[Scilab-users] Scilab 6 bad performances

[Stéphane Mottelet]

Le 19/02/2016 11:41, Samuel Gougeon a écrit :
> Le 19/02/2016 11:31, Stéphane Mottelet a écrit :
>> Le 19/02/2016 11:17, Samuel Gougeon a écrit :
>>> .../...
>>> bench_run() is designed for that: 
>>> https://help.scilab.org/docs/5.5.2/en_US/bench_run.html
>>> On the same computer, you may run it on 5.5.2 for chosen modules, 
>>> run it as well on 6.0, and compare results.
>>>
>> This looks like finding a needle in a haystack...
> .
> Right. Since bench_run() does not return its results -- just displays 
> them --, they can't be post-processed to make a diff and filter it for 
> instance to get only the slower instructions.
> However, bench_run.sci can be edited and modified to do so.
at first sight, I already see the following problem (OSX)

Scilab 6

001/001 - [linear_algebra] bench_inv ....................... 2281.91 ms

Scilab 5.5.2

001/001 - [linear_algebra] bench_inv ....................... 871.99 ms

I never compute inverses or solve full linear systems in my program, but 
seeing this could orient me...

S.
>
> SG
>
> _______________________________________________
> users mailing list
> users at lists.scilab.org
> http://lists.scilab.org/mailman/listinfo/users


-- 
Département de Génie Informatique
EA 4297 Transformations Intégrées de la Matière Renouvelable
Université de Technologie de Compiègne -  CS 60319
60203 Compiègne cedex