enable loop fusion on isl-15

* graphite-optimize-isl.c (optimize_isl): Call
       isl_options_set_schedule_maximize_band_depth.

       * gcc.dg/graphite/fuse-1.c: New.
       * gcc.dg/graphite/fuse-2.c: New.
       * gcc.dg/graphite/interchange-13.c: Remove bogus check.

Co-Authored-By: Sebastian Pop <s.pop@samsung.com>

From-SVN: r229889
This commit is contained in:
Aditya Kumar 2015-11-06 20:43:46 +00:00 committed by Sebastian Pop
parent 1167ebe707
commit 40856c71be
6 changed files with 103 additions and 5 deletions

View File

@ -1,11 +1,17 @@
2015-11-06 Aditya Kumar <aditya.k7@samsung.com>
Sebastian Pop <s.pop@samsung.com>
* graphite-scop-detection.c (scop_detection::merge_sese): Entry
* graphite-optimize-isl.c (optimize_isl): Call
isl_options_set_schedule_maximize_band_depth.
2015-11-06 Aditya Kumar <aditya.k7@samsung.com>
Sebastian Pop <s.pop@samsung.com>
* graphite-scop-detection.c (scop_detection::merge_sese): Entry
and exit edges should not be a part of irreducible loop.
(scop_detection::can_represent_loop_1): Loops should not be
(scop_detection::can_represent_loop_1): Loops should not be
irreducible.
(scop_detection::harmful_stmt_in_region): All the basic block
(scop_detection::harmful_stmt_in_region): All the basic block
should belong to reducible loops.
2015-11-06 Christophe Lyon <christophe.lyon@linaro.org>

View File

@ -404,7 +404,7 @@ optimize_isl (scop_p scop)
isl_options_set_schedule_maximize_band_depth (scop->isl_context, 1);
#ifdef HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS
/* ISL-0.15 or later. */
isl_options_set_schedule_serialize_sccs (scop->isl_context, 1);
isl_options_set_schedule_maximize_band_depth (scop->isl_context, 1);
#else
isl_options_set_schedule_fuse (scop->isl_context, ISL_SCHEDULE_FUSE_MIN);
#endif

View File

@ -1,3 +1,10 @@
2015-11-06 Aditya Kumar <aditya.k7@samsung.com>
Sebastian Pop <s.pop@samsung.com>
* gcc.dg/graphite/fuse-1.c: New.
* gcc.dg/graphite/fuse-2.c: New.
* gcc.dg/graphite/interchange-13.c: Remove bogus check.
2015-11-06 Christophe Lyon <christophe.lyon@linaro.org>
* gcc.target/aarch64/advsimd-intrinsics/vqtbX.c: New test.

View File

@ -0,0 +1,43 @@
/* Check that the two loops are fused and that we manage to fold the two xor
operations. */
/* { dg-options "-O2 -floop-nest-optimize -fdump-tree-forwprop-all" } */
/* { dg-do run } */
/* Make sure we fuse the loops like this:
ISL AST generated by ISL:
for (int c0 = 0; c0 <= 99; c0 += 1) {
S_3(c0);
S_6(c0);
S_9(c0);
}
*/
/* { dg-final { scan-tree-dump-times "ISL AST generated by ISL:.*for (int c0 = 0; c0 <= 99; c0 += 1) \{.*S_.*(c0);.*S_.*(c0);.*S_.*(c0);.*\}" 1 "graphite" } } */
/* Check that after fusing the loops, the scalar computation is also fused. */
/* { dg-final { scan-tree-dump-times "gimple_simplified to\[^\\n\]*\\^ 12" 1 "forwprop4" } } */
#define MAX 100
int A[MAX];
extern void abort ();
int
main (void)
{
int i;
for (i = 0; i < MAX; i++)
A[i] = i;
for(int i=0; i<MAX; i++)
A[i] ^= 4;
for(int i=0; i<MAX; i++)
A[i] ^= 8;
for (i = 0; i < MAX; i++)
if (A[i] != (i ^ 12))
abort ();
return 0;
}

View File

@ -0,0 +1,43 @@
/* Check that the three loops are fused. */
/* { dg-options "-O2 -floop-nest-optimize" } */
/* { dg-do run } */
/* Make sure we fuse the loops like this:
ISL AST generated by ISL:
for (int c0 = 0; c0 <= 99; c0 += 1) {
S_3(c0);
S_6(c0);
S_9(c0);
}
*/
/* { dg-final { scan-tree-dump-times "ISL AST generated by ISL:.*for (int c0 = 0; c0 <= 99; c0 += 1) \{.*S_.*(c0);.*S_.*(c0);.*S_.*(c0);.*\}" 1 "graphite" } } */
#define MAX 100
int A[MAX], B[MAX], C[MAX];
extern void abort ();
int
main (void)
{
int i;
/* The next three loops should be fused. */
for (i = 0; i < MAX; i++)
{
A[i] = i;
B[i] = i + 2;
C[i] = i + 1;
}
for(int i=0; i<MAX; i++)
A[i] += B[i];
for(int i=0; i<MAX; i++)
A[i] += C[i];
for (i = 0; i < MAX; i++)
if (A[i] != 3*i+3)
abort ();
return 0;
}

View File

@ -49,4 +49,3 @@ main (void)
return 0;
}
/* { dg-final { scan-tree-dump "tiled" "graphite" } } */