2 Copyright (C) 2005, 2007, 2008, 2009, 2010, 2011
3 Free Software Foundation, Inc.
5 This file is part of GCC.
7 GCC is free software; you can redistribute it and/or modify it
8 under the terms of the GNU General Public License as published by the
9 Free Software Foundation; either version 3, or (at your option) any
12 GCC is distributed in the hope that it will be useful, but WITHOUT
13 ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
14 FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
17 You should have received a copy of the GNU General Public License
18 along with GCC; see the file COPYING3. If not see
19 <http://www.gnu.org/licenses/>. */
23 #include "coretypes.h"
27 #include "basic-block.h"
29 #include "tree-pretty-print.h"
30 #include "tree-flow.h"
31 #include "tree-dump.h"
34 #include "tree-pass.h"
35 #include "insn-config.h"
38 #include "tree-chrec.h"
39 #include "tree-scalar-evolution.h"
40 #include "diagnostic-core.h"
42 #include "langhooks.h"
43 #include "tree-inline.h"
44 #include "tree-data-ref.h"
47 /* FIXME: Needed for optabs, but this should all be moved to a TBD interface
48 between the GIMPLE and RTL worlds. */
52 /* This pass inserts prefetch instructions to optimize cache usage during
53 accesses to arrays in loops. It processes loops sequentially and:
55 1) Gathers all memory references in the single loop.
56 2) For each of the references it decides when it is profitable to prefetch
57 it. To do it, we evaluate the reuse among the accesses, and determines
58 two values: PREFETCH_BEFORE (meaning that it only makes sense to do
59 prefetching in the first PREFETCH_BEFORE iterations of the loop) and
60 PREFETCH_MOD (meaning that it only makes sense to prefetch in the
61 iterations of the loop that are zero modulo PREFETCH_MOD). For example
62 (assuming cache line size is 64 bytes, char has size 1 byte and there
63 is no hardware sequential prefetch):
66 for (i = 0; i < max; i++)
73 a[187*i + 50] = ...; (5)
76 (0) obviously has PREFETCH_BEFORE 1
77 (1) has PREFETCH_BEFORE 64, since (2) accesses the same memory
78 location 64 iterations before it, and PREFETCH_MOD 64 (since
79 it hits the same cache line otherwise).
80 (2) has PREFETCH_MOD 64
81 (3) has PREFETCH_MOD 4
82 (4) has PREFETCH_MOD 1. We do not set PREFETCH_BEFORE here, since
83 the cache line accessed by (5) is the same with probability only
85 (5) has PREFETCH_MOD 1 as well.
87 Additionally, we use data dependence analysis to determine for each
88 reference the distance till the first reuse; this information is used
89 to determine the temporality of the issued prefetch instruction.
91 3) We determine how much ahead we need to prefetch. The number of
92 iterations needed is time to fetch / time spent in one iteration of
93 the loop. The problem is that we do not know either of these values,
94 so we just make a heuristic guess based on a magic (possibly)
95 target-specific constant and size of the loop.
97 4) Determine which of the references we prefetch. We take into account
98 that there is a maximum number of simultaneous prefetches (provided
99 by machine description). We prefetch as many prefetches as possible
100 while still within this bound (starting with those with lowest
101 prefetch_mod, since they are responsible for most of the cache
104 5) We unroll and peel loops so that we are able to satisfy PREFETCH_MOD
105 and PREFETCH_BEFORE requirements (within some bounds), and to avoid
106 prefetching nonaccessed memory.
107 TODO -- actually implement peeling.
109 6) We actually emit the prefetch instructions. ??? Perhaps emit the
110 prefetch instructions with guards in cases where 5) was not sufficient
111 to satisfy the constraints?
