Re: [Flashcoders] Even odds

2005-12-09 Thread Mattias Johansson 
Geeks ;)

2005/12/9, Mark Winterhalder <[EMAIL PROTECTED]>:
>
> OK, I ran some tests of my own. After repeating it a couple of times,
> with various outcomes, the trend seems to be that a) the cast is about
> 10% faster than the equality, and b) the bitwise and is about 10%
> faster than the modulo.
> I used MTASC to compile, FP8 under wine on Linux, and the ActionStep
> Debug Panel (ActionStep alpha 1 has been released! check it out:
> actionstep.org).
> I decided to go with triggering it onKeyDown, to avoid initialization
> of the player and to be easily able to run repeated tests.
>
> Here's the code:
>
> var bench:Object = {};
> bench.onKeyDown = function () {
> var repeats:Number = 50;
> var i:Number = repeats;
> var foo;
> var start:Number = getTimer();
> while( i-- ) {
> foo = ( Math.random() * 10 )|0;
> }
> var overhead:Number = getTimer() - start;
> trace( "overhead: " + overhead );
> i = repeats;
> start = getTimer();
> while( i-- ) {
> foo = ( Math.random() * 10 )|0 % 2 == 1;
> }
> trace( "modulo, equality: " + ( getTimer() - start - overhead ) );
> i = repeats;
> start = getTimer();
> while( i-- ) {
> foo = Boolean( ( Math.random() * 10 )|0 % 2 );
> }
> trace( "modulo, cast: " + ( getTimer() - start - overhead ) );
> i = repeats;
> start = getTimer();
> while( i-- ) {
> foo = ( Math.random() * 10 )|0 & 1 == 1;
> }
> trace( "and, equality: " + ( getTimer() - start - overhead ) );
> i = repeats;
> start = getTimer();
> while( i-- ) {
> foo = Boolean( ( Math.random() * 10 )|0 & 1 );
> }
> trace( "and, cast: " + ( getTimer() - start - overhead ) );
> }
> Key.addListener( bench );
>
> //mark
>
> --
> http://snafoo.org/
> jabber: [EMAIL PROTECTED]
> ___
> Flashcoders mailing list
> Flashcoders@chattyfig.figleaf.com
> http://chattyfig.figleaf.com/mailman/listinfo/flashcoders
>
___
Flashcoders mailing list
Flashcoders@chattyfig.figleaf.com
http://chattyfig.figleaf.com/mailman/listinfo/flashcoders

RE: [Flashcoders] Even odds

2005-12-09 Thread Scott Hyndman 
Seems my last email got swallowed by the void. Here it is:

Nice tests!

You're totally right. I had no idea that Macromedia did any optimization
during compilation. Very cool. It's also pretty interesting that Java
did not seem to optimize the very same code (the numbers seem to suggest
it).

haha, thanks for the plug.

Scott 

-Original Message-
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On Behalf Of Mark
Winterhalder
Sent: December 9, 2005 12:09 AM
To: Flashcoders mailing list
Subject: Re: [Flashcoders] Even odds

OK, I ran some tests of my own. After repeating it a couple of times,
with various outcomes, the trend seems to be that a) the cast is about
10% faster than the equality, and b) the bitwise and is about 10% faster
than the modulo.
I used MTASC to compile, FP8 under wine on Linux, and the ActionStep
Debug Panel (ActionStep alpha 1 has been released! check it out:
actionstep.org).
I decided to go with triggering it onKeyDown, to avoid initialization of
the player and to be easily able to run repeated tests.

Here's the code:

var bench:Object = {};
bench.onKeyDown = function () {
var repeats:Number = 50;
var i:Number = repeats;
var foo;
var start:Number = getTimer();
while( i-- ) {
foo = ( Math.random() * 10 )|0;
}
var overhead:Number = getTimer() - start;
trace( "overhead: " + overhead );
i = repeats;
start = getTimer();
while( i-- ) {
foo = ( Math.random() * 10 )|0 % 2 == 1;
}
trace( "modulo, equality: " + ( getTimer() - start - overhead )
);
i = repeats;
start = getTimer();
while( i-- ) {
foo = Boolean( ( Math.random() * 10 )|0 % 2 );
}
trace( "modulo, cast: " + ( getTimer() - start - overhead ) );
i = repeats;
start = getTimer();
while( i-- ) {
foo = ( Math.random() * 10 )|0 & 1 == 1;
}
trace( "and, equality: " + ( getTimer() - start - overhead ) );
i = repeats;
start = getTimer();
while( i-- ) {
foo = Boolean( ( Math.random() * 10 )|0 & 1 );
}
trace( "and, cast: " + ( getTimer() - start - overhead ) ); }
Key.addListener( bench );

