[thirdparty/gcc.git] / gcc / tree-loop-linear.c

/* Linear Loop transforms
   Copyright (C) 2003, 2004, 2005, 2007, 2008, 2009
   Free Software Foundation, Inc.
   Contributed by Daniel Berlin <dberlin@dberlin.org>.

This file is part of GCC.

GCC is free software; you can redistribute it and/or modify it under
the terms of the GNU General Public License as published by the Free
Software Foundation; either version 3, or (at your option) any later
version.

GCC is distributed in the hope that it will be useful, but WITHOUT ANY
WARRANTY; without even the implied warranty of MERCHANTABILITY or
FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
for more details.

You should have received a copy of the GNU General Public License
along with GCC; see the file COPYING3.  If not see
<http://www.gnu.org/licenses/>.  */


#include "config.h"
#include "system.h"
#include "coretypes.h"
#include "tm.h"
#include "ggc.h"
#include "tree.h"
#include "target.h"

#include "rtl.h"
#include "basic-block.h"
#include "diagnostic.h"
#include "obstack.h"
#include "tree-flow.h"
#include "tree-dump.h"
#include "timevar.h"
#include "cfgloop.h"
#include "expr.h"
#include "optabs.h"
#include "tree-chrec.h"
#include "tree-data-ref.h"
#include "tree-scalar-evolution.h"
#include "tree-pass.h"
#include "lambda.h"

/* Linear loop transforms include any composition of interchange,
   scaling, skewing, and reversal.  They are used to change the
   iteration order of loop nests in order to optimize data locality of
   traversals, or remove dependences that prevent
   parallelization/vectorization/etc.  

   TODO: Determine reuse vectors/matrix and use it to determine optimal
   transform matrix for locality purposes.
   TODO: Completion of partial transforms.  */

/* Gather statistics for loop interchange.  LOOP is the loop being
   considered. The first loop in the considered loop nest is
   FIRST_LOOP, and consequently, the index of the considered loop is
   obtained by LOOP->DEPTH - FIRST_LOOP->DEPTH
   
   Initializes:
   - DEPENDENCE_STEPS the sum of all the data dependence distances
   carried by loop LOOP,

   - NB_DEPS_NOT_CARRIED_BY_LOOP the number of dependence relations
   for which the loop LOOP is not carrying any dependence,

   - ACCESS_STRIDES the sum of all the strides in LOOP.

   Example: for the following loop,

   | loop_1 runs 1335 times
   |   loop_2 runs 1335 times
   |     A[{{0, +, 1}_1, +, 1335}_2]
   |     B[{{0, +, 1}_1, +, 1335}_2]
   |   endloop_2
   |   A[{0, +, 1336}_1]
   | endloop_1

   gather_interchange_stats (in loop_1) will return 
   DEPENDENCE_STEPS = 3002
   NB_DEPS_NOT_CARRIED_BY_LOOP = 5
   ACCESS_STRIDES = 10694

   gather_interchange_stats (in loop_2) will return 
   DEPENDENCE_STEPS = 3000
   NB_DEPS_NOT_CARRIED_BY_LOOP = 7
   ACCESS_STRIDES = 8010
*/

static void
gather_interchange_stats (VEC (ddr_p, heap) *dependence_relations ATTRIBUTE_UNUSED,
			  VEC (data_reference_p, heap) *datarefs ATTRIBUTE_UNUSED,
			  struct loop *loop ATTRIBUTE_UNUSED,
			  struct loop *first_loop ATTRIBUTE_UNUSED,
			  unsigned int *dependence_steps ATTRIBUTE_UNUSED, 
			  unsigned int *nb_deps_not_carried_by_loop ATTRIBUTE_UNUSED, 
			  double_int *access_strides ATTRIBUTE_UNUSED)
{
  unsigned int i, j;
  struct data_dependence_relation *ddr;
  struct data_reference *dr;

  *dependence_steps = 0;
  *nb_deps_not_carried_by_loop = 0;
  *access_strides = double_int_zero;

  for (i = 0; VEC_iterate (ddr_p, dependence_relations, i, ddr); i++)
    {
      /* If we don't know anything about this dependence, or the distance
	 vector is NULL, or there is no dependence, then there is no reuse of
	 data.  */
      if (DDR_ARE_DEPENDENT (ddr) == chrec_dont_know
	  || DDR_ARE_DEPENDENT (ddr) == chrec_known
	  || DDR_NUM_DIST_VECTS (ddr) == 0)
	continue;

