Zinc:
native StopWatchCreate takes nothing returns integer
native StopWatchMark takes integer stopwatch returns real
native StopWatchDestroy takes integer stopwatch returns nothing
//! textmacro TestPattern takes code
TriggerSleepAction(0);
BJDebugMsg("$code$");
sw = StopWatchCreate();
for(i = 0; i < 1000; i+=1)
{
$code$
$code$
$code$
$code$
$code$
$code$
$code$
$code$
$code$
$code$
}
res= StopWatchMark(sw);
StopWatchDestroy(sw);
BJDebugMsg(R2S(res*1000));
//! endtextmacro
//! zinc
library CollectionTest
{
hashtable table = InitHashtable();
public function Trig_CollectionTest_Actions()
{
integer i;
integer sw;
real res;
integer num;
//! runtextmacro TestPattern("/* exists miss */HaveSavedInteger(table,0,1);")
//! runtextmacro TestPattern("/* Get Miss */ LoadInteger(table,0,1);")
//! runtextmacro TestPattern("/* set */SaveInteger(table,0,2,2);")
//! runtextmacro TestPattern("/* Get hit */ LoadInteger(table,0,2);")
//! runtextmacro TestPattern("/* exists hit */ HaveSavedInteger(table,0,2);")
//! runtextmacro TestPattern("/* flush */ RemoveSavedInteger(table,0,2);")
}
}
//! endzinc
function InitTrig_CollectionTest takes nothing returns nothing
set gg_trg_CollectionTest = CreateTrigger( )
call TriggerRegisterPlayerEventEndCinematic( gg_trg_CollectionTest, Player(0) )
call TriggerAddAction( gg_trg_CollectionTest, function Trig_CollectionTest_Actions )
endfunction
I was testing the hashtable natives if it was worth using the contains method vs a loadinteger and not storing 0.
The means of testing was trying each method 10 times in an iterative loop of 1000 tries. This was to lower the time wasted on simply setting i= i+1. After each test the stopwatch was destroyed and a sleep was used to clense the pop limit.
Each of these results were multiplied by 1000 to make them large enough to notice a difference. Test + numbers::
- HaveSavedInteger(table,0,1); 10.527
- LoadInteger(table,0,1); 9.892
- SaveInteger(table,0,2,2); 10.943
- LoadInteger(table,0,2); 10.248
- HaveSavedInteger(table,0,2); 10.999
- RemoveSavedInteger(table,0,2); 11.355
Admittedly, the numbers seem a bit different then my try with handles, but it seems when integers are concerned, Remove is slower then setting to 0, and contains is slower then just getting.
I didn't assign to variables or test or anything like that to be fair, as if statements in jass seem to be lightening fast, whereas setting is slower.
