Backport from mainline
2013-03-13 Richard Biener <rguenther@suse.de>
PR tree-optimization/56608
* tree-vect-slp.c (vect_schedule_slp): Do not remove scalar
calls when vectorizing basic-blocks.
* gcc.dg/vect/fast-math-bb-slp-call-3.c: New testcase.
2013-03-05 Richard Biener <rguenther@suse.de>
PR tree-optimization/56270
* tree-vect-slp.c (vect_schedule_slp): Clear vectorized stmts
of loads after scheduling an SLP instance.
* gcc.dg/vect/slp-38.c: New testcase.
git-svn-id: svn+ssh://gcc.gnu.org/svn/gcc/branches/gcc-4_7-branch@197096
138bc75d-0d04-0410-961f-
82ee72b054a4
+2013-03-26 Richard Biener <rguenther@suse.de>
+
+ Backport from mainline:
+ 2013-03-13 Richard Biener <rguenther@suse.de>
+
+ PR tree-optimization/56608
+ * tree-vect-slp.c (vect_schedule_slp): Do not remove scalar
+ calls when vectorizing basic-blocks.
+
+ 2013-03-05 Richard Biener <rguenther@suse.de>
+
+ PR tree-optimization/56270
+ * tree-vect-slp.c (vect_schedule_slp): Clear vectorized stmts
+ of loads after scheduling an SLP instance.
+
2013-03-26 Walter Lee <walt@tilera.com>
Backport from mainline:
+2013-03-26 Richard Biener <rguenther@suse.de>
+
+ Backport from mainline
+ 2013-03-13 Richard Biener <rguenther@suse.de>
+
+ PR tree-optimization/56608
+ * gcc.dg/vect/fast-math-bb-slp-call-3.c: New testcase.
+
+ 2013-03-05 Richard Biener <rguenther@suse.de>
+
+ PR tree-optimization/56270
+ * gcc.dg/vect/slp-38.c: New testcase.
+
2013-03-25 Oleg Endo <olegendo@gcc.gnu.org>
Backport from mainline:
--- /dev/null
+#include <stdlib.h>
+#include <math.h>
+
+#define MIN(a, b) (((a) < (b)) ? (a) : (b))
+#define MAX(a, b) (((a) > (b)) ? (a) : (b))
+
+typedef struct {
+ int initialHeight, initialWidth;
+ int rotatedHeight, rotatedWidth;
+ int autoCropHeight, autoCropWidth;
+} ufraw_data;
+
+void __attribute__((noinline,noclone))
+ufraw_test(ufraw_data *uf)
+{
+ int iWidth = uf->initialWidth;
+ int iHeight = uf->initialHeight;
+ double aspectRatio = ((double)iWidth) / iHeight;
+ double midX = iWidth / 2.0 - 0.5;
+ double midY = iHeight / 2.0 - 0.5;
+ double maxX = 0, maxY = 0;
+ double minX = 999999, minY = 999999;
+ double lastX = 0, lastY = 0, area = 0;
+ double scale;
+ int i;
+ for (i = 0; i < iWidth + iHeight - 1; i++)
+ {
+ int x, y;
+ if (i < iWidth) { // Trace the left border of the image
+ x = i;
+ y = 0;
+ } else { // Trace the bottom border of the image
+ x = iWidth - 1;
+ y = i - iWidth + 1;
+ }
+ double srcX = x - midX;
+ double srcY = y - midY;
+ // A digital planimeter:
+ area += srcY * lastX - srcX * lastY;
+ lastX = srcX;
+ lastY = srcY;
+ maxX = MAX(maxX, fabs(srcX));
+ maxY = MAX(maxY, fabs(srcY));
+ if (fabs(srcX / srcY) > aspectRatio)
+ minX = MIN(minX, fabs(srcX));
+ else
+ minY = MIN(minY, fabs(srcY));
+ }
+ scale = sqrt((iWidth - 1) * (iHeight - 1) / area);
+ uf->rotatedWidth = MIN(ceil(2 * maxX + 1.0) * scale, 2 * iWidth);
+ uf->rotatedHeight = MIN(ceil(2 * maxY + 1.0) * scale, 2 * iHeight);
+ uf->autoCropWidth = MIN(floor(2 * minX) * scale, 2 * iWidth);
+ uf->autoCropHeight = MIN(floor(2 * minY) * scale, 2 * iHeight);
+ if (uf->autoCropWidth != 3)
+ abort ();
+}
+
+int main()
+{
+ ufraw_data uf_data;
+ ufraw_data *uf = &uf_data;
+ uf->initialWidth = 4;
+ uf->initialHeight = 5;
+ ufraw_test(uf);
+ return 0;
+}
+
+/* { dg-final { cleanup-tree-dump "slp" } } */
--- /dev/null
+/* { dg-do compile } */
+
+typedef struct {
+ float l, h;
+} tFPinterval;
+
+tFPinterval X[1024];
+tFPinterval Y[1024];
+tFPinterval Z[1024];
+
+void Compute(void)
+{
+ int d;
+ for (d= 0; d < 1024; d++)
+ {
+ Y[d].l= X[d].l + X[d].h;
+ Y[d].h= Y[d].l;
+ Z[d].l= X[d].l;
+ Z[d].h= X[d].h;
+ }
+}
+
+/* { dg-final { scan-tree-dump-times "vectorized 1 loops in function" 1 "vect" { target { vect_float && vect_perm } } } } */
+/* { dg-final { cleanup-tree-dump "vect" } } */
{
VEC (slp_instance, heap) *slp_instances;
slp_instance instance;
- unsigned int i, vf;
+ slp_tree loads_node;
+ unsigned int i, j, vf;
bool is_store = false;
if (loop_vinfo)
/* Schedule the tree of INSTANCE. */
is_store = vect_schedule_slp_instance (SLP_INSTANCE_TREE (instance),
instance, vf);
+
+ /* Clear STMT_VINFO_VEC_STMT of all loads. With shared loads
+ between SLP instances we fail to properly initialize the
+ vectorized SLP stmts and confuse different load permutations. */
+ FOR_EACH_VEC_ELT (slp_tree, SLP_INSTANCE_LOADS (instance), j, loads_node)
+ STMT_VINFO_VEC_STMT
+ (vinfo_for_stmt
+ (VEC_index (gimple, SLP_TREE_SCALAR_STMTS (loads_node), 0))) = NULL;
+
if (vect_print_dump_info (REPORT_VECTORIZED_LOCATIONS)
|| vect_print_dump_info (REPORT_UNVECTORIZED_LOCATIONS))
fprintf (vect_dump, "vectorizing stmts using SLP.");
unsigned int j;
gimple_stmt_iterator gsi;
- vect_remove_slp_scalar_calls (root);
+ /* Remove scalar call stmts. Do not do this for basic-block
+ vectorization as not all uses may be vectorized.
+ ??? Why should this be necessary? DCE should be able to
+ remove the stmts itself.
+ ??? For BB vectorization we can as well remove scalar
+ stmts starting from the SLP tree root if they have no
+ uses. */
+ if (loop_vinfo)
+ vect_remove_slp_scalar_calls (root);
for (j = 0; VEC_iterate (gimple, SLP_TREE_SCALAR_STMTS (root), j, store)
&& j < SLP_INSTANCE_GROUP_SIZE (instance); j++)