      for (j = 0; j < DDR_NUM_DIST_VECTS (ddr); j++)
	{
	  int dist = DDR_DIST_VECT (ddr, j)[loop_depth (loop) - loop_depth (first_loop)];

	  if (dist == 0)
	    (*nb_deps_not_carried_by_loop) += 1;

	  else if (dist < 0)
	    (*dependence_steps) += -dist;

	  else
	    (*dependence_steps) += dist;
	}
    }

  /* Compute the access strides.  */
  for (i = 0; VEC_iterate (data_reference_p, datarefs, i, dr); i++)
    {
      unsigned int it;
      tree ref = DR_REF (dr);
      gimple stmt = DR_STMT (dr);
      struct loop *stmt_loop = loop_containing_stmt (stmt);
      struct loop *inner_loop = first_loop->inner;

      if (inner_loop != stmt_loop 
	  && !flow_loop_nested_p (inner_loop, stmt_loop))
	continue;

      for (it = 0; it < DR_NUM_DIMENSIONS (dr); 
	   it++, ref = TREE_OPERAND (ref, 0))
	{
	  int num = am_vector_index_for_loop (DR_ACCESS_MATRIX (dr), loop->num);
	  int istride = AM_GET_ACCESS_MATRIX_ELEMENT (DR_ACCESS_MATRIX (dr), it, num);
	  tree array_size = TYPE_SIZE (TREE_TYPE (ref));
	  double_int dstride;

	  if (array_size == NULL_TREE 
	      || TREE_CODE (array_size) != INTEGER_CST)
	    continue;

	  dstride = double_int_mul (tree_to_double_int (array_size), 
				    shwi_to_double_int (istride));
	  (*access_strides) = double_int_add (*access_strides, dstride);
	}
    }
}

/* Attempt to apply interchange transformations to TRANS to maximize the
   spatial and temporal locality of the loop.  
   Returns the new transform matrix.  The smaller the reuse vector
   distances in the inner loops, the fewer the cache misses.
   FIRST_LOOP is the loop->num of the first loop in the analyzed loop
   nest.  */


static lambda_trans_matrix
try_interchange_loops (lambda_trans_matrix trans, 
		       unsigned int depth,		       
		       VEC (ddr_p, heap) *dependence_relations,
		       VEC (data_reference_p, heap) *datarefs,
		       struct loop *first_loop)
{
  bool res;
  struct loop *loop_i;
  struct loop *loop_j;
  unsigned int dependence_steps_i, dependence_steps_j;
  double_int access_strides_i, access_strides_j;
  double_int small, large, nb_iter;
  double_int l1_cache_size, l2_cache_size;
  int cmp;
  unsigned int nb_deps_not_carried_by_i, nb_deps_not_carried_by_j;
  struct data_dependence_relation *ddr;

  if (VEC_length (ddr_p, dependence_relations) == 0)
    return trans;

  /* When there is an unknown relation in the dependence_relations, we
     know that it is no worth looking at this loop nest: give up.  */
  ddr = VEC_index (ddr_p, dependence_relations, 0);
  if (ddr == NULL || DDR_ARE_DEPENDENT (ddr) == chrec_dont_know)
    return trans;

  l1_cache_size = uhwi_to_double_int (L1_CACHE_SIZE * 1024);
  l2_cache_size = uhwi_to_double_int (L2_CACHE_SIZE * 1024);

  /* LOOP_I is always the outer loop.  */
  for (loop_j = first_loop->inner; 
       loop_j; 
       loop_j = loop_j->inner)
    for (loop_i = first_loop; 
	 loop_depth (loop_i) < loop_depth (loop_j); 
	 loop_i = loop_i->inner)
      {
	gather_interchange_stats (dependence_relations, datarefs,
				  loop_i, first_loop,
				  &dependence_steps_i, 
				  &nb_deps_not_carried_by_i,
				  &access_strides_i);
	gather_interchange_stats (dependence_relations, datarefs,
				  loop_j, first_loop,
				  &dependence_steps_j, 
				  &nb_deps_not_carried_by_j, 
				  &access_strides_j);
	
	/* Heuristics for loop interchange profitability:

	   0. Don't transform if the smallest stride is larger than
	      the L2 cache, or if the largest stride multiplied by the
	      number of iterations is smaller than the L1 cache.

	   1. (spatial locality) Inner loops should have smallest
              dependence steps.

	   2. (spatial locality) Inner loops should contain more
	   dependence relations not carried by the loop.

