Good Benchmarks?

[tooltiperror]

I was thinking of doing some benchmarks. Are there any functions people need checked?

 

[Lyerae]

Do some tests to see if hashtables slow down after loading shitloads of data into them.

I was going to do that bench, but I epicfail at benching.

 

[Narks]

I was thinking of doing some benchmarks.

We're screwed.

 

[tooltiperror]

Do some tests to see if hashtables slow down after loading shitloads of data into them.

I was going to do that bench, but I epicfail at benching.

-facepalm- I`ll try that one.

We're screwed.

LOLOL.

 

[Azlier]

Benchmark [lJASS]LoadInteger(ht, StringHash("abc"), StringHash("def"))[/lJASS] vs. [lJASS]GetStoredInteger(gc, "abc", "def")[/lJASS].

Store some value like 1100 in each before benchmarking.

 

[Nestharus]

hashtables would slow down when they have bundles of data... they're hashtables....

/facepalm


do a worth while benchmark that people don't already know the answer to ^_^

 

[Azlier]

Who said anything about bundles of data?

I said to store one value in each, the value being something like 1100.

 

[tooltiperror]

Benchmark [lJASS]LoadInteger(ht, StringHash("abc"), StringHash("def"))[/lJASS] vs. [lJASS]GetStoredInteger(gc, "abc", "def")[/lJASS].

Store some value like 1100 in each before benchmarking.

I like that one.

Now I just have to figure out these .dll's and crap...

 

[Nestharus]

I got a good one. Check function call speed between recursive function calls and regular function calls : ). Normally, a recursive function call is pretty much exactly like a loop. I'm really curious to see if a recursive function call is normal in JASS or if it is like a regular function call ;o.

Also please check the memory usage on recursive function calls ; ). We all know that starting up new threads can use up a pretty hefty amount of memory, but what about recursive function calls? Yea, everyone knows that in regular languages they are memory hogs, but in JASS?

Tx tx : ).


Oh, and check a bundle of added conditions to a trigger vs evaluating a bundle of triggers each with a single condition like so-

1 trigger has 10 conditions
10 triggers each have one condition


Also check TriggerClearConditions on 10 conditions vs TriggerRemoveCondition on 10 conditions. Compare trigger actions too.

Also check FlushParentHashtable, FlushChildHashtable, and RemoveSavedInteger in the same styles of comparisons.

Also do single comparisons, removing a single condition vs removing all conditions on a trigger that only has one condition.


: )

 

[weaaddar]

recursion is cheaper than the loop construct.

 

[Nestharus]

So in JASS, recursive calls are faster than plain loops? : o

Remember, this is JASS we're talking about here, not regular programming.

 

[Lyerae]

I think it would be, but is there a way to stop recursion after X times?

 

[Nestharus]

yes... -.-

you can use an if statement to prevent calling the function again ><, or a return