//mark

--
http://snafoo.org/
jabber: [EMAIL PROTECTED]
___
Flashcoders mailing list
Flashcoders@chattyfig.figleaf.com
http://chattyfig.figleaf.com/mailman/listinfo/flashcoders
___
Flashcoders mailing list
Flashcoders@chattyfig.figleaf.com
http://chattyfig.figleaf.com/mailman/listinfo/flashcoders

RE: [Flashcoders] Even odds

2005-12-09 Thread Scott Hyndman 
Nice tests!

You're totally right. I had no idea that Macromedia did any optimization during 
compilation. Very cool. It's also pretty interesting that Java did not seem to 
optimize the very same code (the numbers seem to suggest it).

haha, thanks for the plug.

Scott

-Original Message-
From:   [EMAIL PROTECTED] on behalf of Mark Winterhalder
Sent:   Fri 12/9/2005 12:09 AM
To: Flashcoders mailing list
Cc: 
Subject:    Re: [Flashcoders] Even odds

OK, I ran some tests of my own. After repeating it a couple of times,
with various outcomes, the trend seems to be that a) the cast is about
10% faster than the equality, and b) the bitwise and is about 10%
faster than the modulo.
I used MTASC to compile, FP8 under wine on Linux, and the ActionStep
Debug Panel (ActionStep alpha 1 has been released! check it out:
actionstep.org).
I decided to go with triggering it onKeyDown, to avoid initialization
of the player and to be easily able to run repeated tests.

Here's the code:

var bench:Object = {};
bench.onKeyDown = function () {
var repeats:Number = 50;
var i:Number = repeats;
var foo;
var start:Number = getTimer();
while( i-- ) {
foo = ( Math.random() * 10 )|0;
}
var overhead:Number = getTimer() - start;
trace( "overhead: " + overhead );
i = repeats;
start = getTimer();
while( i-- ) {
foo = ( Math.random() * 10 )|0 % 2 == 1;
}
trace( "modulo, equality: " + ( getTimer() - start - overhead ) );
i = repeats;
start = getTimer();
while( i-- ) {
foo = Boolean( ( Math.random() * 10 )|0 % 2 );
}
trace( "modulo, cast: " + ( getTimer() - start - overhead ) );
i = repeats;
start = getTimer();
while( i-- ) {
foo = ( Math.random() * 10 )|0 & 1 == 1;
}
trace( "and, equality: " + ( getTimer() - start - overhead ) );
i = repeats;
start = getTimer();
while( i-- ) {
foo = Boolean( ( Math.random() * 10 )|0 & 1 );
}
trace( "and, cast: " + ( getTimer() - start - overhead ) );
}
Key.addListener( bench );

//mark

--
http://snafoo.org/
jabber: [EMAIL PROTECTED]
___
Flashcoders mailing list
Flashcoders@chattyfig.figleaf.com
http://chattyfig.figleaf.com/mailman/listinfo/flashcoders



___
Flashcoders mailing list
Flashcoders@chattyfig.figleaf.com
http://chattyfig.figleaf.com/mailman/listinfo/flashcoders

Re: [Flashcoders] Even odds

2005-12-08 Thread Mark Winterhalder 
OK, I ran some tests of my own. After repeating it a couple of times,
with various outcomes, the trend seems to be that a) the cast is about
10% faster than the equality, and b) the bitwise and is about 10%
faster than the modulo.
I used MTASC to compile, FP8 under wine on Linux, and the ActionStep
Debug Panel (ActionStep alpha 1 has been released! check it out:
actionstep.org).
I decided to go with triggering it onKeyDown, to avoid initialization
of the player and to be easily able to run repeated tests.