	   3. (temporal locality) Inner loops should have smallest
	      array access strides.
	*/

	cmp = double_int_ucmp (access_strides_i, access_strides_j);
	small = cmp < 0 ? access_strides_i : access_strides_j;
	large = cmp < 0 ? access_strides_j : access_strides_i;

	if (double_int_ucmp (small, l2_cache_size) > 0)
	  continue;

	res = cmp < 0 ?
	  estimated_loop_iterations (loop_j, false, &nb_iter):
	  estimated_loop_iterations (loop_i, false, &nb_iter);
	large = double_int_mul (large, nb_iter);

	if (res && double_int_ucmp (large, l1_cache_size) < 0)
	  continue;

	if (dependence_steps_i < dependence_steps_j 
	    || nb_deps_not_carried_by_i > nb_deps_not_carried_by_j
	    || cmp < 0)
	  {
	    lambda_matrix_row_exchange (LTM_MATRIX (trans),
					loop_depth (loop_i) - loop_depth (first_loop),
					loop_depth (loop_j) - loop_depth (first_loop));
	    /* Validate the resulting matrix.  When the transformation
	       is not valid, reverse to the previous transformation.  */
	    if (!lambda_transform_legal_p (trans, depth, dependence_relations))
	      lambda_matrix_row_exchange (LTM_MATRIX (trans), 
					  loop_depth (loop_i) - loop_depth (first_loop), 
					  loop_depth (loop_j) - loop_depth (first_loop));
	  }
      }

  return trans;
}

/* Return the number of nested loops in LOOP_NEST, or 0 if the loops
   are not perfectly nested.  */

unsigned int
perfect_loop_nest_depth (struct loop *loop_nest)
{
  struct loop *temp;
  unsigned int depth = 1;

  /* If it's not a loop nest, we don't want it.  We also don't handle
     sibling loops properly, which are loops of the following form:

     | for (i = 0; i < 50; i++)
     |   {
     |     for (j = 0; j < 50; j++)
     |       {
     |        ...
     |       }
     |     for (j = 0; j < 50; j++)
     |       {
     |        ...
     |       }
     |   }
  */

  if (!loop_nest->inner || !single_exit (loop_nest))
    return 0;

  for (temp = loop_nest->inner; temp; temp = temp->inner)
    {
      /* If we have a sibling loop or multiple exit edges, jump ship.  */
      if (temp->next || !single_exit (temp))
	return 0;

      depth++;
    }

  return depth;
}

/* Perform a set of linear transforms on loops.  */

void
linear_transform_loops (void)
{
  bool modified = false;
  loop_iterator li;
  VEC(tree,heap) *oldivs = NULL;
  VEC(tree,heap) *invariants = NULL;
  VEC(tree,heap) *lambda_parameters = NULL;
  VEC(gimple,heap) *remove_ivs = VEC_alloc (gimple, heap, 3);
  struct loop *loop_nest;
  gimple oldiv_stmt;
  unsigned i;

  FOR_EACH_LOOP (li, loop_nest, 0)
    {
      unsigned int depth = 0;
      VEC (ddr_p, heap) *dependence_relations;
      VEC (data_reference_p, heap) *datarefs;
      
      lambda_loopnest before, after;
      lambda_trans_matrix trans;
      struct obstack lambda_obstack;
      struct loop *loop;
      VEC(loop_p,heap) *nest;

      depth = perfect_loop_nest_depth (loop_nest);
      if (depth == 0)
	continue;

      nest = VEC_alloc (loop_p, heap, 3);
      for (loop = loop_nest; loop; loop = loop->inner)
	VEC_safe_push (loop_p, heap, nest, loop);

      gcc_obstack_init (&lambda_obstack);
      VEC_truncate (tree, oldivs, 0);
      VEC_truncate (tree, invariants, 0);
      VEC_truncate (tree, lambda_parameters, 0);

      datarefs = VEC_alloc (data_reference_p, heap, 10);
      dependence_relations = VEC_alloc (ddr_p, heap, 10 * 10);
      if (!compute_data_dependences_for_loop (loop_nest, true, &datarefs,
					      &dependence_relations))
	goto free_and_continue;
      
      lambda_collect_parameters (datarefs, &lambda_parameters);
      if (!lambda_compute_access_matrices (datarefs, lambda_parameters, nest))
	goto free_and_continue;

      if (dump_file && (dump_flags & TDF_DETAILS))
	dump_ddrs (dump_file, dependence_relations);

      /* Build the transformation matrix.  */
      trans = lambda_trans_matrix_new (depth, depth);
      lambda_matrix_id (LTM_MATRIX (trans), depth);
      trans = try_interchange_loops (trans, depth, dependence_relations,
				     datarefs, loop_nest);

      if (lambda_trans_matrix_id_p (trans))
	{
	  if (dump_file)
	   fprintf (dump_file, "Won't transform loop. Optimal transform is the identity transform\n");
	  goto free_and_continue;
	}

