[Bug middle-end/41193] [4.5 Regression] slow compilation with graphite / ice with graphite

[jv244 at cam dot ac dot uk]

------- Comment #6 from jv244 at cam dot ac dot uk  2009-09-18 15:18 -------
(In reply to comment #5)
> If possible please update the performance numbers with recent trunk.

much faster now at -O1 ;-)

fortran -fgraphite -c -O1 -ffast-math -funroll-loops -ftree-vectorize
-march=native -ftime-report PR41193.f90
PR41193.f90: In function ‘__spherical_harmonics_MOD_cvy_lm’:
PR41193.f90:164:0: internal compiler error: in scan_tree_for_params, at
graphite-sese-to-poly.c:826
Please submit a full bug report,
with preprocessed source if appropriate.
See <http://gcc.gnu.org/bugs.html> for instructions.

-O2 is now fast (7s)

-O3 is still very slow (381s)

gfortran -fgraphite -c -O3 -ffast-math -funroll-loops -ftree-vectorize
-march=native -ftime-report PR41193.f90


-- 

jv244 at cam dot ac dot uk changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|                            |ice-on-valid-code
            Summary|[4.5 Regression] slow       |[4.5 Regression] slow
                   |compilation with graphite   |compilation with graphite /
                   |                            |ice with graphite


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=41193