113 A cost model is implemented to determine whether or not prefetching is
114 profitable for a given loop. The cost model has three heuristics:
116 1. Function trip_count_to_ahead_ratio_too_small_p implements a
117 heuristic that determines whether or not the loop has too few
118 iterations (compared to ahead). Prefetching is not likely to be
119 beneficial if the trip count to ahead ratio is below a certain
122 2. Function mem_ref_count_reasonable_p implements a heuristic that
123 determines whether the given loop has enough CPU ops that can be
124 overlapped with cache missing memory ops. If not, the loop
125 won't benefit from prefetching. In the implementation,
126 prefetching is not considered beneficial if the ratio between
127 the instruction count and the mem ref count is below a certain
130 3. Function insn_to_prefetch_ratio_too_small_p implements a
131 heuristic that disables prefetching in a loop if the prefetching
132 cost is above a certain limit. The relative prefetching cost is
133 estimated by taking the ratio between the prefetch count and the
134 total intruction count (this models the I-cache cost).
136 The limits used in these heuristics are defined as parameters with
137 reasonable default values. Machine-specific default values will be
141 -- write and use more general reuse analysis (that could be also used
142 in other cache aimed loop optimizations)
143 -- make it behave sanely together with the prefetches given by user
144 (now we just ignore them; at the very least we should avoid
145 optimizing loops in that user put his own prefetches)
146 -- we assume cache line size alignment of arrays; this could be
149 /* Magic constants follow. These should be replaced by machine specific
152 /* True if write can be prefetched by a read prefetch. */
154 #ifndef WRITE_CAN_USE_READ_PREFETCH
155 #define WRITE_CAN_USE_READ_PREFETCH 1
158 /* True if read can be prefetched by a write prefetch. */
160 #ifndef READ_CAN_USE_WRITE_PREFETCH
161 #define READ_CAN_USE_WRITE_PREFETCH 0
164 /* The size of the block loaded by a single prefetch. Usually, this is
165 the same as cache line size (at the moment, we only consider one level
166 of cache hierarchy). */
168 #ifndef PREFETCH_BLOCK
169 #define PREFETCH_BLOCK L1_CACHE_LINE_SIZE
172 /* Do we have a forward hardware sequential prefetching? */
174 #ifndef HAVE_FORWARD_PREFETCH
175 #define HAVE_FORWARD_PREFETCH 0
178 /* Do we have a backward hardware sequential prefetching? */
180 #ifndef HAVE_BACKWARD_PREFETCH
181 #define HAVE_BACKWARD_PREFETCH 0
184 /* In some cases we are only able to determine that there is a certain
185 probability that the two accesses hit the same cache line. In this
186 case, we issue the prefetches for both of them if this probability
187 is less then (1000 - ACCEPTABLE_MISS_RATE) per thousand. */
189 #ifndef ACCEPTABLE_MISS_RATE
190 #define ACCEPTABLE_MISS_RATE 50
193 #ifndef HAVE_prefetch
194 #define HAVE_prefetch 0
197 #define L1_CACHE_SIZE_BYTES ((unsigned) (L1_CACHE_SIZE * 1024))
198 #define L2_CACHE_SIZE_BYTES ((unsigned) (L2_CACHE_SIZE * 1024))
200 /* We consider a memory access nontemporal if it is not reused sooner than
201 after L2_CACHE_SIZE_BYTES of memory are accessed. However, we ignore
202 accesses closer than L1_CACHE_SIZE_BYTES / NONTEMPORAL_FRACTION,
203 so that we use nontemporal prefetches e.g. if single memory location
204 is accessed several times in a single iteration of the loop. */
205 #define NONTEMPORAL_FRACTION 16
207 /* In case we have to emit a memory fence instruction after the loop that
208 uses nontemporal stores, this defines the builtin to use. */
210 #ifndef FENCE_FOLLOWING_MOVNT
211 #define FENCE_FOLLOWING_MOVNT NULL_TREE
214 /* It is not profitable to prefetch when the trip count is not at
215 least TRIP_COUNT_TO_AHEAD_RATIO times the prefetch ahead distance.