      /* Check whether the transformation is legal.  */
      if (!lambda_transform_legal_p (trans, depth, dependence_relations))
	{
	  if (dump_file)
	    fprintf (dump_file, "Can't transform loop, transform is illegal:\n");
	  goto free_and_continue;
	}

      before = gcc_loopnest_to_lambda_loopnest (loop_nest, &oldivs,
                                                &invariants, &lambda_obstack);

      if (!before)
	goto free_and_continue;

      if (dump_file)
	{
	  fprintf (dump_file, "Before:\n");
	  print_lambda_loopnest (dump_file, before, 'i');
	}
  
      after = lambda_loopnest_transform (before, trans, &lambda_obstack);

      if (dump_file)
	{
	  fprintf (dump_file, "After:\n");
	  print_lambda_loopnest (dump_file, after, 'u');
	}

      lambda_loopnest_to_gcc_loopnest (loop_nest, oldivs, invariants,
				       &remove_ivs,
                                       after, trans, &lambda_obstack);
      modified = true;

      if (dump_file)
	fprintf (dump_file, "Successfully transformed loop.\n");

    free_and_continue:
      obstack_free (&lambda_obstack, NULL);
      free_dependence_relations (dependence_relations);
      free_data_refs (datarefs);
      VEC_free (loop_p, heap, nest);
    }

  for (i = 0; VEC_iterate (gimple, remove_ivs, i, oldiv_stmt); i++)
    remove_iv (oldiv_stmt);

  VEC_free (tree, heap, oldivs);
  VEC_free (tree, heap, invariants);
  VEC_free (gimple, heap, remove_ivs);
  scev_reset ();

