Filed under: iPhone
Speed test comparing iPhone 3G, 3GS, and Palm Pre has surprising results
Gadget fans can fight endlessly about which device is faster than which, without resolution or relief. Fortunately for the weary, Medialets has found common ground between a few of the most popular smart phones to use as a racetrack.
According to Medialets, there is a common benchmarkable technology -- JavaScript execution in a WebKit-based browser -- that runs across four of the most popular phones: the iPhone 3G, the iPhone 3GS, the T-Mobile G1 with Android, and the Palm Pre. By running a benchmark test called SunSpider it is possible to obtain a rough speed comparison between all four devices.
The test was run on six different configurations: Safari on the 3G with iPhone OS v2.2.1, the 3G with OS v3.0, and the 3G S with OS v3.0; "Browser" on the T-Mobile G1 with Android OS v1.5; and "Web" on the Palm Pre with Web OS v.10.2, with a run of Safari 4.0.1 on a MacBook used as a baseline. Read on for the graph of the results.
After putting the phones through their paces, here's how they measured up:
In rendering this type of script, the 3GS was three times as fast as the 3G (an improvement over Apple's estimate, who quoted the 3GS as being only twice as fast as the 3G). The 3G running OS 3.0 was also nearly three times faster than a 3G running OS 2.2.1. The 3G with OS 3.0 clocked almost exactly the same time as the Palm Pre, and both were about twice as fast as the G1. Not only does this test verify the speed improvements in the 3GS over the 3G, it puts to rest, at least to a certain extent, the idea that most of the 3G S speed bump is because OS 3.0 is designed for the 3GS model's hardware rather than the 3G's.
It is important to note that the test studies the core JavaScript language only, and not the DOM or the other browser APIs. More importantly, this test offers no comparison of how each phone and each OS runs its other native applications; also, Apple might be expected to have an edge in optimizing WebKit, as some of the open-source project's core team members work for Apple on Safari. However, when it comes to browsing the Internet and executing JavaScript among smart phones, there is a clear winner and a few losers.
In rendering this type of script, the 3GS was three times as fast as the 3G (an improvement over Apple's estimate, who quoted the 3GS as being only twice as fast as the 3G). The 3G running OS 3.0 was also nearly three times faster than a 3G running OS 2.2.1. The 3G with OS 3.0 clocked almost exactly the same time as the Palm Pre, and both were about twice as fast as the G1. Not only does this test verify the speed improvements in the 3GS over the 3G, it puts to rest, at least to a certain extent, the idea that most of the 3G S speed bump is because OS 3.0 is designed for the 3GS model's hardware rather than the 3G's.
It is important to note that the test studies the core JavaScript language only, and not the DOM or the other browser APIs. More importantly, this test offers no comparison of how each phone and each OS runs its other native applications; also, Apple might be expected to have an edge in optimizing WebKit, as some of the open-source project's core team members work for Apple on Safari. However, when it comes to browsing the Internet and executing JavaScript among smart phones, there is a clear winner and a few losers.
![TUAW [Cafepress]](http://www.blogsmithmedia.com/www.tuaw.com/media/tuaw-cafepress-promo.png)
Reader Comments (Page 1 of 2)
sami said 7:27AM on 6-25-2009
Any idea how Nokia's Symbian phones would compare toward these results...?
Reply
K said 7:32AM on 6-25-2009
Poorly.
Does that help? ;)
Seriously, my mate's E71 on a variety of browsers (Opera, built-in, Skyfire, etc.) is so slow compared to my current 3G (even before 3.0), and don't get me started on rendering issues, or even trying to browse easily.
Robert said 2:03PM on 6-25-2009
I tried it on my Nokia E71. Neither the built in browser or Opera Mini would do the test.
shinrajp said 3:05PM on 6-25-2009
Comparing to Nokia is a good reference, but if you just look at the CPU, the iPhone and Pre are going to WIN way way ahead.
Nokia E71 369 MHz ARM11
Nokia N97 Single CPU, 434 MHz ARM11
iPhone 3G 620 MHz downclock to 412 MHz ARM11
iPhone 3GS 3GS: ARM Cortex-A8 833 MHz underclocked to 600 MHz (Equivalent to 1.2GHz ARM11), PowerVR SGX GPU
Palm Pre: 600 Mhz Texas Instruments OMAP 3430 (ARM Cortex A8 + PowerVR SGX)\
G1: Qualcomm MSM7201A ARM11 @ 528MHz
Do we still need to compare the speed test even they are running in different OS. Obviously iPhone 3GS is the winner base on the specs and improved Java tech.
FTW iPhone 3GS!
Mark said 8:22AM on 6-25-2009
Just ran it on iPod Touch 2g with 3.0= 31.08sec
Reply
macuser said 9:03AM on 6-25-2009
Yah my iPod Touch 2G got 30.741 sec, so roughly the same. And my 13'' unibody MacBook got .7382 sec.
Nathan said 9:38AM on 6-25-2009
my 1g ipod touch on 3.0 got 43.488 seconds, that cant tbe right?
Ellis said 8:27AM on 6-25-2009
Doesnt the G1 and the iPhone use the same java tech except apple get a slightly faster version of it? I like all the new phones that have come out, just ditched my iphone for a magic to see how that does in comparison, so far i like it.
Reply
David V. said 10:46AM on 6-25-2009
I think the G1 still uses the previous JavaScript engine (not Nitro/SquirrelFish).
bravelittlememe said 9:03AM on 6-25-2009
How is this surprising? I thought it was common knowledge by now that the 3GS was faster than a speeding bullet.