216 For example, in a loop with a prefetch ahead distance of 10,
217 supposing that TRIP_COUNT_TO_AHEAD_RATIO is equal to 4, it is
218 profitable to prefetch when the trip count is greater or equal to
219 40. In that case, 30 out of the 40 iterations will benefit from
222 #ifndef TRIP_COUNT_TO_AHEAD_RATIO
223 #define TRIP_COUNT_TO_AHEAD_RATIO 4
226 /* The group of references between that reuse may occur. */
230 tree base
; /* Base of the reference. */
231 tree step
; /* Step of the reference. */
232 struct mem_ref
*refs
; /* References in the group. */
233 struct mem_ref_group
*next
; /* Next group of references. */
236 /* Assigned to PREFETCH_BEFORE when all iterations are to be prefetched. */
238 #define PREFETCH_ALL (~(unsigned HOST_WIDE_INT) 0)
240 /* Do not generate a prefetch if the unroll factor is significantly less
241 than what is required by the prefetch. This is to avoid redundant
242 prefetches. For example, when prefetch_mod is 16 and unroll_factor is
243 2, prefetching requires unrolling the loop 16 times, but
244 the loop is actually unrolled twice. In this case (ratio = 8),
245 prefetching is not likely to be beneficial. */
247 #ifndef PREFETCH_MOD_TO_UNROLL_FACTOR_RATIO
248 #define PREFETCH_MOD_TO_UNROLL_FACTOR_RATIO 4
251 /* Some of the prefetch computations have quadratic complexity. We want to
252 avoid huge compile times and, therefore, want to limit the amount of
253 memory references per loop where we consider prefetching. */
255 #ifndef PREFETCH_MAX_MEM_REFS_PER_LOOP
256 #define PREFETCH_MAX_MEM_REFS_PER_LOOP 200
259 /* The memory reference. */
263 gimple stmt
; /* Statement in that the reference appears. */
264 tree mem
; /* The reference. */
265 HOST_WIDE_INT delta
; /* Constant offset of the reference. */
266 struct mem_ref_group
*group
; /* The group of references it belongs to. */
267 unsigned HOST_WIDE_INT prefetch_mod
;
268 /* Prefetch only each PREFETCH_MOD-th
270 unsigned HOST_WIDE_INT prefetch_before
;
271 /* Prefetch only first PREFETCH_BEFORE
273 unsigned reuse_distance
; /* The amount of data accessed before the first
274 reuse of this value. */
275 struct mem_ref
*next
; /* The next reference in the group. */
276 unsigned write_p
: 1; /* Is it a write? */
277 unsigned independent_p
: 1; /* True if the reference is independent on
278 all other references inside the loop. */
279 unsigned issue_prefetch_p
: 1; /* Should we really issue the prefetch? */
280 unsigned storent_p
: 1; /* True if we changed the store to a
284 /* Dumps information about reference REF to FILE. */
287 dump_mem_ref (FILE *file
, struct mem_ref
*ref
)
289 fprintf (file
, "Reference %p:\n", (void *) ref
);
291 fprintf (file
, " group %p (base ", (void *) ref
->group
);
292 print_generic_expr (file
, ref
->group
->base
, TDF_SLIM
);
293 fprintf (file
, ", step ");
294 if (cst_and_fits_in_hwi (ref
->group
->step
))
295 fprintf (file
, HOST_WIDE_INT_PRINT_DEC
, int_cst_value (ref
->group
->step
));
297 print_generic_expr (file
, ref
->group
->step
, TDF_TREE
);
298 fprintf (file
, ")\n");
300 fprintf (file
, " delta ");
301 fprintf (file
, HOST_WIDE_INT_PRINT_DEC
, ref
->delta
);
302 fprintf (file
, "\n");
304 fprintf (file
, " %s\n", ref
->write_p
? "write" : "read");
306 fprintf (file
, "\n");
309 /* Finds a group with BASE and STEP in GROUPS, or creates one if it does not
312 static struct mem_ref_group
*
313 find_or_create_group (struct mem_ref_group
**groups
, tree base
, tree step
)
315 struct mem_ref_group
*group
;
317 for (; *groups
; groups
= &(*groups
)->next
)
319 if (operand_equal_p ((*groups
)->step
, step
, 0)
320 && operand_equal_p ((*groups
)->base
, base
, 0))
323 /* If step is an integer constant, keep the list of groups sorted