  if (modified)
    rewrite_into_loop_closed_ssa (NULL, TODO_update_ssa_full_phi);
}
Commit	Line	Data
60cfcb79	1	/* Linear Loop transforms
12c697cd	2	Copyright (C) 2003, 2004, 2005, 2007, 2008, 2009
12c697cd	3	Free Software Foundation, Inc.
60cfcb79	4	Contributed by Daniel Berlin <dberlin@dberlin.org>.
	5
	6	This file is part of GCC.
	7
	8	GCC is free software; you can redistribute it and/or modify it under
	9	the terms of the GNU General Public License as published by the Free
8c4c00c1	10	Software Foundation; either version 3, or (at your option) any later
60cfcb79	11	version.
	12
	13	GCC is distributed in the hope that it will be useful, but WITHOUT ANY
	14	WARRANTY; without even the implied warranty of MERCHANTABILITY or
	15	FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
	16	for more details.
	17
	18	You should have received a copy of the GNU General Public License
8c4c00c1	19	along with GCC; see the file COPYING3. If not see
8c4c00c1	20	<http://www.gnu.org/licenses/>. */
60cfcb79	21
	22
	23	#include "config.h"
	24	#include "system.h"
	25	#include "coretypes.h"
	26	#include "tm.h"
60cfcb79	27	#include "ggc.h"
	28	#include "tree.h"
	29	#include "target.h"
	30
	31	#include "rtl.h"
	32	#include "basic-block.h"
	33	#include "diagnostic.h"
3026d558	34	#include "obstack.h"
60cfcb79	35	#include "tree-flow.h"
	36	#include "tree-dump.h"
	37	#include "timevar.h"
	38	#include "cfgloop.h"
	39	#include "expr.h"
	40	#include "optabs.h"
	41	#include "tree-chrec.h"
	42	#include "tree-data-ref.h"
	43	#include "tree-scalar-evolution.h"
	44	#include "tree-pass.h"
60cfcb79	45	#include "lambda.h"
	46
	47	/* Linear loop transforms include any composition of interchange,
	48	scaling, skewing, and reversal. They are used to change the
	49	iteration order of loop nests in order to optimize data locality of
	50	traversals, or remove dependences that prevent
	51	parallelization/vectorization/etc.
	52
	53	TODO: Determine reuse vectors/matrix and use it to determine optimal
	54	transform matrix for locality purposes.
	55	TODO: Completion of partial transforms. */
	56
331e13f9	57	/* Gather statistics for loop interchange. LOOP is the loop being
	58	considered. The first loop in the considered loop nest is
	59	FIRST_LOOP, and consequently, the index of the considered loop is
	60	obtained by LOOP->DEPTH - FIRST_LOOP->DEPTH
50caf588	61
	62	Initializes:
	63	- DEPENDENCE_STEPS the sum of all the data dependence distances
331e13f9	64	carried by loop LOOP,
50caf588	65
50caf588	66	- NB_DEPS_NOT_CARRIED_BY_LOOP the number of dependence relations
331e13f9	67	for which the loop LOOP is not carrying any dependence,
50caf588	68
331e13f9	69	- ACCESS_STRIDES the sum of all the strides in LOOP.
50caf588	70
	71	Example: for the following loop,
	72
	73	\| loop_1 runs 1335 times
	74	\| loop_2 runs 1335 times
	75	\| A[{{0, +, 1}_1, +, 1335}_2]
	76	\| B[{{0, +, 1}_1, +, 1335}_2]
	77	\| endloop_2
	78	\| A[{0, +, 1336}_1]
	79	\| endloop_1
	80
	81	gather_interchange_stats (in loop_1) will return
	82	DEPENDENCE_STEPS = 3002
	83	NB_DEPS_NOT_CARRIED_BY_LOOP = 5
	84	ACCESS_STRIDES = 10694
	85
	86	gather_interchange_stats (in loop_2) will return
	87	DEPENDENCE_STEPS = 3000
	88	NB_DEPS_NOT_CARRIED_BY_LOOP = 7
	89	ACCESS_STRIDES = 8010
84420feb	90	*/
60cfcb79	91
60cfcb79	92	static void
75a70cf9	93	gather_interchange_stats (VEC (ddr_p, heap) *dependence_relations ATTRIBUTE_UNUSED,
	94	VEC (data_reference_p, heap) *datarefs ATTRIBUTE_UNUSED,
	95	struct loop *loop ATTRIBUTE_UNUSED,
	96	struct loop *first_loop ATTRIBUTE_UNUSED,
	97	unsigned int *dependence_steps ATTRIBUTE_UNUSED,
	98	unsigned int *nb_deps_not_carried_by_loop ATTRIBUTE_UNUSED,
	99	double_int *access_strides ATTRIBUTE_UNUSED)
60cfcb79	100	{
1532ec98	101	unsigned int i, j;
41c7a324	102	struct data_dependence_relation *ddr;
41c7a324	103	struct data_reference *dr;
60cfcb79	104
50caf588	105	*dependence_steps = 0;
60cfcb79	106	*nb_deps_not_carried_by_loop = 0;
