https://gcc.gnu.org/bugzilla/show_bug.cgi?id=51119
--- Comment #13 from Thomas Koenig <tkoenig at gcc dot gnu.org> --- (In reply to Dominique d'Humieres from comment #12) > I suppose most modern OS provide such optimized BLAS and, if not, one can > install libraries such as atlas. So I wonder if it would not be more > effective to be able to configure with something such as --with-blas="magic > incantation" and use -fexternal-blas as the default rather than reinventing > the wheel. If -fexternal-blas is supplied, the current implementation defaults to -fblas-matmul-limit=30, which in turn sets -finline-matmul-limit=30 (which is fairly reasonable for the point where an external, optimized BLAS and inlining are equally fast). I would be interested to see where threading moves this intersection.