324 by decreasing step. */
325 if (cst_and_fits_in_hwi ((*groups
)->step
) && cst_and_fits_in_hwi (step
)
326 && int_cst_value ((*groups
)->step
) < int_cst_value (step
))
330 group
= XNEW (struct mem_ref_group
);
334 group
->next
= *groups
;
340 /* Records a memory reference MEM in GROUP with offset DELTA and write status
341 WRITE_P. The reference occurs in statement STMT. */
344 record_ref (struct mem_ref_group
*group
, gimple stmt
, tree mem
,
345 HOST_WIDE_INT delta
, bool write_p
)
347 struct mem_ref
**aref
;
349 /* Do not record the same address twice. */
350 for (aref
= &group
->refs
; *aref
; aref
= &(*aref
)->next
)
352 /* It does not have to be possible for write reference to reuse the read
353 prefetch, or vice versa. */
354 if (!WRITE_CAN_USE_READ_PREFETCH
356 && !(*aref
)->write_p
)
358 if (!READ_CAN_USE_WRITE_PREFETCH
363 if ((*aref
)->delta
== delta
)
367 (*aref
) = XNEW (struct mem_ref
);
368 (*aref
)->stmt
= stmt
;
370 (*aref
)->delta
= delta
;
371 (*aref
)->write_p
= write_p
;
372 (*aref
)->prefetch_before
= PREFETCH_ALL
;
373 (*aref
)->prefetch_mod
= 1;
374 (*aref
)->reuse_distance
= 0;
375 (*aref
)->issue_prefetch_p
= false;
376 (*aref
)->group
= group
;
377 (*aref
)->next
= NULL
;
378 (*aref
)->independent_p
= false;
379 (*aref
)->storent_p
= false;
381 if (dump_file
&& (dump_flags
& TDF_DETAILS
))
382 dump_mem_ref (dump_file
, *aref
);
385 /* Release memory references in GROUPS. */
388 release_mem_refs (struct mem_ref_group
*groups
)
390 struct mem_ref_group
*next_g
;
391 struct mem_ref
*ref
, *next_r
;
393 for (; groups
; groups
= next_g
)
395 next_g
= groups
->next
;
396 for (ref
= groups
->refs
; ref
; ref
= next_r
)
405 /* A structure used to pass arguments to idx_analyze_ref. */
409 struct loop
*loop
; /* Loop of the reference. */
410 gimple stmt
; /* Statement of the reference. */
411 tree
*step
; /* Step of the memory reference. */
412 HOST_WIDE_INT
*delta
; /* Offset of the memory reference. */
415 /* Analyzes a single INDEX of a memory reference to obtain information
416 described at analyze_ref. Callback for for_each_index. */
419 idx_analyze_ref (tree base
, tree
*index
, void *data
)
421 struct ar_data
*ar_data
= (struct ar_data
*) data
;
422 tree ibase
, step
, stepsize
;
423 HOST_WIDE_INT idelta
= 0, imult
= 1;
426 if (!simple_iv (ar_data
->loop
, loop_containing_stmt (ar_data
->stmt
),
432 if (TREE_CODE (ibase
) == POINTER_PLUS_EXPR
433 && cst_and_fits_in_hwi (TREE_OPERAND (ibase
, 1)))
435 idelta
= int_cst_value (TREE_OPERAND (ibase
, 1));
436 ibase
= TREE_OPERAND (ibase
, 0);
438 if (cst_and_fits_in_hwi (ibase
))
440 idelta
+= int_cst_value (ibase
);
441 ibase
= build_int_cst (TREE_TYPE (ibase
), 0);
444 if (TREE_CODE (base
) == ARRAY_REF
)
446 stepsize
= array_ref_element_size (base
);
447 if (!cst_and_fits_in_hwi (stepsize
))
449 imult
= int_cst_value (stepsize
);
450 step
= fold_build2 (MULT_EXPR
, sizetype
,
451 fold_convert (sizetype
, step
),
452 fold_convert (sizetype
, stepsize
));
456 if (*ar_data
->step
== NULL_TREE
)
457 *ar_data
->step
= step
;
459 *ar_data
->step
= fold_build2 (PLUS_EXPR
, sizetype
,
460 fold_convert (sizetype
, *ar_data
->step
),
461 fold_convert (sizetype
, step
));
462 *ar_data
->delta
+= idelta
;
468 /* Tries to express REF_P in shape &BASE + STEP * iter + DELTA, where DELTA and
469 STEP are integer constants and iter is number of iterations of LOOP. The
470 reference occurs in statement STMT. Strips nonaddressable component
471 references from REF_P. */
474 analyze_ref (struct loop
*loop
, tree
*ref_p
, tree
*base
,
475 tree
*step
, HOST_WIDE_INT
*delta
,
478 struct ar_data ar_data
;
480 HOST_WIDE_INT bit_offset
;
486 /* First strip off the component references. Ignore bitfields.
