Hey, you had a Coco too! I still have a Coco 3 in a box. Maybe I'll dig it out and run Linpack on it. How far behind a free feature phone do you think it is?
Why does the system information list 1 core? I thought the big thing about Tegra 2 was that it was a dual core in addition to improved graphics capability.
Quadrant only shows one core for some reason, but Tegra 2 is indeed dual core in this phone. Just one of the many things we're going to find out going forward is what's multithreaded and what isn't.
Your linpack binary doesn't appear to be parallelized - I'd expect linpack to scale well with the number of cores. Is your linpack written in Java? How about the version you run on the iPhone?
We're using Linpack for Android: greenecomputing.com/apps/linpack/
It's currently likely single threaded, but we're undergoing a learning process about what is and isn't mulithreaded in Android through Dalvik. We don't have a version of Linpack on the iPhone ;)
Out of curiosity I ran the sunspider test on mine. Running, 2.2.1, slight overclock to 800 MHz. Came in just behind the EVO. Guess I don't need to upgrade for awhile.
I'm having strong misgivings about putting the focus on performance when evaluating smartphones.
1- I never really think that my phone is slow. I often think it's lacking in other aspects though (ergonomics, functionnality, ruggedness, battery life..). I'd like whatever time your devoting to running so many perfs benchmarks devoted to these aspects instead.
2- Do you choose your car by how fast it can go ? your pizza by how fast it's delivered ? your girl friend by how fast she can run ? did I cover all stupid analogies ? Fast enough ?
3- we're no longer little boys... I'm confident (or is it resigned ?) about my -and my phone's- dick size. Numbers are nice and all, but useless numbers are just that: useless bragging.
As the owner of a phone that is extremely slow to the point of seriously impacting use (HTC Diamond) speed does mean something to me. More importantly for a first look like this though these tests can be run in a few minutes, probably less than an hour to run each of these 3-5 times. As opposed to something like battery life which takes several hours per run. Things like ergonomics and functionality also take days of use to form an opinion on. This is a very early preview, not a final review.
For all your examples, you are probably interested in what they will be like in two years. For phones and other computing devices extra speed is generally a means of future proofing. In other words, really fast now means it will probably still be acceptable by the time I'm 12 months into my phone contract. Plus, reviews generally cover things like battery life, screen quality, etc...
I think this is a good point in general. But better software requires more juice. Graphic intesive ones for sure. Just like general users on the PC don't need the latest and greatest, same would apply here. At least for now. The better the software is, the more addicted the smart phone uses will be.
"(ergonomics, functionnality, ruggedness, battery life..)." -- All of these exexpt functionality, should have been worked out in the mobile phone market over the last 20-30 years. We want to see functionality and juice. The rest is subjective.
Other than 3d games , better software doesn't mean more juice. Better software requires developers to pay more attention to detail and getting better educated about the features they use in their apps.
Especially important with multicore!! Badly written multicore apps can suck a lot more juice than a badly written app running on a single core.
Not as important when running on a vm but can still have a significant effect on battery life.
Android 2.2 does not have official support for dual core CPU's.. 2.3 does so I'm sure that this phone will be quite a bit fast in certain applications as soon as the 2.3 version is ready for it.
But the benchmark numbers are good to at least get a general picture regarding the phone's general performance.. sure, most of it is probably not necessary but it's cool nonetheless ;)
Are there going to be single-core A9 chips? It will be interesting to see if any version of Android prior to Honeycomb can make use of the second core.
http://www.arm.com/products/processors/cortex-a/co... Second paragraph: "Available as either a single core or configurable multicore processor, with both synthesizable or hard-macro implementations available." It seems it can use MPCore to even go up to "4 cores".
I thought the whole thing about the dual core was not to have two CPUs, but to have one logical core and one off-core for other specific tasks. To reduce power and speed up instruction processing. As long as it's faster, I don't care :)
Tegra 2 seems like an amazing piece of silicon (and so other Cortex-A9 chips), buy i will most probably buy nexus S next month. Pure google android FTW!
The Optimus 2X is impressive but nothing above and beyond what is capable with Android devices currently available. Then again the Tegra 2 could have a lot of overclocking potential as well and really blow away my Nexus S, but that remains to be seen. Also as some have pointed out the LG Optimus 2X is running Android 2.2 which doesn't officially support dual-core processors (why they put 2.2 on this device at launch is beyond me), while 2.3 does support dual-core which could really turn up the speed on the Optimus once the Optimus gets upgraded.