Here's the code:

var bench:Object = {};
bench.onKeyDown = function () {
var repeats:Number = 50;
var i:Number = repeats;
var foo;
var start:Number = getTimer();
while( i-- ) {
foo = ( Math.random() * 10 )|0;
}
var overhead:Number = getTimer() - start;
trace( "overhead: " + overhead );
i = repeats;
start = getTimer();
while( i-- ) {
foo = ( Math.random() * 10 )|0 % 2 == 1;
}
trace( "modulo, equality: " + ( getTimer() - start - overhead ) );
i = repeats;
start = getTimer();
while( i-- ) {
foo = Boolean( ( Math.random() * 10 )|0 % 2 );
}
trace( "modulo, cast: " + ( getTimer() - start - overhead ) );
i = repeats;
start = getTimer();
while( i-- ) {
foo = ( Math.random() * 10 )|0 & 1 == 1;
}
trace( "and, equality: " + ( getTimer() - start - overhead ) );
i = repeats;
start = getTimer();
while( i-- ) {
foo = Boolean( ( Math.random() * 10 )|0 & 1 );
}
trace( "and, cast: " + ( getTimer() - start - overhead ) );
}
Key.addListener( bench );

//mark

--
http://snafoo.org/
jabber: [EMAIL PROTECTED]
___
Flashcoders mailing list
Flashcoders@chattyfig.figleaf.com
http://chattyfig.figleaf.com/mailman/listinfo/flashcoders

Re: [Flashcoders] Even odds

2005-12-08 Thread Mark Winterhalder 
On 12/9/05, Scott Hyndman <[EMAIL PROTECTED]> wrote:
> I just did a test. The bitwise approach is not faster...although when you 
> think about it, it really should be.

Hi Scott,

the test won't work because (50010001 % 2 == 1) always resolves to
false and the compiler will optimize it (at least MMC, not sure about
MTASC). basically, the same is true for Boolean(50010001 & 1), but
maybe it just optimizes the (50010001 & 1) part because Boolean might
be a custom replacement for the intrinsic class?

mark


> Here is the test code:
>
> // Modulo - Time ~225ms
> var time:Number = getTimer();
> for (var i:Number = 0; i < 8; i++) {
> var isOdd:Boolean = 50010001 % 2 == 1;
> }
> trace(getTimer() - time);
>
> // Bitwise And - Time: ~356ms
> time = getTimer();
> for (var i:Number = 0; i < 8; i++) {
> var isOdd:Boolean = Boolean(50010001 & 1);
> }
> trace(getTimer() - time);
>
> I also did a quick test in Java. The bitwise AND approach is about 50% 
> faster...really makes you wonder. Maybe AS3 is an improvement.
>
> Scott
>
> -Original Message-
> From:   [EMAIL PROTECTED] on behalf of Boon Chew
> Sent:   Thu 12/8/2005 9:34 PM
> To: Flashcoders mailing list
> Cc:
> Subject:Re: [Flashcoders] Even odds
>
> Yet another way (and probably the fastest):
>
> var isOdd:Boolean = Boolean(num & 1);
>
>   - boon
>
> Mike Boutin <[EMAIL PROTECTED]> wrote:  Is it possible to check a number to 
> see if its even / odd?
>
>
> Thanks!
> ___
> Flashcoders mailing list
> Flashcoders@chattyfig.figleaf.com
> http://chattyfig.figleaf.com/mailman/listinfo/flashcoders
>
>
>
>
> -
> Yahoo! Shopping
>  Find Great Deals on Holiday Gifts at Yahoo! Shopping
> ___
> Flashcoders mailing list
> Flashcoders@chattyfig.figleaf.com
> http://chattyfig.figleaf.com/mailman/listinfo/flashcoders
>
>
>
>
>
> ___
> Flashcoders mailing list
> Flashcoders@chattyfig.figleaf.com
> http://chattyfig.figleaf.com/mailman/listinfo/flashcoders
>
>
>


--
http://snafoo.org/
jabber: [EMAIL PROTECTED]
___
Flashcoders mailing list
Flashcoders@chattyfig.figleaf.com
http://chattyfig.figleaf.com/mailman/listinfo/flashcoders

RE: [Flashcoders] Even odds

2005-12-08 Thread Scott Hyndman 
I just did a test. The bitwise approach is not faster...although when you think 
about it, it really should be. 

Here is the test code:

// Modulo - Time ~225ms
var time:Number = getTimer();
for (var i:Number = 0; i < 8; i++) {
var isOdd:Boolean = 50010001 % 2 == 1;
}
trace(getTimer() - time);

// Bitwise And - Time: ~356ms
time = getTimer();
for (var i:Number = 0; i < 8; i++) {
var isOdd:Boolean = Boolean(50010001 & 1);
}
trace(getTimer() - time);

I also did a quick test in Java. The bitwise AND approach is about 50% 
faster...really makes you wonder. Maybe AS3 is an improvement.

