Mini-putt for a slow day on korg ... - KiXtart.org

You are not logged in. [Log In] KiXtart.org website » Forums » KiXtart » Lounge » Mini-putt for a slow day on korg ...

Page 2 of 3

Topic Options

#106508 - 2003-11-03 04:32 PM Re: Mini-putt for a slow day on korg ...
Shawn Shawn Administrator Registered: 1999-08-13 Posts: 8611	yip - you got it ... I can see where the guys went astray though ... when I mention a number like &00DDEE i am just using the & to denote a hex number in kix, kinda like saying 0x00DDEE in other langauges - ultimately its all decimal ... got yah now.
Top

#106509 - 2003-11-03 04:38 PM

Re: Mini-putt for a slow day on korg ...

Howard Bullock Offline

KiX Supporter

Registered: 2000-09-15
Posts: 5809
Loc: Harrisburg, PA USA

code:

function FadeToBlack($color, $level)

	Dim $B, $G, $R
	$B=$color/65536
	$G=($color mod 65536)/256
	$R=$color-($B*65536)-($G*256)


	$B = IIF($B>$level, $level, $B)
	$G = IIF($G>$level, $level, $G)
	$R = IIF($R>$level, $level, $R)

;by matching level these checks are no longer required.
	;$B = IIF($B<0, 0, $B)
	;$G = IIF($G<0, 0, $G)
	;$R = IIF($R<0, 0, $R)

	$FadeToBlack = ($B*65536)+($G*256)+$R

endfunction

[ 03. November 2003, 16:41: Message edited by: Howard Bullock ]

_________________________
Home page: http://www.kixhelp.com/hb/

Top

#106510 - 2003-11-03 04:41 PM Re: Mini-putt for a slow day on korg ...
Shawn Shawn Administrator Registered: 1999-08-13 Posts: 8611	Howard - PERFECT ! Jlo - you got another submission ? Want to do speed tests now.
Top

#106511 - 2003-11-03 04:54 PM Re: Mini-putt for a slow day on korg ...
Howard Bullock Howard Bullock KiX Supporter Registered: 2000-09-15 Posts: 5809 Loc: Harrisburg, PA USA	Shawn did you catch the edit above? _________________________ Home page: http://www.kixhelp.com/hb/
Top

#106512 - 2003-11-03 04:59 PM Re: Mini-putt for a slow day on korg ...
Shawn Shawn Administrator Registered: 1999-08-13 Posts: 8611	yes I did, thank you. plus - I took the liberty of renaming all your internal $vars to single letters, to make it afap.
Top

#106513 - 2003-11-03 05:04 PM Re: Mini-putt for a slow day on korg ...
Howard Bullock Howard Bullock KiX Supporter Registered: 2000-09-15 Posts: 5809 Loc: Harrisburg, PA USA	Like this? code: function FadeToBlack($c, $) Dim $B, $G, $R $B=$c/65536 $G=($c mod 65536)/256 $R=$c-($B65536)-($G256) $B = IIF($B>$, $, $B) $G = IIF($G>$, $, $G) $R = IIF($R>$, $, $R) $FadeToBlack = $B65536+$G256+$R endfunction [ 03. November 2003, 17:06: Message edited by: Howard Bullock ] _________________________ Home page: http://www.kixhelp.com/hb/
Top

#106514 - 2003-11-03 05:11 PM Re: Mini-putt for a slow day on korg ...
Howard Bullock Howard Bullock KiX Supporter Registered: 2000-09-15 Posts: 5809 Loc: Harrisburg, PA USA	Shawn, I would think that this would be faster. The IIF has to rest the value of the color variable even if it did not need to change it. code: function FadeToBlack($c, $) Dim $B, $G, $R $B=$c/65536 $G=($c mod 65536)/256 $R=$c-($B65536)-($G256) if $B>$ $B=$ endif if $G>$ $G=$ endif if $R>$ $R=$ endif $FadeToBlack = $B65536+$G256+$R endfunction _________________________ Home page: http://www.kixhelp.com/hb/
Top

