mirror of
git://gcc.gnu.org/git/gcc.git
synced 2025-03-07 08:37:11 +08:00
Fix PR42334: correct the update of the LST on loop interchange and distribution.
2009-12-15 Sebastian Pop <sebastian.pop@amd.com>
PR middle-end/42178
PR middle-end/42334
* graphite-interchange.c (lst_perfect_nestify): Reset to NULL the LSTs
that are empty.
(lst_do_interchange_1): Renamed lst_interchange_select_inner.
(lst_try_interchange): Reimplemented.
(lst_interchange_select_inner): Same.
(lst_do_interchange): Renamed lst_interchange_select_outer.
Reimplemented.
(scop_do_interchange): Update use of lst_interchange_select_outer.
* graphite-interchange.c (lst_try_interchange): Do not increment the
the OUTER index when there is no AFTER kernel. Do not increment the
OUTER index for after processing the AFTER kernel.
(lst_interchange_select_inner): Call lst_try_interchange only on loops.
(lst_interchange_select_outer): Do not pass in a pointer to the OUTER
index. Do not pass to lst_interchange_select_inner the OUTER index.
(scop_do_interchange): Update use of lst_interchange_select_outer.
* graphite-interchange.c (lst_try_interchange): Do not modify OUTER
index. Call lst_interchange_select_inner only once.
(lst_interchange_select_inner): Update use of lst_try_interchange.
(lst_interchange_select_outer): Update.
* testsuite/g++.dg/graphite/pr42130.C: Add -fgraphite-identity.
* testsuite/gcc.dg/graphite/block-0.c: Un-XFAILed.
* testsuite/gcc.dg/graphite/pr42211.c: New.
* testsuite/gfortran.dg/graphite/pr42334.f90: New.
* testsuite/gfortran.dg/graphite/graphite.exp
(DEFAULT_FLAGS_GRAPHITE_IDENTITY): Remove -fdump-tree-graphite-all.
* testsuite/gfortran.dg/graphite/interchange-1.f: Add comment. Clean
the graphite dump file.
* testsuite/gfortran.dg/graphite/interchange-2.f: Same.
* testsuite/gfortran.dg/graphite/pr42334-1.f: New.
From-SVN: r155418
This commit is contained in:
Sebastian Pop
Sebastian Pop
parent
4b216ab08a
commit
070ba48343
@ -1,3 +1,49 @@
|
||||
2009-12-17 Sebastian Pop <sebastian.pop@amd.com>
|
||||
|
||||
PR middle-end/42178
|
||||
PR middle-end/42334
|
||||
* graphite-interchange.c (lst_try_interchange): Do not modify OUTER
|
||||
index. Call lst_interchange_select_inner only once.
|
||||
(lst_interchange_select_inner): Update use of lst_try_interchange.
|
||||
(lst_interchange_select_outer): Update.
|
||||
|
||||
2009-12-15 Sebastian Pop <sebastian.pop@amd.com>
|
||||
|
||||
PR middle-end/42178
|
||||
PR middle-end/42334
|
||||
* graphite-interchange.c (lst_try_interchange): Do not increment the
|
||||
the OUTER index when there is no AFTER kernel. Do not increment the
|
||||
OUTER index for after processing the AFTER kernel.
|
||||
(lst_interchange_select_inner): Call lst_try_interchange only on loops.
|
||||
(lst_interchange_select_outer): Do not pass in a pointer to the OUTER
|
||||
index. Do not pass to lst_interchange_select_inner the OUTER index.
|
||||
(scop_do_interchange): Update use of lst_interchange_select_outer.
|
||||
|
||||
* testsuite/gfortran.dg/graphite/graphite.exp
|
||||
(DEFAULT_FLAGS_GRAPHITE_IDENTITY): Remove -fdump-tree-graphite-all.
|
||||
* testsuite/gfortran.dg/graphite/interchange-1.f: Add comment. Clean
|
||||
the graphite dump file.
