[Gambas-user] Wow, Gambas is twice as fast as (Free) Pascal

[Benoît Minisini]

Le 11/10/2014 18:32, T Lee Davidson a écrit :
>
> On 10/11/2014 08:52 AM, Benoît Minisini wrote:
>> You're right, I thought that the number of occurrences was printed, not
>> the result. Always look at the code before answering!:-)
>>
>> Well, as Fabien says, shows the Pascal code. I find strange that a
>> compiled Pascal program is twice slower than an interpreted Gambas
>> program. Or there is something weird in the Lazarus compiler?
>>
>> Regards,
>>
>> -- Benoît Minisini
>
> I thought it was strange too, Benoît. But, at the same time, also
> thought it was highly impressive on Gambas' (and its developers')
> account. :-)
>
>
> I have tried compiling the Pascal program with {$OPTIMIZATION ON} and
> also {$RANGECHECKS OFF} (even though that should be the default) with no
> improvement in performance.
>
> Here's the Pascal code:
>
> program Polynom;
> {$mode objfpc}
>
> var
>     z : integer;
>
> function DoIt(x : double) : double;
>     var Mu : double = 10.0;
>     var Pu, Su : double;
>     var I, J, N : integer;
>     var aPoly : array [0..99] of double;
>
> begin
>     N := 500000;
>     Pu := 0;
>
>     For I := 0 To N-1 do
>     begin
>       For J := 0 To 99 do
>       begin
>         Mu :=  (Mu + 2.0) / 2.0;
>         aPoly[J] := Mu;
>       end;
>       Su := 0.0;
>       For J := 0 To 99 do
>       begin
>         Su := X * Su + aPoly[J];
>       end;
>       Pu := Pu + Su;
>     end;
>
>     DoIt := Pu;
> end;
>
> Begin
>
> For z := 1 To 10 do
> begin
> 	writeln( DoIt(0.2) );
> end;
>
> End.
>

It seems to be a perfect equivalent. So Gambas is faster, cool...

If you succeed in converting the other benchmarks, I will add them to 
the wiki.

Regards,

-- 
Benoît Minisini