MacOSX Performance

Karsten Wolf karstenwo at web.de
Tue Jan 25 00:20:06 UTC 2005


Am 24.01.2005 um 22:10 schrieb John M McIntosh:

>
> On Jan 24, 2005, at 12:39 PM, Karsten Wolf wrote:
>
>>
>> Regions don't affect performance that much. Rectangles tend to grow 
>> too fast. Consider two small updates: one at 0 at 0 the other at 800 at 800 
>> moving 64000 pixels instead of a few.
>
> Post some sample code suggestions, and we'll compare.
> --

Hhm, my actual tests seem to make my statent an understatement. But of 
course I've chosen snippets accordingly...



Here it comes:
machine TiBook 867MHz, squeak diplay:32bit mac display 32bit; updater 
set to minFPS:30 maxFPS:40


the snippet:

| bic |
TimeProfileBrowser onBlock: [
bic _ Pen new.
bic color: Color yellow.
5 timesRepeat:[
	Display fillBlack.
	0 to: 63 do:
	[:i|
		bic up; goto: (16 * i)@0; down;goto: 1023@(i * 12).
		bic up; goto: (1023-(16 * i))@767; down; goto: 0@(767-(i * 12)).
		bic up; goto: (1023-(16 * i))@0; down; goto: 0@(i * 12).
		bic up; goto: (16 * i)@767; down; goto: 1023@(767-(i * 12))
	]]]

VM3.8.1b2: 16126
Using rectangles for damageCollection: 5463 msec
Using regions for damageCollection: 3156 msec

Using a list of rectangles with maximum size of 128 at 128 for caching: 
2289 msec



This one requires the Commander class from the purple book (attached)

the snippet:
| bic |
TimeProfileBrowser onBlock: [
	Display fillWhite.
	bic _ Commander new: 8.
	bic fanOut.
	bic do: [:eachPen| eachPen up. eachPen go: 100. eachPen down].
	bic do: [:eachPen| eachPen spiral:200 angle: 121]
]

VM3.8.1b2: 7406
Using rectangles for damageCollection: 1101 msec
Using regions for damageCollection: 539 msec

Using a list of rectangles with maximum size of 128 at 128 for caching: 328


regards
karsten

-------------- next part --------------
A non-text attachment was scrubbed...
Name: Commander.st.gz
Type: application/x-gzip
Size: 668 bytes
Desc: not available
Url : http://lists.squeakfoundation.org/pipermail/squeak-dev/attachments/20050125/63f7544e/Commander.st.bin
-------------- next part --------------




More information about the Squeak-dev mailing list