Re: [gfortran,patch] New option to use BLAS routines for matrix multiplication

[Brooks Moses <bmoses at stanford dot edu>]

FX Coudert wrote:
> > > Where is the documentation of the new options and this new feature?
> >
> > I'm currently writing it, and will post it ASAP. I should have  
> > stated that in my patch submission.
>
> Here is the doc patch. Tested with "make info". As I'm not a very  
> good writer (even in french, let alone english), alternate wordings  
> are welcome!

It looks good to me, with a couple of small comments:

> +@cindex -fblas-matmul-limit
> +@item -fblas-matmul-limit=@var{n}
> +Only significant when @command{-fexternal-blas} is in effect.
> +Matrix multiplication of matrices with size larger than (or equal to) @var{n}
> +will be performed by calls to BLAS functions, while others will be
> +handled by @command{gfortran} internal algorithms. If the matrices
> +involved are not square, the size comparison is performed using the
> +geometric mean of the three matrices dimensions.

The last sentence is a bit unclear -- it took me a moment to realize 
that you meant the dimensions of the three matrices involved (the two 
inputs and the result), rather than three dimensions of one matrix, and 
so I was trying to think of what the third one might be.  Perhaps 
"geometric mean of the dimensions of the argument and result matrices"?

Also, this isn't completely clear on what happens if the matmul-limit 
option isn't specified.  Is there a default value of n, or is n 
effectively zero as a default?

- Brooks