From: John Navas on
Google's Android was expected to become the world's second most-used
smartphone OS by 2012, after the Symbian OS, but now accelerating sales
will help it get there this year, according to analysts.

MORE:
<http://www.computerworld.com/s/article/9180197/Android_growing_much_faster_than_expected_say_analysts>
From: Thomas T. Veldhouse on
In alt.cellular.verizon John Navas <spamfilter1(a)navasgroup.com> wrote:
> Google's Android was expected to become the world's second most-used
> smartphone OS by 2012, after the Symbian OS, but now accelerating sales
> will help it get there this year, according to analysts.
>
> MORE:
> <http://www.computerworld.com/s/article/9180197/Android_growing_much_faster_than_expected_say_analysts>

When Froyo (2.2) comes out with the JIT embedded in the operating system,
these huge platform phones are going to leap out of their shells with
additional capabilities. Android is on its ascendency and shows no signs of
peaking. The iPhone on the other hand may very well have peaked.

--
Thomas T. Veldhouse

Religion is a crutch, but that's okay... humanity is a cripple.
From: John Navas on
On 6 Aug 2010 15:50:22 GMT, in <8c2p9uFrtkU5(a)mid.individual.net>,
"Thomas T. Veldhouse" <veldy71(a)gmail.com> wrote:

>In alt.cellular.verizon John Navas <spamfilter1(a)navasgroup.com> wrote:
>> Google's Android was expected to become the world's second most-used
>> smartphone OS by 2012, after the Symbian OS, but now accelerating sales
>> will help it get there this year, according to analysts.
>>
>> MORE:
>> <http://www.computerworld.com/s/article/9180197/Android_growing_much_faster_than_expected_say_analysts>
>
>When Froyo (2.2) comes out with the JIT embedded in the operating system,
>these huge platform phones are going to leap out of their shells with
>additional capabilities. Android is on its ascendency and shows no signs of
>peaking. The iPhone on the other hand may very well have peaked.

JIT (Just In Time compilation) has advantages for certain types of
applications, but not for all types, and there is no free lunch -- the
tradeoff is increased startup delay, which can result in degraded
responsiveness, a big reason why Froyo (2.2) is questionable (IMnsHO)
on earlier, slower hardware.

--
John

"Assumption is the mother of all screw ups."
[Wethern�s Law of Suspended Judgement]
From: Thomas T. Veldhouse on
In alt.cellular.verizon John Navas <spamfilter1(a)navasgroup.com> wrote:
> On 6 Aug 2010 15:50:22 GMT, in <8c2p9uFrtkU5(a)mid.individual.net>,
> "Thomas T. Veldhouse" <veldy71(a)gmail.com> wrote:
>
>>In alt.cellular.verizon John Navas <spamfilter1(a)navasgroup.com> wrote:
>>> Google's Android was expected to become the world's second most-used
>>> smartphone OS by 2012, after the Symbian OS, but now accelerating sales
>>> will help it get there this year, according to analysts.
>>>
>>> MORE:
>>> <http://www.computerworld.com/s/article/9180197/Android_growing_much_faster_than_expected_say_analysts>
>>
>>When Froyo (2.2) comes out with the JIT embedded in the operating system,
>>these huge platform phones are going to leap out of their shells with
>>additional capabilities. Android is on its ascendency and shows no signs of
>>peaking. The iPhone on the other hand may very well have peaked.
>
> JIT (Just In Time compilation) has advantages for certain types of
> applications, but not for all types, and there is no free lunch -- the
> tradeoff is increased startup delay, which can result in degraded
> responsiveness, a big reason why Froyo (2.2) is questionable (IMnsHO)
> on earlier, slower hardware.

That goes without saying. It is also clear that any app that has code used
more than one time will benefit and that will just about all apps and
certainly the operating system itself. Look at the performance benchmarks for
the Nexus One compared to other Android based phones (still on 2.1) and it is
at least 4x higher (I have the link on my Facebook page somewhere ... I can
provide it if desired). In any event, in all but a few rare cases, JIT will
be a HUGE improvement (4x - 5x general performance improvement excluding GPU
based graphics). Look at the JRE and .NET and JIT is built in and that's that
.... it simply is a no brainer; just use it. The worst affect I can see is
extended load times for certain types of apps [but with the 1GHz processors
out these days, that will be barely noticable which is indeed the case as
experienced by a friend of mine who has the Nexus One and another who has the
EVO which was just updated to 2.2 yesterday].

--
Thomas T. Veldhouse

Religion is a crutch, but that's okay... humanity is a cripple.
From: John Navas on
On 6 Aug 2010 21:08:13 GMT, in <8c3bttFfmrU1(a)mid.individual.net>,
"Thomas T. Veldhouse" <veldy71(a)gmail.com> wrote:

>In alt.cellular.verizon John Navas <spamfilter1(a)navasgroup.com> wrote:

>> JIT (Just In Time compilation) has advantages for certain types of
>> applications, but not for all types, and there is no free lunch -- the
>> tradeoff is increased startup delay, which can result in degraded
>> responsiveness, a big reason why Froyo (2.2) is questionable (IMnsHO)
>> on earlier, slower hardware.
>
>That goes without saying. It is also clear that any app that has code used
>more than one time will benefit

It takes much more than a single repetition for JIT to pay off, part of
the reason more sophisticated engines only compile code loops that have
reached a certain repetition count.

>and that will just about all apps and

There are large sections of code that do not benefit from JIT, such as
linear code, and code that largely just glues together OS function
calls. I've profiled many apps that show less than 10% gain.

>certainly the operating system itself.

The OS is pre-optimized.

>Look at the performance benchmarks for
>the Nexus One compared to other Android based phones (still on 2.1) and it is
>at least 4x higher (I have the link on my Facebook page somewhere ... I can
>provide it if desired).

Please do -- there's no way to assess your statement without the source.
In general, such benchmarks tend to be selected to show the maximum
benefit, and aren't a substitute for proper code profiling.

>In any event, in all but a few rare cases, JIT will
>be a HUGE improvement (4x - 5x general performance improvement excluding GPU
>based graphics).

Net real-world gains rarely exceed 2-3x in my experience, and are
typically less. Most app code does not max out the processor.

>Look at the JRE and .NET and JIT is built in and that's that
>... it simply is a no brainer; just use it.

JRE has a sophisticated system that only applies JIT when there's a good
chance of performance gain, and even then it guesses wrong a fair
percentage of the time. .NET is worse. And both still suffer from
annoying startup delays. I'd call it anything but a no brainer.

Where JIT really pays off is in server applications that will be
executed repeatedly in a very performance sensitive environment.

>The worst affect I can see is
>extended load times for certain types of apps [but with the 1GHz processors
>out these days, that will be barely noticable which is indeed the case as
>experienced by a friend of mine who has the Nexus One and another who has the
>EVO which was just updated to 2.2 yesterday].

Even the 1 GHz machines have some annoying lag issues.

--
John

"Assumption is the mother of all screw ups."
[Wethern�s Law of Suspended Judgement]