a75122be	107	*access_strides = double_int_zero;
50caf588	108
41c7a324	109	for (i = 0; VEC_iterate (ddr_p, dependence_relations, i, ddr); i++)
60cfcb79	110	{
84420feb	111	/* If we don't know anything about this dependence, or the distance
	112	vector is NULL, or there is no dependence, then there is no reuse of
	113	data. */
1532ec98	114	if (DDR_ARE_DEPENDENT (ddr) == chrec_dont_know
	115	\|\| DDR_ARE_DEPENDENT (ddr) == chrec_known
	116	\|\| DDR_NUM_DIST_VECTS (ddr) == 0)
84420feb	117	continue;
60cfcb79	118
1532ec98	119	for (j = 0; j < DDR_NUM_DIST_VECTS (ddr); j++)
1532ec98	120	{
9e3536f4	121	int dist = DDR_DIST_VECT (ddr, j)[loop_depth (loop) - loop_depth (first_loop)];
1532ec98	122
	123	if (dist == 0)
	124	(*nb_deps_not_carried_by_loop) += 1;
	125
	126	else if (dist < 0)
	127	(*dependence_steps) += -dist;
	128
	129	else
	130	(*dependence_steps) += dist;
	131	}
50caf588	132	}
	133
	134	/* Compute the access strides. */
41c7a324	135	for (i = 0; VEC_iterate (data_reference_p, datarefs, i, dr); i++)
50caf588	136	{
50caf588	137	unsigned int it;
28d4f5a1	138	tree ref = DR_REF (dr);
75a70cf9	139	gimple stmt = DR_STMT (dr);
50caf588	140	struct loop *stmt_loop = loop_containing_stmt (stmt);
331e13f9	141	struct loop *inner_loop = first_loop->inner;
28d4f5a1	142
331e13f9	143	if (inner_loop != stmt_loop
331e13f9	144	&& !flow_loop_nested_p (inner_loop, stmt_loop))
50caf588	145	continue;
28d4f5a1	146
	147	for (it = 0; it < DR_NUM_DIMENSIONS (dr);
	148	it++, ref = TREE_OPERAND (ref, 0))
50caf588	149	{
b79b3386	150	int num = am_vector_index_for_loop (DR_ACCESS_MATRIX (dr), loop->num);
b79b3386	151	int istride = AM_GET_ACCESS_MATRIX_ELEMENT (DR_ACCESS_MATRIX (dr), it, num);
28d4f5a1	152	tree array_size = TYPE_SIZE (TREE_TYPE (ref));
a75122be	153	double_int dstride;
28d4f5a1	154
b79b3386	155	if (array_size == NULL_TREE
28d4f5a1	156	\|\| TREE_CODE (array_size) != INTEGER_CST)
50caf588	157	continue;
28d4f5a1	158
a75122be	159	dstride = double_int_mul (tree_to_double_int (array_size),
b79b3386	160	shwi_to_double_int (istride));
a75122be	161	(access_strides) = double_int_add (access_strides, dstride);
50caf588	162	}
60cfcb79	163	}
	164	}
	165
84420feb	166	/* Attempt to apply interchange transformations to TRANS to maximize the
84420feb	167	spatial and temporal locality of the loop.
60cfcb79	168	Returns the new transform matrix. The smaller the reuse vector
50caf588	169	distances in the inner loops, the fewer the cache misses.
	170	FIRST_LOOP is the loop->num of the first loop in the analyzed loop
	171	nest. */
	172
60cfcb79	173
	174	static lambda_trans_matrix
	175	try_interchange_loops (lambda_trans_matrix trans,
	176	unsigned int depth,
41c7a324	177	VEC (ddr_p, heap) *dependence_relations,
41c7a324	178	VEC (data_reference_p, heap) *datarefs,
331e13f9	179	struct loop *first_loop)
60cfcb79	180	{
2fcf1fbb	181	bool res;
331e13f9	182	struct loop *loop_i;
331e13f9	183	struct loop *loop_j;
50caf588	184	unsigned int dependence_steps_i, dependence_steps_j;
a75122be	185	double_int access_strides_i, access_strides_j;
2fcf1fbb	186	double_int small, large, nb_iter;
	187	double_int l1_cache_size, l2_cache_size;
	188	int cmp;
60cfcb79	189	unsigned int nb_deps_not_carried_by_i, nb_deps_not_carried_by_j;
	190	struct data_dependence_relation *ddr;
	191
e0cf06e1	192	if (VEC_length (ddr_p, dependence_relations) == 0)
	193	return trans;
	194
60cfcb79	195	/* When there is an unknown relation in the dependence_relations, we
60cfcb79	196	know that it is no worth looking at this loop nest: give up. */
41c7a324	197	ddr = VEC_index (ddr_p, dependence_relations, 0);
60cfcb79	198	if (ddr == NULL \|\| DDR_ARE_DEPENDENT (ddr) == chrec_dont_know)
60cfcb79	199	return trans;
2fcf1fbb	200
	201	l1_cache_size = uhwi_to_double_int (L1_CACHE_SIZE * 1024);
	202	l2_cache_size = uhwi_to_double_int (L2_CACHE_SIZE * 1024);
	203
60cfcb79	204	/* LOOP_I is always the outer loop. */
331e13f9	205	for (loop_j = first_loop->inner;
	206	loop_j;
	207	loop_j = loop_j->inner)