Scott

-Original Message-
From:   [EMAIL PROTECTED] on behalf of Boon Chew
Sent:   Thu 12/8/2005 9:34 PM
To: Flashcoders mailing list
Cc: 
Subject:    Re: [Flashcoders] Even odds

Yet another way (and probably the fastest):
  
var isOdd:Boolean = Boolean(num & 1);
  
  - boon

Mike Boutin <[EMAIL PROTECTED]> wrote:  Is it possible to check a number to see 
if its even / odd?


Thanks!
___
Flashcoders mailing list
Flashcoders@chattyfig.figleaf.com
http://chattyfig.figleaf.com/mailman/listinfo/flashcoders




-
Yahoo! Shopping
 Find Great Deals on Holiday Gifts at Yahoo! Shopping 
___
Flashcoders mailing list
Flashcoders@chattyfig.figleaf.com
http://chattyfig.figleaf.com/mailman/listinfo/flashcoders



___
Flashcoders mailing list
Flashcoders@chattyfig.figleaf.com
http://chattyfig.figleaf.com/mailman/listinfo/flashcoders

Re: [Flashcoders] Even odds

2005-12-08 Thread Boon Chew 
Yet another way (and probably the fastest):
  
var isOdd:Boolean = Boolean(num & 1);
  
  - boon

Mike Boutin <[EMAIL PROTECTED]> wrote:  Is it possible to check a number to see 
if its even / odd?


Thanks!
___
Flashcoders mailing list
Flashcoders@chattyfig.figleaf.com
http://chattyfig.figleaf.com/mailman/listinfo/flashcoders




-
Yahoo! Shopping
 Find Great Deals on Holiday Gifts at Yahoo! Shopping 
___
Flashcoders mailing list
Flashcoders@chattyfig.figleaf.com
http://chattyfig.figleaf.com/mailman/listinfo/flashcoders

Re: [Flashcoders] Even odds

2005-12-08 Thread Mike Boutin 

Awesome, thank you both


Scott Hyndman wrote:


var isOdd:Boolean = number % 2 == 1;

-Original Message-
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On Behalf Of Mike
Boutin
Sent: December 8, 2005 7:40 PM
To: Flashcoders mailing list
Subject: [Flashcoders] Even odds

Is it possible to check a number to see if its even / odd?


Thanks!
___
Flashcoders mailing list
Flashcoders@chattyfig.figleaf.com
http://chattyfig.figleaf.com/mailman/listinfo/flashcoders
___
Flashcoders mailing list
Flashcoders@chattyfig.figleaf.com
http://chattyfig.figleaf.com/mailman/listinfo/flashcoders

 



___
Flashcoders mailing list
Flashcoders@chattyfig.figleaf.com
http://chattyfig.figleaf.com/mailman/listinfo/flashcoders

RE: [Flashcoders] Even odds

2005-12-08 Thread Scott Hyndman 
var isOdd:Boolean = number % 2 == 1;

-Original Message-
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On Behalf Of Mike
Boutin
Sent: December 8, 2005 7:40 PM
To: Flashcoders mailing list
Subject: [Flashcoders] Even odds

Is it possible to check a number to see if its even / odd?


Thanks!
___
Flashcoders mailing list
Flashcoders@chattyfig.figleaf.com
http://chattyfig.figleaf.com/mailman/listinfo/flashcoders
___
Flashcoders mailing list
Flashcoders@chattyfig.figleaf.com
http://chattyfig.figleaf.com/mailman/listinfo/flashcoders

RE: [Flashcoders] Even odds

2005-12-08 Thread Ralph Caraveo 
Use your friend the modulus operater:

If( I % 2 == 0)
{
//even number code
}
Else
{
//odd number code
} 

This operater returns the remainder of a numbers division.

(stupid email app is trying to capitilize my code)

-Original Message-
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On Behalf Of Mike
Boutin
Sent: Thursday, December 08, 2005 4:40 PM
To: Flashcoders mailing list
Subject: [Flashcoders] Even odds

Is it possible to check a number to see if its even / odd?


Thanks!
___
Flashcoders mailing list
Flashcoders@chattyfig.figleaf.com
http://chattyfig.figleaf.com/mailman/listinfo/flashcoders


___
Flashcoders mailing list
Flashcoders@chattyfig.figleaf.com
http://chattyfig.figleaf.com/mailman/listinfo/flashcoders