I'm a very verbal opponent of Quadrant as a benchmark as it does a terrible job of producing values that reflect real-world results. It's an easily manipulated (I got my Epic 4G on 2.2 to produce a score of 2597) and has real issues producing accurate I/O results on phones using the RFS file system.
I hope to see some newer benchmarks used in the future. BaseMark looks very promising, and I'd also recommend SmartBench 2010 as he's been working closely with the XDA Developer community to produce an excellent benchmark. I know that he's also working on dual-core support in SmartBench 2011.
Benchmark comparisons are only as accurate as the software we use to run the test. I'd be interested to see an article that went into comparing the accuracy of benchmark software. This could be accomplished benchmarking phones with different benchmark applications against one another using a number of different applications and looking for outliers in the benchmark scores.
We’ve updated our terms. By continuing to use the site and/or by logging into your account, you agree to the Site’s updated Terms of Use and Privacy Policy.
28 Comments
Back to Article
jeffbui - Thursday, January 27, 2011 - link
Those synthetic benchmarks are entertaining if only for nostalgia's sake. Compare some of those numbers to your first computer.jeffbui - Thursday, January 27, 2011 - link
Here's a link:http://www.roylongbottom.org.uk/linpack%20results....
Ushio01 - Thursday, January 27, 2011 - link
Oh god I was using a pentium mmx 233 till 2004 and now it's surpassed by a phone chip.DanNeely - Thursday, January 27, 2011 - link
I didn't realize any of these benchmarks had been ported to Motorola 6809 architecture. :)The Hardcard - Friday, January 28, 2011 - link
Hey, you had a Coco too! I still have a Coco 3 in a box. Maybe I'll dig it out and run Linpack on it. How far behind a free feature phone do you think it is?Zokudu - Thursday, January 27, 2011 - link
Why does the system information list 1 core? I thought the big thing about Tegra 2 was that it was a dual core in addition to improved graphics capability.Brian Klug - Thursday, January 27, 2011 - link
I'm glad someone else saw that too ;)Quadrant only shows one core for some reason, but Tegra 2 is indeed dual core in this phone. Just one of the many things we're going to find out going forward is what's multithreaded and what isn't.
-Brian
pookguy88 - Thursday, January 27, 2011 - link
what carrier/frequencies does it have? I want it for T-Mobile!!BryanC - Thursday, January 27, 2011 - link
Your linpack binary doesn't appear to be parallelized - I'd expect linpack to scale well with the number of cores.Is your linpack written in Java? How about the version you run on the iPhone?
Brian Klug - Thursday, January 27, 2011 - link
We're using Linpack for Android: greenecomputing.com/apps/linpack/It's currently likely single threaded, but we're undergoing a learning process about what is and isn't mulithreaded in Android through Dalvik. We don't have a version of Linpack on the iPhone ;)
-Brian
JonnyBlaze - Thursday, January 27, 2011 - link
Gets 41.667 in linpack. It is overclocked a little tho.notposting - Thursday, January 27, 2011 - link
Out of curiosity I ran the sunspider test on mine. Running, 2.2.1, slight overclock to 800 MHz. Came in just behind the EVO. Guess I don't need to upgrade for awhile.mwhk1983 - Thursday, January 27, 2011 - link
Does it have an LED Notifications light ??bplewis24 - Thursday, January 27, 2011 - link
I believe there is one. Click on the large photograph at the top of the article and it looks like there is one on the top left.Brandon
StormyParis - Thursday, January 27, 2011 - link
I'm having strong misgivings about putting the focus on performance when evaluating smartphones.1- I never really think that my phone is slow. I often think it's lacking in other aspects though (ergonomics, functionnality, ruggedness, battery life..). I'd like whatever time your devoting to running so many perfs benchmarks devoted to these aspects instead.
2- Do you choose your car by how fast it can go ? your pizza by how fast it's delivered ? your girl friend by how fast she can run ? did I cover all stupid analogies ? Fast enough ?
3- we're no longer little boys... I'm confident (or is it resigned ?) about my -and my phone's- dick size. Numbers are nice and all, but useless numbers are just that: useless bragging.
strikeback03 - Friday, January 28, 2011 - link
As the owner of a phone that is extremely slow to the point of seriously impacting use (HTC Diamond) speed does mean something to me. More importantly for a first look like this though these tests can be run in a few minutes, probably less than an hour to run each of these 3-5 times. As opposed to something like battery life which takes several hours per run. Things like ergonomics and functionality also take days of use to form an opinion on. This is a very early preview, not a final review.nafhan - Friday, January 28, 2011 - link
For all your examples, you are probably interested in what they will be like in two years.For phones and other computing devices extra speed is generally a means of future proofing. In other words, really fast now means it will probably still be acceptable by the time I'm 12 months into my phone contract.