|
||||
* testsuite/gfortran.dg/graphite/interchange-2.f: Same.
|
||||
* testsuite/gfortran.dg/graphite/pr42334-1.f: New.
|
||||
|
||||
2009-12-15 Sebastian Pop <sebastian.pop@amd.com>
|
||||
|
||||
PR middle-end/42178
|
||||
PR middle-end/42334
|
||||
* graphite-interchange.c (lst_perfect_nestify): Reset to NULL the LSTs
|
||||
that are empty.
|
||||
(lst_do_interchange_1): Renamed lst_interchange_select_inner.
|
||||
(lst_try_interchange): Reimplemented.
|
||||
(lst_interchange_select_inner): Same.
|
||||
(lst_do_interchange): Renamed lst_interchange_select_outer.
|
||||
Reimplemented.
|
||||
(scop_do_interchange): Update use of lst_interchange_select_outer.
|
||||
|
||||
* testsuite/g++.dg/graphite/pr42130.C: Add -fgraphite-identity.
|
||||
* testsuite/gcc.dg/graphite/block-0.c: Un-XFAILed.
|
||||
* testsuite/gcc.dg/graphite/pr42211.c: New.
|
||||
* testsuite/gfortran.dg/graphite/pr42334.f90: New.
|
||||
|
||||
2009-12-14 Sebastian Pop <sebastian.pop@amd.com>
|
||||
|
||||
PR middle-end/42181
|
||||
|
||||
@ -585,6 +585,13 @@ lst_perfect_nestify (lst_p loop1, lst_p loop2, lst_p *before,
|
||||
|
||||
lst_remove_all_before_excluding_pbb (*nest, first, true);
|
||||
lst_remove_all_before_excluding_pbb (*nest, last, false);
|
||||
|
||||
if (lst_empty_p (*before))
|
||||
*before = NULL;
|
||||
if (lst_empty_p (*after))
|
||||
*after = NULL;
|
||||
if (lst_empty_p (*nest))
|
||||
*nest = NULL;
|
||||
}
|
||||
|
||||
/* Try to interchange LOOP1 with LOOP2 for all the statements of the
|
||||
@ -650,92 +657,91 @@ lst_try_interchange_loops (scop_p scop, lst_p loop1, lst_p loop2,
|
||||
return false;
|
||||
}
|
||||
|
||||
static bool lst_do_interchange_1 (scop_p, lst_p, int *);
|
||||
static bool lst_interchange_select_inner (scop_p, lst_p, int, lst_p);
|
||||
|
||||
/* Try to interchange LOOP with all the loops contained in the body of
|
||||
LST. Return true if it did interchanged some loops. INDEX points
|
||||
to the next element to be processed by lst_do_interchange. */
|
||||
/* Try to interchange loop OUTER of LST_SEQ (OUTER_FATHER) with all
|
||||
the loop INNER and with all the loops contained in the body of
|
||||
INNER. Return true if it did interchanged some loops. */
|
||||
|
||||
static bool
|
||||
lst_try_interchange (scop_p scop, lst_p loop, lst_p lst, int *index)
|
||||
lst_try_interchange (scop_p scop, lst_p outer_father, int outer, lst_p inner)
|
||||
{
|
||||
int i;
|
||||
lst_p l;
|
||||
lst_p before, nest, after;
|
||||
bool res;
|
||||
lst_p loop1 = VEC_index (lst_p, LST_SEQ (outer_father), outer);
|
||||
lst_p loop2 = inner;
|
||||
|
||||
if (!lst || !LST_LOOP_P (lst))
|
||||
return false;
|
||||
gcc_assert (LST_LOOP_P (loop1)
|
||||
&& LST_LOOP_P (loop2));
|
||||
|
||||
res = lst_try_interchange_loops (scop, loop, lst, &before, &nest, &after);
|
||||
res = lst_try_interchange_loops (scop, loop1, loop2, &before, &nest, &after);
|
||||
|
||||
if (before)
|
||||
{
|
||||
res |= lst_do_interchange_1 (scop, before, index);
|
||||
(*index)++;
|
||||
}
|
||||
|
||||
if (nest)
|
||||
res |= lst_do_interchange_1 (scop, nest, index);
|
||||
res |= lst_interchange_select_inner (scop, outer_father, outer, before);
|
||||
else if (nest)
|
||||
res |= lst_interchange_select_inner (scop, outer_father, outer, nest);
|
||||
else
|
||||
for (i = 0; VEC_iterate (lst_p, LST_SEQ (lst), i, l); i++)
|
||||
res |= lst_try_interchange (scop, loop, l, index);
|
||||
res |= lst_interchange_select_inner (scop, outer_father, outer, loop2);
|
||||
|
||||
if (after)
|
||||
{
|
||||
res |= lst_do_interchange_1 (scop, after, index);
|
||||
(*index)++;
|
||||
}
|
||||
|
||||
(*index)++;
|
||||
return res;
|
||||
}
|
||||
|
||||
/* Interchanges all the loops of LOOP that are considered profitable
|
||||
to interchange. Return true if it did interchanged some loops.
