This is the mail archive of the
gcc-bugs@gcc.gnu.org
mailing list for the GCC project.
[Bug libfortran/51119] MATMUL slow for large matrices
- From: "Joost.VandeVondele at mat dot ethz.ch" <gcc-bugzilla at gcc dot gnu dot org>
- To: gcc-bugs at gcc dot gnu dot org
- Date: Fri, 29 Mar 2013 08:47:39 +0000
- Subject: [Bug libfortran/51119] MATMUL slow for large matrices
- Auto-submitted: auto-generated
- References: <bug-51119-4 at http dot gcc dot gnu dot org/bugzilla/>
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=51119
Joost VandeVondele <Joost.VandeVondele at mat dot ethz.ch> changed:
What |Removed |Added
----------------------------------------------------------------------------
Last reconfirmed|2011-11-14 00:00:00 |2013-03-29
--- Comment #10 from Joost VandeVondele <Joost.VandeVondele at mat dot ethz.ch> 2013-03-29 08:47:39 UTC ---
What about compiling the fortran runtime library with vectorization, and all
the fancy options that come with graphite (loop-blocking in particular). If
they don't work for a matrix multiplication pattern .... what's their use ?
Further naivety would be to provide an lto'ed runtime, allowing matrix
multiplication to be inlined for known small bounds ... kind of the ultimate
dogfooding ?