This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
[graphite] Fix 453.povray from CPU2006 runtime test with -floop-block
- From: Sebastian Pop <sebpop at gmail dot com>
- To: GCC Patches <gcc-patches at gcc dot gnu dot org>, Richard Guenther <rguenther at suse dot de>
- Date: Fri, 16 Jan 2009 19:37:08 -0600
- Subject: [graphite] Fix 453.povray from CPU2006 runtime test with -floop-block
Hi,
This patch fixes the last runtime fail for CPU2006 on amd64-linux.
Loop blocking is applied only on perfect loop nests of depth >= 2.
I committed this patch to the graphite branch. This is still under
test on trunk with BOOT_CFLAGS="-g -O2 -fgraphite-identity
-floop-block". Okay for trunk once regstrap passes?
Thanks,
Sebastian Pop
--
AMD - GNU Tools
2009-01-16 Sebastian Pop <sebastian.pop@amd.com>
Tobias Grosser <tobi.grosser@amd.com>
* graphite.c (graphite_trans_scop_block): Do not block single
nested loops.
Index: graphite.c
===================================================================
--- graphite.c (revision 143412)
+++ graphite.c (working copy)
@@ -5964,7 +5964,7 @@ graphite_trans_scop_block (scop_p scop)
j++;
/* Found perfect loop nest. */
- if (last_nb_loops - j > 0)
+ if (last_nb_loops - j >= 2)
transform_done |= graphite_trans_loop_block (bbs, last_nb_loops - j);
VEC_free (graphite_bb_p, heap, bbs);