|
||||
INDEX points to the next element to be processed by
|
||||
lst_do_interchange. */
|
||||
/* Selects the inner loop in LST_SEQ (INNER_FATHER) to be interchanged
|
||||
with the loop OUTER in LST_SEQ (OUTER_FATHER). */
|
||||
|
||||
static bool
|
||||
lst_do_interchange_1 (scop_p scop, lst_p loop, int *index)
|
||||
lst_interchange_select_inner (scop_p scop, lst_p outer_father, int outer,
|
||||
lst_p inner_father)
|
||||
{
|
||||
int i;
|
||||
lst_p l;
|
||||
bool res = false;
|
||||
int inner;
|
||||
|
||||
if (!loop || !LST_LOOP_P (loop))
|
||||
return false;
|
||||
gcc_assert (outer_father
|
||||
&& LST_LOOP_P (outer_father)
|
||||
&& LST_LOOP_P (VEC_index (lst_p, LST_SEQ (outer_father), outer))
|
||||
&& inner_father
|
||||
&& LST_LOOP_P (inner_father));
|
||||
|
||||
for (i = 0; VEC_iterate (lst_p, LST_SEQ (loop), i, l); i++)
|
||||
res |= lst_try_interchange (scop, loop, l, index);
|
||||
for (inner = 0; VEC_iterate (lst_p, LST_SEQ (inner_father), inner, l); inner++)
|
||||
if (LST_LOOP_P (l))
|
||||
res |= lst_try_interchange (scop, outer_father, outer, l);
|
||||
|
||||
return res;
|
||||
}
|
||||
|
||||
/* Interchanges all the loops of LOOP and the loops of its body that
|
||||
are considered profitable to interchange. Return true if it did
|
||||
interchanged some loops. INDEX points to the next element to be
|
||||
processed in the LST_SEQ (LOOP) vector. */
|
||||
interchanged some loops. OUTER is the index in LST_SEQ (LOOP) that
|
||||
points to the next outer loop to be considered for interchange. */
|
||||
|
||||
static bool
|
||||
lst_do_interchange (scop_p scop, lst_p loop, int *index)
|
||||
lst_interchange_select_outer (scop_p scop, lst_p loop, int outer)
|
||||
{
|
||||
lst_p l;
|
||||
bool res = false;
|
||||
int i = 0;
|
||||
lst_p father;
|
||||
|
||||
if (!loop || !LST_LOOP_P (loop))
|
||||
return false;
|
||||
|
||||
if (lst_depth (loop) >= 0)
|
||||
res = lst_do_interchange_1 (scop, loop, index);
|
||||
father = LST_LOOP_FATHER (loop);
|
||||
if (father)
|
||||
{
|
||||
res = lst_interchange_select_inner (scop, father, outer, loop);
|
||||
|
||||
while (VEC_iterate (lst_p, LST_SEQ (loop), *index, l))
|
||||
if (LST_LOOP_P (l))
|
||||
res |= lst_do_interchange (scop, l, index);
|
||||
else
|
||||
(*index)++;
|
||||
if (VEC_length (lst_p, LST_SEQ (father)) <= (unsigned) outer)
