) They also failed to mention the RAM bump for the iPhone 4.
Current example: Resolution. They beat the competition, and they present it as a distinct experience benefit.
By and large, they neither strongly compete nor communicate numbers elsewhere. In fact they go so far as to obscure or hide numbers and instead focus exclusively on the quality and experience angles instead.
I'm not a fruit fanboy, but I do think that the way to play the numbers game is to show why they matter in ways that the layman can comprehend. On that front, if Google aren't using this advantage in the sunspiders tests to communicate on a user experience angle then they're missing the advantage of being ahead in the numbers game.
http://arstechnica.com/open-source/reviews/2010/07/android-2...
This particular problem is with the handset, not the OS, but it's basically the flagship device. Ugh.
Then the choice of where to install apps being left to developers.
Last, the choice to expose a method for shuffling apps around. Sure it's nice to have because of the default limited space for apps. However the correct solution isn't to burden users with this low-level crap, but to fix the design so this kind of thing is unnecessary.
If they insist on supporting a removable SD card then they should install apps wherever there is room, perhaps notifying the user that if they remove the SD card then they will no longer be able to run apps installed to it.
It seems like a small thing, but these kinds of hacks pile up on each other and soon your left with a Windows-esque system where users are burdened with system maintenance crap instead of the device being more like an appliance that they can just use.
If a user goes to install an app and the device says there's no room when they have a 16GB SD card, most would just throw up their hands and think that it's broken, not that they should contact the developer of the app to request that they enable installation to SD cards, or that they should change the install location or whatever.
Yeah, I'm not thrilled about that. They have a point that some apps types fundamentally won't work well when run from the SD card, but I'd prefer it to be an opt-out rather than opt-in process.
"One obvious downside of external application storage is that the software installed on the SD card will only be available when the card is mounted on the device. If you plug your phone into your computer and enable USB mass storage mode, for example, the software installed on the SD card will not be accessible. Due to this limitation, Google's developer documentation warns that long-running software like background services, live wallpapers, and widgets should not allow SD installation."
Can you imagine explaining this to end users? "Well, if you install to the SD card to get more storage, you can't use this app when your phone is plugged into your computer." This is just... stupid.
Plugged in is fine, it's only when the SD card is explicitly mounted as a USB drive that the apps are unavailable. Still mildly annoying, but not as annoying as the iPhone refusing to let you access its filesystem at all.
I would assume that most of the time would be wasted on rendering and downloading, and that javascript runs a fraction, even unmeasurable time.
From SunSpider website: "This includes tests to generate a tagcloud from JSON input, a 3D raytracer, cryptography tests, code decompression, and many more "
How often are this things really done in a browser, and are they really the bottleneck?
However, you raise a valid point: Mobile websites, the kinds most likely to be served to you via MobileSafari, will likely not have things like Tag clouds or 3D raytracers for the foreseeable future.
Maybe there should be a mobile edition of the JS benchmarks run?
And also, as much as speed is an issue, what about a benchmark for compliance? Safari/Webkit is still (to some sites) a red headed step child. The "it works in FF and IE" mentality applies there. My bank is an example of that. It has gotten better, but there are still plenty of web apps where webkit is unsupported.
I hope one of these days we will get to the point where web apps are standards complaint, rather than rendering engine complaint.
We've largely won the battle on earning recognition and support for Gecko, but now we have to do most of it over again for Webkit? What new and exciting engine will we be griping about in two years? Does this make sense?
To do this im rendering my scene to one canvas buffer, and then using that as the stencil to run a 3x3 convolution filter to produced a 'carved' look on a wood texture. I want the map editing to update the carved surface as quickly as possible, so yes javascript performance is a bottle neck.
ideally i want this to run on the iPad so its useful at the table as well as post game. The current implementation isn't fast enough on iPad so the browser kills the script.
Stuff like you are talking about isn't common at the moment, but its only going to become more viable and more prevalent. If a guy like me can knock together something like the above in a couple of days, imagine the possiblities for someone working full time on a real project.
http://www.pcworld.com/article/200041/google_android_22_froy...
"Today is one of those days that has my heart racing; we’ve just released the source code for Android 2.2."
...
"Posted by Tim Bray on 23 June 2010"
Regardless, your statement seems to contradict itself. If you can't install 2.2 on your phone, then it isn't really "released," is it?
It could become an advantage over iPhone if Google plans to boost web apps on the platform. Otherwise this benchmark is meaningless to the average user, in my opinion.
In fact, the iPad and iPhone 4 are very similar in hardware as well. The most significant difference is in RAM, where the iPhone has twice as much as the iPad.
Mentioning his or her experience on the iPad is completely valid.
That said, of course faster javascript is a benefit. Shaving even fractions of a second off web operations has huge benefits for users and results in much higher usage and happier customers.
When the performance bumps are imperceptible, it will be time to make a distinction. But on mobile devices that's still quite a ways out.
Its likely that the experience will be faster on the iPad when iOS 4 comes out for it in October.
“The high pixel density of the display is marvelous for reading text. Letterforms are, as you’d expect, very crisp. My biggest gripe about the Nexus One display is that certain colors are way over-saturated. All skin tones look very orange to me. Everyone gets that spray-on tan look. Reds, pinks, and especially oranges all go fluorescent. In short, I love the pixel density and brightness (and, so far, the battery life), but I do not like the color reproduction. I don’t know if that’s the nature of OLED, or if it’s specific to the Nexus One.“
— http://daringfireball.net/linked/2010/01/19/nexus-one-oled
Right now this information is cool (and an engineering triumph, yet another data point for my love of VMs and how they are the future of all modern computing), but put this in perspective: it only affects a very small number of users with an expensive and difficult-to-obtain handset. From the perspective of the average customer, it's pretty meaningless. Gruber probably won't respond until many phones have 2.2 and suddenly it's a public issue that "iPhone is slow."
If the end goal is enhancing user experience, why not benchmark against those metrics?
then pick ten web pages and collect the data.
then pick ten other common UI actions (not in the web browser) and collect that data too. there's some fudge here because the phones have different UIs so you won't necessarily be able to find identical actions but you could get a good idea.
it's entirely possible that the dev teams have instrumented builds that give them this info automatically. i would be surprised if they didn't, in fact.
I don't understand your argument - true shitty engineers can ruin a UI, as well as they can ruin a JS interpreter. The point of testing and benchmarking is to evaluate the relative "performance" in each of these domains.
Measuring page loads on an array of popular, representative websites, or measuring the latency between user interactions and some visible response are some examples.
Also, having a group of real people give subjective ratings I think carries more weight than people give it credit for.
Don't get me wrong, it's good that these sort of enhancements are made and that the platforms drive each other on in this area, but it's not actually going to sell any phones because the gap will never become big enough to become a genuine selling point.
Maybe this is coincidental, but I am not sure anymore...
On the right of flag we could add "merge with" and take a news.yc url and if enough people submitted the same url a mod could merge all the comments into the old story.
Had to take it with my iPhone 4, so it's not included in the picture. Before you ask, I'm a mobile developer.