properties vs local variable

[dunbarx at aol.com]

Alex. I had tried lines 3, 3000 and 7500.


I did indeed only report the first value, likely because I wanted to slant my point about different types of processes overwhelming others in my favor. It is a character flaw.



But I got very similar results with the other two:



For line 3: 0 and 2 ticks
For line 3000: 57 and 59 ticks


For line 7500: 135 and 137 ticks


The same offset, if you will, as the 0 and 2 that I got with line 3.


Not sure why yours are so different. I had, with my 100,000 line list in "the testProp" of the card:


on mouseUp --test variable
   get the ticks
put the testProp of this cd into temp
   repeat 10000
      put line 7500 of  temp into xyz
      --put any line of temp into xyz --that other "process"
   end repeat
   answer the ticks - it
end mouseUp


on mouseUp --test custom Prop
   get the ticks
   repeat 10000
      put line 7500 of the testProp of this cd into temp
      --put any line of the testprop of this cd into temp  --that other "process"
   end repeat
   answer the ticks - it
end mouseUp


Craig




-----Original Message-----
From: Alex Tweedly <alex at tweedly.net>
To: use-livecode <use-livecode at lists.runrev.com>
Sent: Mon, Mar 10, 2014 12:45 pm
Subject: Re: properties vs local variable


On 10/03/2014 15:33, dunbarx at aol.com wrote:
> I tried a simple test. I placed 100,000 lines of text into a custom property 
and into a variable. I timed how long it took to directly access a single line, 
10,000 times. It took 2 ticks for the custom property, and almost no time at all 
for the variable.
The question is - *which* "single line" did you test with ?

with the same circumstances, accessing
line       1   takes 3 msec
line 5000   takes 349 msec

>
> So the custom property is slower, but still pretty fast. What really matters 
is other stuff. For example, in the same pair of setups, if I asked for "any 
line" of either the prop or the variable, the process took about 30 seconds. 
That function call took its sweet time. So the lesson here is to watch carefully 
what is a first order issue, and what is a fourth order issue.
It's not just the "any" that takes the time, it's the fact that on 
average it will be equivalent to the middle line.

-- Alex.

_______________________________________________
use-livecode mailing list
use-livecode at lists.runrev.com
Please visit this url to subscribe, unsubscribe and manage your subscription 
preferences:
http://lists.runrev.com/mailman/listinfo/use-livecode