|
||||
return res;
|
||||
|
||||
loop = VEC_index (lst_p, LST_SEQ (father), outer);
|
||||
}
|
||||
|
||||
if (LST_LOOP_P (loop))
|
||||
for (i = 0; VEC_iterate (lst_p, LST_SEQ (loop), i, l); i++)
|
||||
if (LST_LOOP_P (l))
|
||||
res |= lst_interchange_select_outer (scop, l, i);
|
||||
|
||||
(*index)++;
|
||||
return res;
|
||||
}
|
||||
|
||||
@ -744,8 +750,8 @@ lst_do_interchange (scop_p scop, lst_p loop, int *index)
|
||||
bool
|
||||
scop_do_interchange (scop_p scop)
|
||||
{
|
||||
int i = 0;
|
||||
bool res = lst_do_interchange (scop, SCOP_TRANSFORMED_SCHEDULE (scop), &i);
|
||||
bool res = lst_interchange_select_outer
|
||||
(scop, SCOP_TRANSFORMED_SCHEDULE (scop), 0);
|
||||
|
||||
lst_update_scattering (SCOP_TRANSFORMED_SCHEDULE (scop));
|
||||
|
||||
|
||||
@ -1,4 +1,4 @@
|
||||
/* { dg-options "-O2 -fno-tree-ch" } */
|
||||
/* { dg-options "-O2 -fgraphite-identity -fno-tree-ch" } */
|
||||
#include <vector>
|
||||
|
||||
using std::vector;
|
||||
@ -9,11 +9,11 @@ vector<unsigned> & __attribute__((noinline, noclone)) foo(unsigned n)
|
||||
return *vv;
|
||||
}
|
||||
|
||||
|
||||
int main()
|
||||
{
|
||||
foo(0);
|
||||
return 0;
|
||||
}
|
||||
|
||||
/* { dg-do run } */
|
||||
|
||||
|
||||
@ -19,5 +19,5 @@ main()
|
||||
return toto();
|
||||
}
|
||||
|
||||
/* { dg-final { scan-tree-dump-times "will be loop blocked" 1 "graphite" { xfail *-*-* } } } */
|
||||
/* { dg-final { scan-tree-dump-times "will be loop blocked" 1 "graphite" } } */
|
||||
/* { dg-final { cleanup-tree-dump "graphite" } } */
|
||||
|
||||
22
gcc/testsuite/gcc.dg/graphite/pr42211.c
Normal file
22
gcc/testsuite/gcc.dg/graphite/pr42211.c
Normal file
@ -0,0 +1,22 @@
|
||||
/* { dg-options "-O3 -floop-interchange" } */
|
||||
|
||||
typedef unsigned char uint8_t;
|
||||
|
||||
void border_mirror(uint8_t *outer_img, int w, int h, int rb, int border)
|
||||
{
|
||||
uint8_t *img = outer_img + border * rb + border;
|
||||
int x, y;
|
||||
|
||||
for (y = -border; y < 0; y++) {
|
||||
for (x = -border; x < 0; x++)
|
||||
img[y*rb + x] = img[(-y)*rb + (-x)];
|
||||
|
||||
for (x = 0; x < w; x++)
|
||||
img[y*rb + x] = img[(-y)*rb + x];
|
||||
}
|
||||
}
|
||||
|
||||
void border_mirror_480(uint8_t *outer_img)
|
||||
{
|
||||
border_mirror(outer_img, 640, 480, 640 + 16*2, 16);