#106515 - 2003-11-03 05:45 PM Re: Mini-putt for a slow day on korg ...
Shawn Shawn Administrator Registered: 1999-08-13 Posts: 8611	Howard, just working on my short game here. Wondering if this might be shorter/faster ... code: $r = $c & &FF $g = ($c & &FF00) / 256 $b = ($c & &FF0000) / 65536 plus, I concatenated your IIF's (which I like) into a big long expression: code: $FadeToBlack = IIF($B>$, $, $B) * 65536 + IIF($G>$, $, $G) * 256 + IIF($R>$, $, $R) -Shawn [ 03. November 2003, 17:48: Message edited by: Shawn ]
Top

#106516 - 2003-11-03 06:16 PM Re: Mini-putt for a slow day on korg ...
Shawn Shawn Administrator Registered: 1999-08-13 Posts: 8611	hmmm, combining both strategies yields: code: function FadeToBlack($c, $) dim $b, $g, $r $b = $c / 65536 $g = ($c & &FF00) / 256 $r = $c & &FF $FadeToBlack = iif($b>$, $, $b) * 65536 + iif($g>$, $, $g) * 256 + IIF($r>$, $, $r) endfunction
Top

#106517 - 2003-11-03 06:29 PM Re: Mini-putt for a slow day on korg ...
Shawn Shawn Administrator Registered: 1999-08-13 Posts: 8611	Think this is actually four chars shorter. Even with the redundancy ... code: function FadeToBlack($c, $) $FadeToBlack = iif($c/65536>$, $, $c/65536) * 65536 + iif(($c & 65280) / 256>$, $, ($c & 65280) / 256) * 256 + iif($c & 255>$, $, $c & 255) endfunction [ 03. November 2003, 18:29: Message edited by: Shawn ]
Top

#106518 - 2003-11-03 06:36 PM Re: Mini-putt for a slow day on korg ...
Howard Bullock Howard Bullock KiX Supporter Registered: 2000-09-15 Posts: 5809 Loc: Harrisburg, PA USA	But is your goal the least characters used or the most efficient code with he fastest execution? Do you have any benchmarks for each version? [ 03. November 2003, 18:45: Message edited by: Howard Bullock ] _________________________ Home page: http://www.kixhelp.com/hb/
Top

#106519 - 2003-11-03 06:41 PM Re: Mini-putt for a slow day on korg ...
Shawn Shawn Administrator Registered: 1999-08-13 Posts: 8611	yeah, i'm just playing around with the solution now - trying to get the smallest solution. Will end-up using the fastest solution at the end of the day. Will be fun to compare all the solutions.
Top

#106520 - 2003-11-03 06:47 PM Re: Mini-putt for a slow day on korg ...
Shawn Shawn Administrator Registered: 1999-08-13 Posts: 8611	well, here are some benchmarks the last snippet posted (the scrunched IIF's) for 10,000 iterations takes (avg) 2293 msecs. the second-last snippet posted (the combined solution with BGR and IIF's seperate) takes (avg) 2874 msecs for 10,000 iterations. So it would seem that smaller and scrunched helps with the numbers. Should try the IF-ENDIF solution for speed though. [ 03. November 2003, 18:49: Message edited by: Shawn ]
Top

#106521 - 2003-11-03 06:50 PM Re: Mini-putt for a slow day on korg ...
Howard Bullock Howard Bullock KiX Supporter Registered: 2000-09-15 Posts: 5809 Loc: Harrisburg, PA USA	Yes, what is that result??? _________________________ Home page: http://www.kixhelp.com/hb/
Top

#106522 - 2003-11-03 06:52 PM Re: Mini-putt for a slow day on korg ...
Shawn Shawn Administrator Registered: 1999-08-13 Posts: 8611	You last IF/ENDIF solution for 10,000 iterations takes (avg) 3615 msecs.
Top

