try running this simpler benchmark on the tablets

[Stephen Bannasch]

I'm curious how fast this Energy2D JavaScript engine implemented *just* as a benchmark without any graphics runs on tablets and
smartphones:

https://bug636096.bugzilla.mozilla.org/attachment.cgi?id=514432

Here are results from my MacBook Pro. 2.66 GHz Intel Core i&, system 10.6.7

Browsers that support JavaScript Typed Arrays:

FireFox V4: 100 steps: 2965ms (34 fps)
Webkit Nightly (84467) 100 steps: 4653ms (21 fps)
Chrome 10.0.648.127 100 steps: 2434ms (41 fps)

Interesting that after updating Minefield it slowed by 25%

Minefield 4.0pre23 100 steps: 2452ms (41 fps)
Minefield 6.0a1 (2011-04-21) 100 steps: 3114ms (32 fps)

Older browsers

FireFox 3.6.15 100 steps: 2758ms (36 fps)
Safari Version 5.0.4 (6533.20.27) 100 steps: 24420ms (4 fps)

Interesting that FireFox 3.6.15 (without Typed Arrays) is faster than FireFox v4 (with typed arrays).

[Paul Horwitz]

Also interesting that Safari 5.0.4 is ten times slower than the "best of show"!

Paul

--
----
post message :cc-dev...@googlegroups.com
unsubscribe: cc-developer...@googlegroups.com
more options: http://groups.google.com/group/cc-developers?hl=en

--
Dr. Paul Horwitz
The Concord Consortium
25 Love Lane
Concord, MA  01742
978-405-3223

[Noah Paessel]

Mobile safari on IOS 4.3.2 doesn't display a benchmark for me. It does
'sit there' for at least a 5 count after I push the button.
The error message is (heh heh) Error On Line 769: JavaScriptExecution
exceeded timeout.

- n

[Dan Damelin]

Tried to run on my iPod touch but only saw a "benchmark" button. When I pushed it, it stayed "pressed" for several seconds, but I saw no output indicating a measurement.
-Dan

[Noah Paessel]

For IOS people out there, you can turn on your javascript console in
safari by following these instructions:
http://maisonbisson.com/blog/post/14447/step-by-step-turn-on-the-iphone-ipads-web-debugging-console/

You will probably want to turn it off again, because it takes up
significant real-estate.

- noah

[Stephen Bannasch]

At 5:07 PM -0400 4/21/11, Dan Damelin wrote:
>Tried to run on my iPod touch but only saw a "benchmark" button. When I pushed it, it stayed "pressed" for several seconds, but I saw no output indicating a measurement.
>-Dan

On Safari it takes 24s to return a value (4 models steps per second * 100 model steps).

I think the iPad 2 runs it at about 1 model step per second -- so if it didn't time out it might take almost 2 minutes to return.

I suppose an iPod touch would be even slower.

[Scott Cytacki]

on a first generation motorola droid (android) I got:

164s  

I had given up but then went back to look. :)

Scott

[Sam Fentress]

At first glance, seeing the results in ms, I though I had almost the exact same result as Scott's, which made sense as we have the same phone. Then I realized there was an extra decimal place.

I get 1611s (about 26 minutes) on my Droid running Android's Browser 2.2

Sam

[Paul Horwitz]

That seems incredibly slow! Remind me again, what were the times in a browser? Weren't they less than a minute? No wonder those phones get so much battery life!

Paul

[Aaron Unger]

Droid 2, Android 2.2:

android browser: 79-80s.
dolphin hd: 78-80s.
firefox 4: 69-71s (skewed slightly, because it would ask every 20 seconds if I wanted to stop the script)

-- Aaron

[Stephen Bannasch]

>That seems incredibly slow! Remind me again, what were the times in a browser? Weren't they less than a minute? No wonder thosephones get so much battery life!

Paul,

In modern fast browsers on my MacBook Pro the benchmark takes about 3s.

Are you viewing this email thread with an application which let's you easily see the beginning of the conversation?

This is quoting from my first message:

[Sam Fentress]

I noticed that my phone had some updates pending, so I upgraded to 2.2, restarted, and ran the tests again. This time I got a much more reasonable 159 seconds.

It's also quite possible that the phone went to sleep while performing the test the previous time. I'm not sure whether Android will continue running js scripts when the phone is asleep.

Sam

On Fri, Apr 22, 2011 at 11:31 AM, Stephen Bannasch <stephen....@deanbrook.org> wrote:

>That seems incredibly slow! Remind me again, what were the times in a browser? Weren't they less than a minute? No wonder thosephones get so much battery life!

Paul,

In modern fast browsers on my MacBook Pro the benchmark takes about 3s.

Are you viewing this email thread with an application which let's you easily see the beginning of the conversation?

This is quoting from my first message:

At 4:44 PM -0400 4/21/11, Stephen Bannasch wrote:

>Browsers that support JavaScript Typed Arrays:
>
>FireFox V4:                         100 steps: 2965ms (34 fps)
>Webkit Nightly (84467)              100 steps: 4653ms (21 fps)
>Chrome 10.0.648.127                 100 steps: 2434ms (41 fps)
>
>Interesting that after updating Minefield it slowed by 25%
>
>Minefield 4.0pre23                  100 steps: 2452ms (41 fps)
>Minefield 6.0a1 (2011-04-21)        100 steps: 3114ms (32 fps)
>
>Older browsers
>
>FireFox 3.6.15                     100 steps: 2758ms (36 fps)
>Safari Version 5.0.4 (6533.20.27)   100 steps: 24420ms (4 fps)

[Paul Horwitz]

It's also quite possible that the phone went to sleep while performing the test the previous time. I'm not sure whether Android will continue running js scripts when the phone is asleep.

Ah, that sounds likely!

Paul

[Stephen Bannasch]

>I noticed that my phone had some updates pending, so I upgraded to 2.2, restarted, and ran the tests again. This time I got a much more reasonable 159 seconds.
>

>It's also quite possible that the phone went to sleep while performing the test the previous time. I'm not sure whether Androidwill continue running js scripts when the phone is asleep.

Adam and I just tested the new Xoom and the iPad2:

The iPad 2 never responded after clicking the "Benchmark" button.

Motorala Xoom
Android 3.01
Kernel 2.6.36.3
Dual Core Nvidea Tegra-2

we tested 5 different browsers the results are the time in seconds to complete 100 models steps

Opera Mobile: v11.00
41.096 seconds

webkit-based browsers:

Android Browser
47.212

Dolphin Browser HD v4.6.1
47.054

SkyFire: v4.0
46.975

FireFox: v4.00 for Android
58.569s

[Stephen Bannasch]

If you do decide to build the laullon branch (and already have thebrotherbard branch cloned) take a look at this thread on the
git mailing list to see how to deal with the fact that they use different repos for the same submodule:

http://www.spinics.net/lists/git/msg155344.html