This is the mail archive of the mailing list for the GCC project.

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [wwwdocs] Add "-fexternal-blas" to gcc-4.3/changes.html

On Mon, Oct 23, 2006 at 11:05:02AM +0200, Tobias Burnus wrote:
> Richard Guenther wrote:
> > Does it work with BLAS only or with any lapack style library?  In this
> > case at least the documentation should read like "calls to a linear
> > algebra library
> > like BLAS or LAPACK for matrix".  Of course in that case I also
> > dislike the option name, but
> > I don't have a particular better idea either.


> But if you think that many people think of only Netlib's BLAS when one
> uses the term "BLAS", we indeed need to find a better wording.

My impression of the situation is that the name "BLAS" is a general name
for libraries implmementing a specified set of routines, rather than a
specific implementation -- just as "Fortran" is a name of a language
rather than a specific compiler.  "Netlib's BLAS" is often referred to
as "the reference implmementation".  Since that implmementation is the
one that defines what routines a BLAS library is supposed to implement
and how they are supposed to behave, it could perhaps be considered to
be a little more BLAS than other BLASes.  OTOH, the very idea, if I have
understood it correctly, is that that implementation is just a reference
implementation, i.e. a kind of a standard expressed as code, and that
others can (and are encouraged to) provide better optimized
implementations.  Therefore, both the wording of the docs and the
option are IMHO OK.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]