#106523 - 2003-11-03 06:56 PM Re: Mini-putt for a slow day on korg ...
Howard Bullock Howard Bullock KiX Supporter Registered: 2000-09-15 Posts: 5809 Loc: Harrisburg, PA USA	Crap! I still think the If...Endif should be faster than IIF. Keep the If..Endif's and toss the "mod" operator in favor of the "&" bitwise operator. The slowness is probably in the additional math I do. _________________________ Home page: http://www.kixhelp.com/hb/
Top

#106524 - 2003-11-03 06:59 PM Re: Mini-putt for a slow day on korg ...
Howard Bullock Howard Bullock KiX Supporter Registered: 2000-09-15 Posts: 5809 Loc: Harrisburg, PA USA	code: function FadeToBlack($c, $) Dim $B, $G, $R $B=$c/65536 $G=($c & &FF00) / 256 $R=$c & &FF if $B>$ $B=$ endif if $G>$ $G=$ endif if $R>$ $R=$ endif $FadeToBlack = $B65536+$G256+$R endfunction _________________________ Home page: http://www.kixhelp.com/hb/
Top

#106525 - 2003-11-03 07:02 PM

Re: Mini-putt for a slow day on korg ...

Shawn

Administrator

Registered: 1999-08-13
Posts: 8611

Runs about 3275 msecs on my machine, so the math seems to have an impact. Here is the benchmark script, just rename the function as required. Don't think the order of functions has an impact:

code:

break on
 
$start = @TICKS
 
for $i = 0 to 10000
 $c = fadeToBlack2(&FFAADD,&DD)
next
 
?"elapsed=" @TICKS-$start
 
exit 1
 
function FadeToBlack($c, $)
 
 dim $B, $G, $R
 
 $b = $c / 65536
 $g = ($c & &FF00) / 256
 $r = $c & &FF
 
 if $B>$	$B=$ endif
 if $G>$	$G=$ endif
 if $R>$	$R=$ endif
  
 $FadeToBlack = $B*65536+$G*256+$R
 
endfunction
 
function FadeToBlack2($c, $)
 
 $FadeToBlack2 = 
	iif($c/65536>$, $, $c/65536) * 65536 + 
	iif(($c & 65280) / 256>$, $, ($c & 65280) / 256) * 256 + 
	iif($c & 255>$, $, $c & 255)
 
endfunction

Top

#106526 - 2003-11-03 07:25 PM Re: Mini-putt for a slow day on korg ...
Howard Bullock Howard Bullock KiX Supporter Registered: 2000-09-15 Posts: 5809 Loc: Harrisburg, PA USA	Shawn, after some experimentation. I have concluded that the high overhead operation is the instantiation of variables. Your FadeToBlack2 avoids the creation of variables as much as seems possible. The additional calculations appear to add less burden to the process than the creation of variables. Good job optimizing. [ 03. November 2003, 19:28: Message edited by: Howard Bullock ] _________________________ Home page: http://www.kixhelp.com/hb/
Top

#106527 - 2003-11-03 07:34 PM Re: Mini-putt for a slow day on korg ...
Shawn Shawn Administrator Registered: 1999-08-13 Posts: 8611	The other thing that may contribute to the greater speed of IIF is the following (just speculating here). Even though the IIF's have redundant calculations, the "checking" and the "result" are internalized within the IIF function itself. Unlike the IF/ENDIF code - these statements have to be "parsed-around" by the interpreter (even though it is supposed to be byte-code internally) - yeah, maybe ?
Top

Page 2 of 3

Previous Topic View All Topics

Index Next Topic

Moderator: Arend_, Allen, Jochen, Radimus, Glenn Barnas, ShaneEP, Ruud van Velsen, Mart

Hop to:

Shout Box

Search the board with:
superb Board Search
or try with google:

	Web kixtart.org