	208	for (loop_i = first_loop;
9e3536f4	209	loop_depth (loop_i) < loop_depth (loop_j);
331e13f9	210	loop_i = loop_i->inner)
60cfcb79	211	{
50caf588	212	gather_interchange_stats (dependence_relations, datarefs,
	213	loop_i, first_loop,
	214	&dependence_steps_i,
	215	&nb_deps_not_carried_by_i,
	216	&access_strides_i);
	217	gather_interchange_stats (dependence_relations, datarefs,
	218	loop_j, first_loop,
	219	&dependence_steps_j,
	220	&nb_deps_not_carried_by_j,
	221	&access_strides_j);
60cfcb79	222
60cfcb79	223	/* Heuristics for loop interchange profitability:
50caf588	224
2fcf1fbb	225	0. Don't transform if the smallest stride is larger than
	226	the L2 cache, or if the largest stride multiplied by the
	227	number of iterations is smaller than the L1 cache.
	228
50caf588	229	1. (spatial locality) Inner loops should have smallest
	230	dependence steps.
	231
	232	2. (spatial locality) Inner loops should contain more
	233	dependence relations not carried by the loop.
	234
2fcf1fbb	235	3. (temporal locality) Inner loops should have smallest
50caf588	236	array access strides.
60cfcb79	237	*/
2fcf1fbb	238
	239	cmp = double_int_ucmp (access_strides_i, access_strides_j);
	240	small = cmp < 0 ? access_strides_i : access_strides_j;
	241	large = cmp < 0 ? access_strides_j : access_strides_i;
	242
	243	if (double_int_ucmp (small, l2_cache_size) > 0)
	244	continue;
	245
	246	res = cmp < 0 ?
	247	estimated_loop_iterations (loop_j, false, &nb_iter):
	248	estimated_loop_iterations (loop_i, false, &nb_iter);
	249	large = double_int_mul (large, nb_iter);
	250
	251	if (res && double_int_ucmp (large, l1_cache_size) < 0)
	252	continue;
	253
50caf588	254	if (dependence_steps_i < dependence_steps_j
50caf588	255	\|\| nb_deps_not_carried_by_i > nb_deps_not_carried_by_j
2fcf1fbb	256	\|\| cmp < 0)
60cfcb79	257	{
331e13f9	258	lambda_matrix_row_exchange (LTM_MATRIX (trans),
9e3536f4	259	loop_depth (loop_i) - loop_depth (first_loop),
9e3536f4	260	loop_depth (loop_j) - loop_depth (first_loop));
60cfcb79	261	/* Validate the resulting matrix. When the transformation
50caf588	262	is not valid, reverse to the previous transformation. */
60cfcb79	263	if (!lambda_transform_legal_p (trans, depth, dependence_relations))
331e13f9	264	lambda_matrix_row_exchange (LTM_MATRIX (trans),
9e3536f4	265	loop_depth (loop_i) - loop_depth (first_loop),
9e3536f4	266	loop_depth (loop_j) - loop_depth (first_loop));
60cfcb79	267	}
60cfcb79	268	}
50caf588	269
60cfcb79	270	return trans;
	271	}
	272
538dd0ca	273	/* Return the number of nested loops in LOOP_NEST, or 0 if the loops
	274	are not perfectly nested. */
	275
255b6be7	276	unsigned int
538dd0ca	277	perfect_loop_nest_depth (struct loop *loop_nest)
	278	{
	279	struct loop *temp;
	280	unsigned int depth = 1;
	281
	282	/* If it's not a loop nest, we don't want it. We also don't handle
	283	sibling loops properly, which are loops of the following form:
	284
	285	\| for (i = 0; i < 50; i++)
	286	\| {
	287	\| for (j = 0; j < 50; j++)
	288	\| {
	289	\| ...
	290	\| }
	291	\| for (j = 0; j < 50; j++)
	292	\| {
	293	\| ...
	294	\| }
	295	\| }
	296	*/
	297
	298	if (!loop_nest->inner \|\| !single_exit (loop_nest))
	299	return 0;
	300
	301	for (temp = loop_nest->inner; temp; temp = temp->inner)
	302	{
	303	/* If we have a sibling loop or multiple exit edges, jump ship. */
	304	if (temp->next \|\| !single_exit (temp))
	305	return 0;
	306
	307	depth++;
	308	}
	309
	310	return depth;
	311	}
	312
7194de72	313	/* Perform a set of linear transforms on loops. */
60cfcb79	314
60cfcb79	315	void
7194de72	316	linear_transform_loops (void)
60cfcb79	317	{
dfe210e4	318	bool modified = false;
17519ba0	319	loop_iterator li;
81bd7518	320	VEC(tree,heap) *oldivs = NULL;
81bd7518	321	VEC(tree,heap) *invariants = NULL;
b79b3386	322	VEC(tree,heap) *lambda_parameters = NULL;
75a70cf9	323	VEC(gimple,heap) *remove_ivs = VEC_alloc (gimple, heap, 3);
17519ba0	324	struct loop *loop_nest;
75a70cf9	325	gimple oldiv_stmt;
