[JASS] How to do X to the 2nd Power?

[Titanhex]

Found it. It's Pow.

 

[D4RK_G4ND4LF]

x * x is faster

 

[Titanhex]

Only if the variable is a small variable like i.

My variables contained arrays, unfortunately and I had 2 to sort out on a single line.

 

[Element of Water]

x*x*x*x*x*x*x*x*x*x*x*x*x*x*x*x*x*x*x*x*x*x*x*x*x*x*x*x*x*x is faster than Pow(x, 30)

 

[Titanhex]

Are we talking processor speed, or time to enter in the information?

If it's just time to enter in the information, I prefer the far more organized look of Pow(x, 2) than x*x

If there's something more significant to it than I'm interested and all ears.

 

[Element of Water]

We're talking processor speed. Pow (I think, someone correct me if I'm wrong) uses a conditional loop to multiply x, which means it has to evaluate <index> conditions, or more, making it much slower than straight multiplication. If processing speed doesn't matter to you, however, feel free to use Pow, but i personally think x*x looks more organised, however I can see why anything over a power of about 4 might tempt you to use the function.

 

[Hemlock]

Correct, however when not used oftenly, don't be bothered to use the pow function inside the editor, it's nothing significant really + if you want to have x to the power of 50, you might get sore finger from clicking or typing once your done.

 

[Element of Water]

In something that runs 60 times per second, when there are 100 instances of it, making a total of 6000 runs per second, using Pow could hit performance significantly.

 

[PurplePoot]

We're talking processor speed. Pow (I think, someone correct me if I'm wrong) uses a conditional loop to multiply x, which means it has to evaluate <index> conditions, or more, making it much slower than straight multiplication. If processing speed doesn't matter to you, however, feel free to use Pow, but i personally think x*x looks more organised, however I can see why anything over a power of about 4 might tempt you to use the function.

No, it uses a power series. A for() in C may very well be (in fact, has incredibly good odds of being) faster than that many multiplications in Jass.

 

[The Reborn Devil]

Vexorian should make NewGen pack replace "Pow()" calls with x * x * x .....
It can't be that hard to put in

 

[PurplePoot]

Pow(x,-3.2)
Pow(x,4.142341)
Pow(x,y)

Good luck.

 

[The Reborn Devil]

Pow(x,y)

ohsh-

Pow(x,-3.2)

1/x^3.2
which happens to be 1/x^(16/5) which is 1/ sixteenth root of (x*x*x*x*x)

Pow(x,4.142341)

Similar to the one above just without the "1/" and something else than 16/5

Besides, he could make it ignore those ^^

 

[Antares]

In the other thread you guys said, that even Sine and Cosine functions are nothing to worry about when processed quite often. Im confused.
As far as I know, computers approximate Sine and Cosine with a taylor series (correct me if im wrong). that takes alot more processing time then a simple potentialization (hope that word exists).

 

[Element of Water]

@Poot

Why the hell would JASS multiplication be that slow?

 

[Eccho]

Because Jass in an emulated language?

 

[PurplePoot]

@Poot

Why the hell would JASS multiplication be that slow?

Because interpreters are by definition slower than truly compiled languages?

 

[Element of Water]

True, but do a few string operations plus the multiplication cost more than evaluating lots of conditions plus multiplication?

 

[PurplePoot]

A few string operations? What?

Also, the cost of > is extremely low.

 

[xxdingo93xx]

x * x is faster

X^Pow
So far Pow is a whole number it's not a big difference.
If Pow is a floating point number its slow and x * x is better indeed.

 

[PurplePoot]

Uuh, you have that so completely and utterly confused that I don't even know how to correct it without rewriting it from scratch.

 

[Element of Water]

I was under the impression that interpreters looked at the script (a big string) and performed actions based on it?

 

[PurplePoot]

That's a very simplistic way of describing a subset of interpreters which I'm fairly sure the JASS VM does not fall into.

 

[Mooglefrooglian]

@Poot

Why the hell would JASS multiplication be that slow?

JASS is interpreted into bytecode when the map loads. So like Python, Lua, and every other interpreted language, it's slow as molasses compared to something like C++.

It's still blazing fast though. It may be 5x slower, but 5 more nanoseconds isn't noticeable at all.

 

[PurplePoot]

The greater problem is that he overestimates the cost of loops and conditionals.

 

[Berb]

Correct, however when not used oftenly, don't be bothered to use the pow function inside the editor, it's nothing significant really + if you want to have x to the power of 50, you might get sore finger from clicking or typing once your done.

set i = 0
loop
    exitwhen i == 50
    set x = x * x
    set i = i+1
endloop

 

[Element of Water]

a) That would be a lot slower than Pow.
b) It doesn't work right, because it multiplies x by x, so, for example 2*2=4, then multiplies that by x again, so 4*4=16, when it should multiply it by 2 again.

 

[The Reborn Devil]

set i = 0
set i2 = x
loop
    exitwhen i == 50
    set x = x * i2
    set i = i+1
endloop

Fixed.

 

[Element of Water]

But it's still slower.

 

[PurplePoot]

And, more importantly, who cares?

 

[The Reborn Devil]

But it's still slower.

Indeed.

And, more importantly, who cares?

Umm, no one?

 

[Berb]

It was actually a joke. Nobody laughed.