GroupRemoveUnit and IsUnitInGroup

01-07-2009, 09:47 PM

if we need to remove GetTriggerUnit(), do we really need for speed freak to check if the unit is in the group before it, like that

JASS:

if IsUnitInGroup(GetTriggerUnit(),<group>) then
   call GroupRemoveUnit(<group>,GetTriggerUnit())
endif

or can we do simply that with the same speed, or maybe faster ?

JASS:

call GroupRemoveUnit(<group>,GetTriggerUnit())

EDIT : Ofc, if we don't need to know that the unit was on the group or not.

01-08-2009, 12:25 AM	#2
Zerzax	You'll probably know whether or not the unit is in the group anyway. You should, at least. It's totally based on context, tell more of what your script is trying to do.

01-08-2009, 03:29 AM	#3
DioD	You dont need any checks for this. Just remove no errors will come.

01-08-2009, 03:31 PM

Troll-Brain

Quote:

Originally Posted by Zerzax

You'll probably know whether or not the unit is in the group anyway. You should, at least. It's totally based on context, tell more of what your script is trying to do.

No really i don't need it, believe me.

@ Diod : Yes i know that no error will come, but in fact my question is for speed freak only.

IsUnitIsGroup vs GroupRemoveUnit.

01-08-2009, 04:15 PM	#5
DioD	For group check is near instant, but if+else+then slow. Sadly there is no way to build if+else+then without if no jumps around same function lines allow. Ah jump if equal + section markings for jass == divine power.

01-08-2009, 04:16 PM	#6
Zerzax	GroupRemoveUnit would be faster, because IsUnitInGroup would require using conditionals in any application.

01-08-2009, 04:53 PM

Troll-Brain

Quote:

Originally Posted by Zerzax

GroupRemoveUnit would be faster, because IsUnitInGroup would require using conditionals in any application.

Are you sure about that, because in fact, most of the times the unit won't be in the group.
So if IsUnitInGroup is faster than GroupRemoveUnit, when the unit is not in the group, then i will use both, like i've said in the first post of this thread.

Could someone benchmark plz ? Because I can't use japi, thx.

01-08-2009, 05:18 PM	#8
DioD	if else then is much more slower then any native. benchmark will show it.

01-08-2009, 05:27 PM

Troll-Brain

Quote:

Originally Posted by DioD

if else then is much more slower then any native.

benchmark will show it.

ok but just to know, how many times ?

01-08-2009, 06:02 PM

#10

Alexander244

JASS:

scope Test initializer Init

    globals
        private timer t = CreateTimer()
        private group g = CreateGroup()
        private unit u
        private integer i = 0
    endglobals

    function Bench takes nothing returns nothing
        local integer sw = StopWatchCreate()
        
        //Test 1 & 2:
        if IsUnitInGroup(u,g) then
            call GroupRemoveUnit(g,u)
        endif
        //
        
        //Test 3 & 4:
        call GroupRemoveUnit(g,u)
        //
        
        call DebugPrint(R2S(StopWatchMark(sw)*1000000))
        call StopWatchDestroy(sw)
        
        //Test 1 & 3:
        call GroupAddUnit(g,u)
        //
        
        set i = i+1
        if i > 50000 then
            call PauseTimer(t)
            call BJDebugMsg("Done")
        endif
    endfunction

    private function Init takes nothing returns nothing
        call DebugPrint("=============================")
        set u = CreateUnit(Player(0), 'hpea', 0., 0., 0.)
        
        //Test 1 & 3:
        call GroupAddUnit(g,u)
        //
        
        call TimerStart(t, 0., true, function Bench)
    endfunction
    
endscope

war3err off, done three times, time for null-test subtracted, etc.

Times in microseconds:

Test 1:	Test 2:	Test 3:	Test 4:
3.249	1.708	1.728	1.653

01-08-2009, 06:25 PM	#11
Troll-Brain	Thx, is it the same if the unit is not in the group, i dunno how groups are handle, but i guess it should take more time, no ?

01-08-2009, 06:41 PM	#12
Alexander244	Tests 2 & 4 don't have the unit in the group, tests 1 & 2 do. Otherwise tests 1 and 2 are the same, and tests 3 and 4 are the same.

01-08-2009, 06:57 PM

#13

Troll-Brain

Quote:

Originally Posted by Alexander244

Tests 2 & 4 don't have the unit in the group, tests 1 & 2 do. Otherwise tests 1 and 2 are the same, and tests 3 and 4 are the same.

You want mean that ?

Quote:

Tests 2 & 4 don't have the unit in the group, tests 1 & 3 do. Otherwise tests 1 and 2 are the same, and tests 3 and 4 are the same

01-08-2009, 06:59 PM	#14
Alexander244	opps, yes.

01-08-2009, 07:06 PM	#15
Troll-Brain	Thx for testing.