SC Array Speed Test

[Gan]

SC Array Speed Test
Video of the tests in action.

This compares a nthfield array to a real array. It runs each test 10,000 times. Storing the number 5 in each array and retrieving the number from the array.

For these tests on my iMac I have this data:
- 7 ticks for the real array test.
- 136 ticks for the nthfield array test.

Real arrays are about 20* faster than nthfield arrays.
This is a substantial lag for projects that use nthfield arrays a lot.

[Connors]

Wow, I thought it would be a bit slower but 20* is sort of surprising.

[EqwanoX]

daaaaaammmnn! thats insane!

lol, it takes 56ticks for number array and 519 ticks for string on my g5

HOLY SHlT, ive used a FOR loop before for replaceing specific fields, but not as a method, just raw code, and even though its virtually the SAME, for some reason redds code as a method runs insanly faster, i couldnt even get a reading on it, it just kept going, i had to force quit, thats so fucking wierd, now i have to go back and modify my for loops as methods

to be fair though, i dont see a circumstance where you would constantly replace specific fields like this, i think it would only be for specific circumstances like purchasing something that increases a stat .

holy shit, useing number arrays is faster than useing number variables! so let x(1)=5 is faster than let x=5

same with strings, its about 20% faster for me! so you can make your game faster by makeing all strings and variables as arrays even if your storing just one value

[Gan]

How interesting...

[Silverwind]

same with strings, its about 20% faster for me! so you can make your game faster by makeing all strings and variables as arrays even if your storing just one value

Aye, but with the possible exception of TD I doubt players can physically click things fast enough to notice a difference.

[GMG Mike]

There's no logical reason why number arrays would be faster than number variables. It takes more work (within the runtime) to modify a value in the number array.

[Silverwind]

Logic be damned; this isn't brain surgery, it's computer science!

[Connors]

Maybe arrays are faster because it's a more specific location for the computer to look up. Otherwise it gets the variable's value from a larger group (similar to what slows down fake string arrays).

[EqwanoX]

its true! check it out, in some cases 50% faster


http://www.mediafire.com/?owd5uspt8c2kxz6

[Gan]

Hmm. I just tested your project and I'm getting normal variables are faster than arrays.

[Connors]

Say whaaaaaaaaat?

[EqwanoX]

you might need a restart or quit some other apps, cause even on my intel its 40% faster

[Gan]

Did you have the variable window opened or closed?
I'm gonna try another test.

[Al Staffieri]

Arrays were faster for me with EQ's test.

[GMG Mike]

On the Mac Pro both of eq's tests take 1 tick  

I increased the number of times from 1,000 to 100,000. This gives a more accurate result.

SilverCreator v2.0b3.1

VARIABLES WINDOW OPEN
Variable: 114 ticks
Variable Array: 73 ticks

VARIABLES WINDOW CLOSED
Variable: 15 ticks
Variable Array: 19 ticks

COMPILED GAME:
Variable: 14 ticks
Variable Array: 17 ticks