5aaf9e39	326	unsigned i;
3026d558	327
17519ba0	328	FOR_EACH_LOOP (li, loop_nest, 0)
60cfcb79	329	{
60cfcb79	330	unsigned int depth = 0;
41c7a324	331	VEC (ddr_p, heap) *dependence_relations;
41c7a324	332	VEC (data_reference_p, heap) *datarefs;
b79b3386	333
60cfcb79	334	lambda_loopnest before, after;
60cfcb79	335	lambda_trans_matrix trans;
3026d558	336	struct obstack lambda_obstack;
2e54c85d	337	struct loop *loop;
12c697cd	338	VEC(loop_p,heap) *nest;
3026d558	339
538dd0ca	340	depth = perfect_loop_nest_depth (loop_nest);
538dd0ca	341	if (depth == 0)
60cfcb79	342	continue;
538dd0ca	343
12c697cd	344	nest = VEC_alloc (loop_p, heap, 3);
2e54c85d	345	for (loop = loop_nest; loop; loop = loop->inner)
	346	VEC_safe_push (loop_p, heap, nest, loop);
	347
19af51e2	348	gcc_obstack_init (&lambda_obstack);
046bfc77	349	VEC_truncate (tree, oldivs, 0);
046bfc77	350	VEC_truncate (tree, invariants, 0);
b79b3386	351	VEC_truncate (tree, lambda_parameters, 0);
60cfcb79	352
41c7a324	353	datarefs = VEC_alloc (data_reference_p, heap, 10);
41c7a324	354	dependence_relations = VEC_alloc (ddr_p, heap, 10 * 10);
b79b3386	355	if (!compute_data_dependences_for_loop (loop_nest, true, &datarefs,
b79b3386	356	&dependence_relations))
19af51e2	357	goto free_and_continue;
b79b3386	358
b79b3386	359	lambda_collect_parameters (datarefs, &lambda_parameters);
2e54c85d	360	if (!lambda_compute_access_matrices (datarefs, lambda_parameters, nest))
19af51e2	361	goto free_and_continue;
41c7a324	362
60cfcb79	363	if (dump_file && (dump_flags & TDF_DETAILS))
41c7a324	364	dump_ddrs (dump_file, dependence_relations);
60cfcb79	365
60cfcb79	366	/* Build the transformation matrix. */
	367	trans = lambda_trans_matrix_new (depth, depth);
	368	lambda_matrix_id (LTM_MATRIX (trans), depth);
50caf588	369	trans = try_interchange_loops (trans, depth, dependence_relations,
331e13f9	370	datarefs, loop_nest);
50caf588	371
	372	if (lambda_trans_matrix_id_p (trans))
	373	{
	374	if (dump_file)
	375	fprintf (dump_file, "Won't transform loop. Optimal transform is the identity transform\n");
62761183	376	goto free_and_continue;
50caf588	377	}
60cfcb79	378
	379	/* Check whether the transformation is legal. */
	380	if (!lambda_transform_legal_p (trans, depth, dependence_relations))
	381	{
	382	if (dump_file)
	383	fprintf (dump_file, "Can't transform loop, transform is illegal:\n");
62761183	384	goto free_and_continue;
60cfcb79	385	}
41c7a324	386
7194de72	387	before = gcc_loopnest_to_lambda_loopnest (loop_nest, &oldivs,
3026d558	388	&invariants, &lambda_obstack);
41c7a324	389
60cfcb79	390	if (!before)
62761183	391	goto free_and_continue;
62761183	392
60cfcb79	393	if (dump_file)
	394	{
	395	fprintf (dump_file, "Before:\n");
	396	print_lambda_loopnest (dump_file, before, 'i');
	397	}
	398
3026d558	399	after = lambda_loopnest_transform (before, trans, &lambda_obstack);
41c7a324	400
60cfcb79	401	if (dump_file)
	402	{
	403	fprintf (dump_file, "After:\n");
	404	print_lambda_loopnest (dump_file, after, 'u');
	405	}
41c7a324	406
60cfcb79	407	lambda_loopnest_to_gcc_loopnest (loop_nest, oldivs, invariants,
5aaf9e39	408	&remove_ivs,
3026d558	409	after, trans, &lambda_obstack);
dfe210e4	410	modified = true;
41c7a324	411
557ef5d8	412	if (dump_file)
557ef5d8	413	fprintf (dump_file, "Successfully transformed loop.\n");
41c7a324	414
62761183	415	free_and_continue:
3026d558	416	obstack_free (&lambda_obstack, NULL);
60cfcb79	417	free_dependence_relations (dependence_relations);
60cfcb79	418	free_data_refs (datarefs);
2e54c85d	419	VEC_free (loop_p, heap, nest);
60cfcb79	420	}
41c7a324	421
75a70cf9	422	for (i = 0; VEC_iterate (gimple, remove_ivs, i, oldiv_stmt); i++)
5aaf9e39	423	remove_iv (oldiv_stmt);
5aaf9e39	424
81bd7518	425	VEC_free (tree, heap, oldivs);
81bd7518	426	VEC_free (tree, heap, invariants);
75a70cf9	427	VEC_free (gimple, heap, remove_ivs);
557ef5d8	428	scev_reset ();
dfe210e4	429
	430	if (modified)
	431	rewrite_into_loop_closed_ssa (NULL, TODO_update_ssa_full_phi);
60cfcb79	432	}