iPhone 3GS Performance: A Significant Performance Bump
by Anand Lal Shimpi on June 19, 2009 12:00 AM EST- Posted in
- Smartphones
- Mobile
You may have noticed today's review of the Palm Pre. While I strongly suggest reading it if you're at all interested, I do have some new data for your consumption. After the announcement of the iPhone 3GS I posted an article discussing the hardware specs of the 3GS. I outlined the single chip CPU/GPU SoC as follows:
| iPhone 3G (ARM11) | iPhone 3GS (ARM Cortex A8) | |
| Manufacturing Process | 90nm | 65nm |
| Architecture | In-Order | In-Order |
| Issue Width | 1-issue | 2-issue |
| Pipeline Depth | 8-stage | 13-stage |
| Clock Speed | 412MHz | 600MHz |
| L1 Cache Size | 16KB I-Cache + 16KB D-Cache | 32KB I-Cache + 32KB D-Cache |
| L2 Cache Size | N/A | 256KB |
The iPhone 3GS uses an ARM Cortex A8 processor running at 600MHz, much like the Palm Pre. Many weren't confident that the 3GS used the new ARM A8 core instead of a higher clocked ARM11, so after waiting in line to grab a 3GS this morning I decided to run a few tests (the iPhone 3G tests were using OS 3.0 and the Pre tests used OS 1.0.2).
Update: Thanks to the many readers who have pointed out my incorrect wording of the 3GS' performance improvements. The tables in the article have been updated to reflect the correct percentages. I appreciate the corrections :)
| WiFi | Apple iPhone 3G | Apple iPhone 3GS | Palm Pre | T-Mobile G1 |
| anandtech.com | 16.3 s | 7.8 s | 8.2 s | 17.2 s |
| arstechnica.com | 17.7 s | 6.3 s | 7.8 s | 17.8 s |
| hothardware.com | 35.2 s | 14.7 s | 11.2 s | 24.4 s |
| pcper.com | 33.3 s | 15.0 s | 18.0 s | 34.0 s |
| digg.com | 34.3 s | 15.0 s | 22.1 s | 40.0 s |
| techreport.com | 24.1 s | 9.6 s | 9.0 s | 20.5 s |
| tomshardware.com | 21.4 s | 16.4 s | 13.8 s | 26.0 s |
| slashdot.org | 26.0 s | 10.0 s | 20.9 s | 46.0 s |
| facebook.com | 31.7 s | 13.5 s | 19.6 s | 37.7 s |
| iPhone 3GS Advantage over Palm Pre | 21% | |||
| iPhone 3GS Advantage over iPhone 3G | 122% |
The new 3GS renders web pages 128% faster, on average, than the old iPhone 3G. The 45% clock speed boost alone isn't enough to generate such a large performance increase, this is a new microarchitecture. Also, note that the 3GS' performance mimics that of the Palm Pre - another Cortex A8 based phone.
Not too surprising given the just-released nature of the Pre's webOS, the 3GS is actually able to render webpages slightly faster than the Pre in some cases. The overall performance advantage ends up being 22.6% in favor of the 3GS over the Pre.
Application launch time has also improved (I've updated the results with data from the iPhone OS 3.0):
| Application Launch Time in Seconds | Web Browser | Dialer | Google Maps | Camera | |
| Apple iPhone 3GS | 0.7 s | 0.7 s | 2.7 s | 2.8 s | 0.8 s |
| Apple iPhone 3G | 0.8 s | 1.2 s | 3.3 s | 3.9 s | 1.2 s |
| Palm Pre | 3.0 s | 1.5 s | 8.6 s | 4.4 s | 3.3 s |
| T-Mobile G1 | 5.4 s | 2.0 s | 4.4 s | 4.9 s | 2.0 s |
| iPhone 3GS vs. 3G Performance Advantage | 14% | 71.5% | 22.2% | 39.4% | 50% |
While the old iPhone 3G was no slouch, the 3GS is anywhere from 14 - 72% faster in basic application load times. It's the magic of a brand new CPU architecture.
I'll be working on the 3GS all weekend and hope to provide a more thorough look at CPU, GPU and battery life performance with the new hardware. Enjoy.
Facebook
Twitter
RSS
75 Comments
View All Comments
DerekWilson - Friday, June 19, 2009 - link
the advantage of the 3gs over the 3g is ~50% (things take half as long)the disadvantage of the 3g compared to the 3gs is ~100% (things take twice as long)
...
it is quite easy to manipulate statistics and we always try to frame things in the most real world, useful and non-misleading way possible.
statistics are a tool, not an end in themselves ... whenever someone tries to sell you on statistics without the empirical data you should be seriously skeptical.
DeepBlue1975 - Friday, June 19, 2009 - link
You are comparing speeds.If it was about a 1000kb site,
0.7" = 1428kb/s
1.7" = 588kb/s
Thus, the speed of the 3gs is 2.43x that of the original 3g.
This is a common mistake I see everywhere... The comparison is made backwards because he who makes the article forgets that when you are talking about time to compare speeds, the relationship is inverse, not direct.
That being said... I still need a phone with a real keyboard. I'd like to see how the next n97 and HTC touch pro2 do against all of these, specially in the OS and application response time department... not too interested in bandwidth measurements as I use the phone too little when it comes to browsing.
DeepBlue1975 - Friday, June 19, 2009 - link
Pardon me, forgot to say that, in this article, you are not actually measuring speed, but rather the time it takes for the newcomer to get things done when compared to the original one.In that respect, saying that the 3gs takes only 59% of the time of the original one is pretty correct... What is not correct, is to say that it's 59% faster.
Tristano - Friday, June 19, 2009 - link
Why is the first iPhone ignored in these speed comparisons? Millions of first-generation iPhone owners are contemplating the upgrade, and I have no idea how the first generation phone compares in speed to either the 3g or the 3gs.jimhsu - Friday, June 19, 2009 - link
"Faster" in itself is ambiguous. Takes "half the time", a 54% reduction in launch time, "twice as fast" ... there's got to be a better alternative.DerekWilson - Friday, June 19, 2009 - link
You know, I've always advocated for using "percent difference" ... It would look like this:abs(test1 - test2) / ( (test1+test2) / 2) )
Right now we tend to look at either "percent increase" or "percent decrease" which are relative to either the higher or lower value in the testing ...
These two options do definitely have their place, but all three are useful for different purposes.
Cacolaco - Friday, June 19, 2009 - link
100% reduction in time?That's the sort of alternative I'd be more than happy to pay for...
DerekWilson - Friday, June 19, 2009 - link
you have got to be tired. get some freaking rest already!But also, those are some pretty solid improvements ... very impressive stuff.
I'm hoping the G2 will bring some improvements to google's platform since I have t-mobile and don't want to break my contract ...
Cacolaco - Friday, June 19, 2009 - link
I can't understand why the Blackberry Storm isn't included in comparisons like this. There was a new update a couple of weeks ago, and, at least in terms of loading applications, the storm is just as fast as the 3G S, if not faster. I can say, for a fact, that the Storm is faster than the numbers listed for the 3G, Pre, and G1.I'm not saying the Storm is better. The iPhone has a LOT of things going for it, but the Storm is a popular phone on one of the nations largest carriers, and it performs well enough (now, if not when it was released) to be included in comparisons such as this.
bocafit - Sunday, January 10, 2010 - link
I think the comparison really comes down to what you're using the iPod for. It actually comes with a lot of business apps, which really wouldn't make a difference in the S.