Reply
DistortedLoop said 9:11AM on 6-25-2009
Benchmarks are fine on one level, but they only reflect conditions in the lab, and not necessarily the "real" "user experience" that any particular phone will provide.
That cheesy CNET comparison of the speeds of a 3G, 3GS, and Pre, which garnered a lot of criticism in the comments because of not using WiFi to compare Internet speeds, did point out one thing that benchmarks like this fail to: hardware aside, the network you use is what makes the experience.
So what if the iPhone 3GS can run java 3x as fast as anything else in the lab under controlled conditions? It won't do you any good if AT&T can't deliver the bandwidth. In the CNET tests, the Pre/Sprint combo accomplished several tasks quicker because in THAT particular location, Sprint delivered a better network connection.
It doesn't matter to me what the hardware CAN do on some test bench, it matters to me what the hardware ACTUALLY does when I am out and about and need to get something done. For some people in some locations, the Pre will beat the iPhone, in others the iPhone the Pre, AT&T will beat Sprint, in others vice/versa.
I was in Bullhead City, AZ over the weekend with my new 3GS and my laptop with a Sprint CDMA card. The Sprint could pick up a 3G signal from Sprint in areas the iPhone couldn't even get service. Here at home, though, the tethering via my iPhone to AT&T provided 2-3x the speed on speedtest.net tests vs my Sprint card on 3G.
My overall point is that it's easy to get all wrapped up in some bars on benchmark and forget that in the real world, they don't necessarily mean all that much.
Reply
tekkenshihan said 9:34AM on 6-25-2009
That's all nice and all, but what this benchmark is telling you is that when either phone ends up in it's ideal location with good service, the iPhone is always going to win because ultimately it is the fastest at rendering the script.
And since where I live and the places I travel, I always have 3G and almost full bars, that means my user experience is always going to be better than using the Pre.
I mean you can't really make sweeping generalizations about this stuff because a user's experience is so dependent on their location and their service. If AT&T has horrible service where you live, they get the Pre.
I mean come on, use your head.
DistortedLoop said 9:47AM on 6-25-2009
@tekkenshihan -
Use your head, dude That's exactly my point, the "experience" isn't something you can measure on a SINGLE benchmark (how fast javascript runs on the browser), it's a combination of many things. How many people are going to feel so justified they bought a 3GS because they don't understand this benchmark and what it does, and does not tell you?
If you want to crow on about how fast the iPhone 3GS is at running javascript, or if it will help you sleep better at night because it runs javascript faster, more power to you, but don't be deluded into thinking this means the 3GS is the best phone out there all things considered.
That said, perhaps you missed that I own the 3GS, so clearly I think it's the best choice out there, or I wouldn't have bought it, but I'm not going to get wrapped up in a single benchmark and try to convince friends and clients to get one based on that. I'm going to point out that it's big, very fast in real world use at all my tasks, and comes with AT&T service, which, in our area, is faster than Sprint since the recent AT&T tower upgrades were installed.
Reply
hmlong said 9:57AM on 6-25-2009
No one says to rely on a single benchmark, but the speed test is valid, nonetheless.
And your counter-arguments only apply to network-based applications, btw. Games and many other local or synced applications benefit just as well.
Heck, even network-based applications benefit. The 3G connection may lag, but if Safari can render the page 3X faster, you may not even notice the difference.
DistortedLoop said 10:09AM on 6-25-2009
@hmlong - Did I say the benchmark was invalid? I don't think so.
All I said was it only represented a small part of the overall user experience. I don't disagree with anything you're saying, just pointing out that it's easy to get wrapped up in some single benchmark that is just some small part of the experience.
As far as network-based arguments, go, isn't this benchmark something that tests something that will ONLY run over a network (javascript execution in the web browser)? Because of that, the network connection MUST be a consideration when considering/comparing real world user experience.
So yes, the iPhone 3GS will run javascript apps faster than the compared smartphones IN THE LAB, but that doesn't mean that when I'm sitting at a restaurant in downtown San Francisco that I'll be able to get what time the movie starts tonight faster than my friend with a Pre on Sprint can. Chances are he will be able to BECAUSE of the network connection.
If you want to talk about rendering 3D, running games, etc, that's a whole different point, and not relevant to the benchmark being discussed, unless of course it's a javascript based game that can be ran without a network connection.
Ese said 11:35AM on 6-25-2009
@ hmlong - It's valid why? Because Apple won? You all know good and damn well if the Pre had won, you all would be splittin wigs..
Blu-Sam said 10:03AM on 6-25-2009
My workplace on PowerMac G5 goes at 5.56 sec
But question is... does it make different between wi-fi and 3G/Ev???
I tested my Palm Pre which appear to be 49.54 sec over Ev.
Reply
Danny Zumwalt said 4:47PM on 6-25-2009
835.8ms +/- 2.1% on my iMac 2.8GHz Core 2 Extreme 4GB RAM
Reply
hansning said 11:56AM on 6-25-2009
my iphone 2g on 3.0 got 45.7s
Reply
kbotc said 12:07PM on 6-25-2009
Sprint's EV-DO has such a wide range compared to AT&T's HSPA network. I couldn't believe how much time I spent out of 3G when I was traveling after I switched from my Sprint MM-900A compared to the iPhone. I wouldn't give up the iPhone's features though.
Reply