Stupid noob Q. - AVS Forums

mIcrO_brAIn

24th December 2004 23:02 UTC

Stupid noob Q.
Ok, I think I've got this right but if someone would verify/explain this to me it would be appreciated.

It's in the use of the % operator in AVS. The explanation in "expression help" doesn't make a whole lot of sense.

If I were to use 5%2 then I would get the remainder, which would be .5 right?

And reading the AVS programming guide, in the first explanation about 3d scopes, the code point=(point+1)%4 is used. How does the % operator affect this code?

It's probably a dumb question but I just didn't seem to be able to figure this one out.

^..^

24th December 2004 23:53 UTC

The modulus operator % actually gives you the remainder of the division 5/2. But it wouldn't be 0.5; its 1!
The remainder of that divison means: what is left over in order to avoid decimals?

5/2 = 2 + 1/5

Mostly this operator is used, to generate a variable that repeatedly counts to a specific value and afterwards starts with 0 again:

point=(point+1)%4 (only convenient when set in "pixel")

means that the variable point increases by one and jumps back to 0 after reaching 3:

0 1 2 3 0 1 2 3 0 1...

For example: 4 can be divided by 4 without a remainder. 5 is by 1 greater than 4; 6 is 2 greater than 4; 7 -> 3, but 8 can be divided by 4 without a rest again, and so on.

Clear?? If not, just ask again!

P.S: merry christmas

mIcrO_brAIn

25th December 2004 02:39 UTC

Might it be the same thing as saying:
point=point+1;
point=if(equal(point,4),0,point);
So that when point hit 4 it would revert back to 0.

WalkerP

25th December 2004 04:43 UTC

Yes! Exactly right!

^..^

25th December 2004 11:14 UTC

Originally posted by mIcrO_brAIn
Might it be the same thing as saying:
point=point+1;
point=if(equal(point,4),0,point);
So that when point hit 4 it would revert back to 0.

This would be the same, but

point=(point+1)%4

is much faster cause its only one equation. (Probably its a imperceptible difference, but if you've got a big bunch of code it'll be worth)

sidd

26th December 2004 00:48 UTC

it also has many other uses, because unlike the if equal combination, you can perform it on any number at any time, it doesnt have to be a count.

for example, anyNumber%2 will tell you if the number is odd, because if it is, the remainder will be 1.

the downside is, the % operator only works on integers, not decimals, so decimal precision is lost when you use it.

MaTTFURY

26th December 2004 07:43 UTC

what f*cking genius coded the compabilities of the processor :p

TomyLobo

26th December 2004 11:04 UTC

all intel coprocessors since the 8087 and all intel processors since the 80486DX (not SX) have an instruction called FPREM
this instruction does a floating point remainder. (80387 and above additionally have FPREM1, which is should be used instead)
so much about the "compabilities" of processors ;)

back to the topic:
if you want a point counter go from 0 to n-1, write p=0; into frame and p=p+1; at the end of the point block.
if n is 7, for example, p will go this way:
0 1 2 3 4 5 6
in the next frame, p is reset to 0 and the whole thing will start from the beginning.
the advantage over just writing p=(p+1)%n; at the end of the point block is that the other method is much faster , because % is a division, which is generally not that fast ;)

that's why you normally write a=b*.5 instead of a=b/2;

PAK-9

26th December 2004 12:42 UTC

Originally posted by TomyLobo
back to the topic:
if you want a point counter go from 0 to n-1, write p=0; into frame and p=p+1; at the end of the point block.
if n is 7, for example, p will go this way:
0 1 2 3 4 5 6
in the next frame, p is reset to 0 and the whole thing will start from the beginning...

unfortunately I dont want a variable to go from 0 to n-1, I want a variable to go from 0 to n-1 then 0. However as people have pointed out it is a little poorly explained and arguably inefficient. A better line would be:

frame:
point=0;
point:
...
point=if(equal(point,3),0,point+1);

And I shall modify the guide accordingly, thank you.

----- techy bit ------

On an additional and rather anal point I believe its marginally more efficient to have:

frame:
point=3
point:
..
point=if(bnot(point),3,point-1);

It draws the square backwards but of course that doesnt matter, its a lot more ugly to understand tho. Also, since I think if's are more expensive than multiplications it may be more efficient to have:

frame:
point=3
point:
..
point=point-1+bnot(point)*3;

but its pretty gosh darn ugly.

^..^

26th December 2004 22:39 UTC

you got the point pak! :D

jheriko

6th January 2005 02:05 UTC

i thought * is slower than if by a lot... otherwise why have if?

[Ishan]

6th January 2005 13:02 UTC

um....i'm very tech here but actually what Jheriko said above i always thought the same thing. I was told by someone....er...i think Shreyas told me that * is slower than if.

Warrior of the Light

6th January 2005 13:50 UTC

/agrees that if() is faster than *
(if() doesn't caclulate, but compares only)

PAK-9

6th January 2005 15:32 UTC

Originally posted by jheriko
i thought * is slower than if by a lot... otherwise why have if?

To do conditional processing :weird:

To be honest I dont really know which is more expensive I just assumed it was 'if'. But now that I think about it its really just a jump-if-not-zero in asm... and since it short circuits you are removing some of the computation. Okay I'm sold, it probably is cheaper... but someone should really check the evallib and prove it. I'd do it myself but... well I can't really be arsed.

PJ-Avser

7th January 2005 01:52 UTC

I've tried it out on a ridculusly numpointeds scc, and * is faster, BUT THE DIFFENECE IS SO TINY, IT DOES'NT REALLY MATTER

PAK-9

7th January 2005 19:56 UTC

pwnt

^..^

8th January 2005 00:41 UTC

wow, all that expenditure for nothing... what a waste!

S-uper_T-oast

8th January 2005 02:59 UTC

It depends on the use though. If() will be faster for some things and * will be faster for others.

PJ-Avser

8th January 2005 05:54 UTC

Each time it did the if it gained 0.00000001 fps over times.
On my PC

jheriko

8th January 2005 09:04 UTC

you'd think avs would optimise if(a,b,0) to a*b if if was slower than a * :P