|
||||
}
|
||||
@ -49,7 +49,7 @@ gfortran-dg-runtest $block_files $DEFAULT_FLAGS_GRAPHITE_BLOCK
|
||||
foreach block_file $block_files {lremove wait_to_run_files $block_file}
|
||||
|
||||
# Flags using for id-* files.
|
||||
set DEFAULT_FLAGS_GRAPHITE_IDENTITY "-O2 -fgraphite-identity -fdump-tree-graphite-all"
|
||||
set DEFAULT_FLAGS_GRAPHITE_IDENTITY "-O2 -fgraphite-identity"
|
||||
set id_files [lsort [glob -nocomplain $srcdir/$subdir/id-*.\[fF\]{,90,95,03,08} ] ]
|
||||
gfortran-dg-runtest $id_files $DEFAULT_FLAGS_GRAPHITE_IDENTITY
|
||||
foreach id_file $id_files {lremove wait_to_run_files $id_file}
|
||||
|
||||
@ -37,4 +37,9 @@
|
||||
end
|
||||
|
||||
|
||||
! We should be able to interchange this as the number of iterations is
|
||||
! known to be 4 in the inner two loops. See interchange-2.f for the
|
||||
! kernel from bwaves.
|
||||
|
||||
! { dg-final { scan-tree-dump-times "will be interchanged" 1 "graphite" { xfail *-*-* } } }
|
||||
! { dg-final { cleanup-tree-dump "graphite" } }
|
||||
|
||||
@ -32,7 +32,12 @@
|
||||
enddo
|
||||
enddo
|
||||
enddo
|
||||
enddo
|
||||
enddo
|
||||
return
|
||||
end
|
||||
|
||||
! This is the kernel extracted from bwaves: this cannot be interchanged
|
||||
! as the number of iterations for f4 is not known.
|
||||
|
||||
! { dg-final { scan-tree-dump-times "will be interchanged" 0 "graphite" } }
|
||||
! { dg-final { cleanup-tree-dump "graphite" } }
|
||||
|
||||
16
gcc/testsuite/gfortran.dg/graphite/pr42334-1.f
Normal file
16
gcc/testsuite/gfortran.dg/graphite/pr42334-1.f
Normal file
@ -0,0 +1,16 @@
|
||||
! { dg-options "-O2 -floop-interchange" }
|
||||
|
||||
subroutine linel(icmdl,stre,anisox)
|
||||
real*8 stre(6),tkl(3,3),ekl(3,3),anisox(3,3,3,3)
|
||||
do m1=1,3
|
||||
do m2=1,m1
|
||||
do m3=1,3
|
||||
do m4=1,3
|
||||
tkl(m1,m2)=tkl(m1,m2)+
|
||||
& anisox(m1,m2,m3,m4)*ekl(m3,m4)
|
||||
enddo
|
||||
enddo
|
||||
enddo
|
||||
enddo
|
||||
stre(1)=tkl(1,1)
|
||||
end
|
||||
20
gcc/testsuite/gfortran.dg/graphite/pr42334.f90
Normal file
20
gcc/testsuite/gfortran.dg/graphite/pr42334.f90
Normal file
@ -0,0 +1,20 @@
|
||||
! { dg-options "-O2 -floop-interchange -ftree-loop-distribution" }
|
||||
|
||||
subroutine blockdis(bl1eg,bl2eg)
|
||||
implicit real*8 (a-h,o-z)
|
||||
parameter(nblo=300)
|
||||
common/str /mblo
|
||||
common/str2 /mel(nblo)
|
||||
dimension h(nblo,2,6),g(nblo,2,6)
|
||||
dimension bl1eg(nblo,2,6),bl2eg(nblo,2,6)
|
||||
do k=1,mblo
|
||||
jm=mel(k)
|
||||
do l=1,2
|
||||
do m=1,6
|
||||
bl1eg(k,l,m)=h(jm,l,m)
|
||||
bl2eg(k,l,m)=g(jm,l,m)
|
||||
enddo
|
||||
enddo
|
||||
enddo
|
||||
return
|
||||
end subroutine blockdis
|
||||
Loading…
Reference in New Issue
Block a user