Optimises ISEVEN() , ISODD() macro

[EvilDragonfiend]

About The Pull Request

Optimises ISEVEN() , ISODD() macro

calculation is expensive than comparison bitwise operator. When we know first bitflag makes a value even or odd, we don't have to calculate to get 0 by divide 2.

Division is extremely cheap. What isn't extremely cheap is finding the remainder of division. Secondly, you aren't comparing, you're doing a bitwise math operation, comaprison is ==. Bitwise math is actually really slow in byond due to how numbers work in the engine, but it is funny it's slower than modulo.

Kapu's comment

Why It's Good For The Game

optimisation

Testing Photographs and Procedure

simulation code

#define ISEVEN(x) (!((x) & 1))
#define ISODD(x) ((x) & 1)
#define ISEVENMOD(x) (x % 2 == 0)
#define ISODDMOD(x) (x % 2 != 0)

/proc/main()
    set background = 1
    var/const/repeat = 15000000
    var/t1
    var/t2
    for(var/i in 1 to 100)
        TEST1(0)
        TEST2(0)

    var/time1 = world.timeofday 
    for(var/i in 1 to repeat)
        TEST1(i)
    t1 = world.timeofday-time1
    world.log << "bitflag &: [t1] taken"

    var/time2 = world.timeofday
    for(var/i in 1 to repeat)
        TEST2(i)
    t2 = world.timeofday-time2
    world.log << "mod %: [t2] taken"

/proc/TEST1(val)
    return ISODD(val)

/proc/TEST2(val)
    return ISODDMOD(val)

Note

It has the same result for decimal numbers

Changelog

🆑
code: improved ISEVEN() and ISODD() macro
/:cl: