Re: Polyhedron shootout between g95 and gfortran

[Paul Thomas <paulthomas2 at wanadoo dot fr>]

Janne and Dominique,

> IMHO a good solution would be to always inline (thus getting rid of
> dot_product in libgfortran), and have an option to use an optimized
> BLAS version for large arrays (with a user-selectable cross-over
> switch to override the default) if applicable (see PR26025).
>  
If one of or both of you would like to implement this, then please do 
so.  The inline code is ready to go and works with anything, as far as I 
have been able to determine.  The standard allows to override an 
intrinsic with a local version that has the same type and type 
properties, as the attached shows. So I am not sure that I see the 
purpose in a switch.

> One reason to keep the libgfortran library version would be if we'd
> like to avoid inlining for -Os, but I suspect that issue won't matter
> anyway considering how Fortran is used in practice.
>
>  
Let's scrub the library version.  It is not worth worrying about Os; 
especially when the performance for large arrays will be more or less 
the same.

Anyway, I have more on my plate than I can cope with right now, so 
please insert my offering of a few months back and do the necessary tidy 
up in the library and the makefile.

Paul