Plus, reviews generally cover things like battery life, screen quality, etc...
bhougha10 - Friday, January 28, 2011 - link
I think this is a good point in general. But better software requires more juice. Graphic intesive ones for sure.Just like general users on the PC don't need the latest and greatest, same would apply here. At least for now.
The better the software is, the more addicted the smart phone uses will be.
"(ergonomics, functionnality, ruggedness, battery life..)." -- All of these exexpt functionality, should have been worked out in the mobile phone market over the last 20-30 years. We want to see functionality and juice. The rest is subjective.
alxx - Monday, January 31, 2011 - link
Get real.Other than 3d games , better software doesn't mean more juice.
Better software requires developers to pay more attention to detail
and getting better educated about the features they use in their apps.
Especially important with multicore!!
Badly written multicore apps can suck a lot more juice than a badly written app running on a single core.
Not as important when running on a vm but can still have a significant effect on battery life.
stureandre - Friday, January 28, 2011 - link
Android 2.2 does not have official support for dual core CPU's.. 2.3 does so I'm sure that this phone will be quite a bit fast in certain applications as soon as the 2.3 version is ready for it.But the benchmark numbers are good to at least get a general picture regarding the phone's general performance.. sure, most of it is probably not necessary but it's cool nonetheless ;)
KaarlisK - Friday, January 28, 2011 - link
It seems that soon enough netbook CPU's should start to appear in the benchmark graphs...strikeback03 - Friday, January 28, 2011 - link
Are there going to be single-core A9 chips? It will be interesting to see if any version of Android prior to Honeycomb can make use of the second core.vol7ron - Friday, January 28, 2011 - link
http://www.arm.com/products/processors/cortex-a/co...Second paragraph: "Available as either a single core or configurable multicore processor, with both synthesizable or hard-macro implementations available." It seems it can use MPCore to even go up to "4 cores".
I thought the whole thing about the dual core was not to have two CPUs, but to have one logical core and one off-core for other specific tasks. To reduce power and speed up instruction processing. As long as it's faster, I don't care :)
http://www.anandtech.com/Gallery/Album/926#4
Wow - I know the icons aren't exactly the same, but did they have to use the same color scheme as the iPhone?
bplewis24 - Monday, January 31, 2011 - link
Does Apple have a patent on icon color schemes now?alphacheez - Friday, January 28, 2011 - link
YieldsTotal: 47256.0ms +/- 2.0%
so I'm looking forward to an upgrade in the first half of this year probably
jasn - Sunday, January 30, 2011 - link
Tegra 2 seems like an amazing piece of silicon (and so other Cortex-A9 chips), buy i will most probably buy nexus S next month. Pure google android FTW!scottb75 - Tuesday, February 1, 2011 - link
On my overclocked Nexus S (1.4 GHz Trinity CM7 Kernal) I ran some of these benchies myself and scored the following:Sunspider 0.9.1 benchmark: 3147.4ms
Rightware BrowserMark: 57144
Linpack: 19.379 Mflops
Quadrant CPU Benchmark: 7357 (3136 overall score)
The Optimus 2X is impressive but nothing above and beyond what is capable with Android devices currently available. Then again the Tegra 2 could have a lot of overclocking potential as well and really blow away my Nexus S, but that remains to be seen. Also as some have pointed out the LG Optimus 2X is running Android 2.2 which doesn't officially support dual-core processors (why they put 2.2 on this device at launch is beyond me), while 2.3 does support dual-core which could really turn up the speed on the Optimus once the Optimus gets upgraded.
Electrofreak - Saturday, February 5, 2011 - link
I'm a very verbal opponent of Quadrant as a benchmark as it does a terrible job of producing values that reflect real-world results. It's an easily manipulated (I got my Epic 4G on 2.2 to produce a score of 2597) and has real issues producing accurate I/O results on phones using the RFS file system.I hope to see some newer benchmarks used in the future. BaseMark looks very promising, and I'd also recommend SmartBench 2010 as he's been working closely with the XDA Developer community to produce an excellent benchmark. I know that he's also working on dual-core support in SmartBench 2011.
Benchmark comparisons are only as accurate as the software we use to run the test. I'd be interested to see an article that went into comparing the accuracy of benchmark software. This could be accomplished benchmarking phones with different benchmark applications against one another using a number of different applications and looking for outliers in the benchmark scores.
Just a suggestion! :)