[Scilab-users] For loop - maximum size

[Calixte Denizet]

Hi Remi,

but the better way to do this is
timer();A=zeros(1,100000);for i=1:100000,A(i)=i;end;timer()

it is probably linked with how Scilab handles its memory

Calixte

On 06/11/2012 16:48, Tingsten wrote:
> Hello everybody,
>
> I assume this kind of talk about the effectiveness of the For loop in Scilab
> has been discussed many times but I was not able to find a recent topic
> about it....
>
> I am usually post-processing big csv files and by reading them I noticed
> some important time change depending of the architecture of the loop.
>
> So i did some very quick and simple tests with the following results :
>
> 1)timer();A=[];for i=1:100000,A=[A;i];,end;timer()
>   ans  =
>       16.598506
>
> 2)timer();B=[];for j=1:10,A=[];for
> i=1:10000,A=[A;i];,end;B=[B;A];,end;timer()
>   ans  =
>       1.0452067
>
> 3)timer();B=[];for j=1:100,A=[];for
> i=1:1000,A=[A;i];,end;B=[B;A];,end;timer()
>   ans  =
>       0.4992032
>
> 4)timer();B=[];for j=1:1000,A=[];for
> i=1:100,A=[A;i];,end;B=[B;A];,end;timer()
>   ans  =
>       0.9204059
>
> 5)timer();C=[];for h=1:10,B=[];for j=1:100,A=[];for
> i=1:100,A=[A;i];,end;B=[B;A];,end,C=[C;B];,end;timer()
>   ans  =
>       0.4524029
>
> It is probably explainable but interesting to know that 3) is 30 time better
> than 1).
> So I would think that it is totally forbidden to be efficient to do loop
>> 10000.
> Any comment or advice about that?
>
> Thx you a lot,
> Rémi Mercier
>
>
>
> --
> View this message in context: http://mailinglists.scilab.org/For-loop-maximum-size-tp4025172.html
> Sent from the Scilab users - Mailing Lists Archives mailing list archive at Nabble.com.
> _______________________________________________
> users mailing list
> users at lists.scilab.org
> http://lists.scilab.org/mailman/listinfo/users


-- 
Calixte Denizet
Software Development Engineer
-----------------------------------------------------------
Scilab Enterprises
143bis rue Yves Le Coz - 78000 Versailles, France
http://www.scilab-enterprises.com