487 Also strip off the real and imagine parts of a complex, so that
488 they can have the same base. */
489 if (TREE_CODE (ref
) == REALPART_EXPR
490 || TREE_CODE (ref
) == IMAGPART_EXPR
491 || (TREE_CODE (ref
) == COMPONENT_REF
492 && DECL_NONADDRESSABLE_P (TREE_OPERAND (ref
, 1))))
494 if (TREE_CODE (ref
) == IMAGPART_EXPR
)
495 *delta
+= int_size_in_bytes (TREE_TYPE (ref
));
496 ref
= TREE_OPERAND (ref
, 0);
501 for (; TREE_CODE (ref
) == COMPONENT_REF
; ref
= TREE_OPERAND (ref
, 0))
503 off
= DECL_FIELD_BIT_OFFSET (TREE_OPERAND (ref
, 1));
504 bit_offset
= TREE_INT_CST_LOW (off
);
505 gcc_assert (bit_offset
% BITS_PER_UNIT
== 0);
507 *delta
+= bit_offset
/ BITS_PER_UNIT
;
510 *base
= unshare_expr (ref
);
514 ar_data
.delta
= delta
;
515 return for_each_index (base
, idx_analyze_ref
, &ar_data
);
518 /* Record a memory reference REF to the list REFS. The reference occurs in
519 LOOP in statement STMT and it is write if WRITE_P. Returns true if the
520 reference was recorded, false otherwise. */
523 gather_memory_references_ref (struct loop
*loop
, struct mem_ref_group
**refs
,
524 tree ref
, bool write_p
, gimple stmt
)
528 struct mem_ref_group
*agrp
;
530 if (get_base_address (ref
) == NULL
)
533 if (!analyze_ref (loop
, &ref
, &base
, &step
, &delta
, stmt
))
535 /* If analyze_ref fails the default is a NULL_TREE. We can stop here. */
536 if (step
== NULL_TREE
)
539 /* Stop if the address of BASE could not be taken. */
540 if (may_be_nonaddressable_p (base
))
543 /* Limit non-constant step prefetching only to the innermost loops. */
544 if (!cst_and_fits_in_hwi (step
) && loop
->inner
!= NULL
)
547 /* Now we know that REF = &BASE + STEP * iter + DELTA, where DELTA and STEP
548 are integer constants. */
549 agrp
= find_or_create_group (refs
, base
, step
);
550 record_ref (agrp
, stmt
, ref
, delta
, write_p
);
555 /* Record the suitable memory references in LOOP. NO_OTHER_REFS is set to
556 true if there are no other memory references inside the loop. */
558 static struct mem_ref_group
*
559 gather_memory_references (struct loop
*loop
, bool *no_other_refs
, unsigned *ref_count
)
561 basic_block
*body
= get_loop_body_in_dom_order (loop
);
564 gimple_stmt_iterator bsi
;
567 struct mem_ref_group
*refs
= NULL
;
569 *no_other_refs
= true;
572 /* Scan the loop body in order, so that the former references precede the
574 for (i
= 0; i
< loop
->num_nodes
; i
++)
577 if (bb
->loop_father
!= loop
)
580 for (bsi
= gsi_start_bb (bb
); !gsi_end_p (bsi
); gsi_next (&bsi
))
582 stmt
= gsi_stmt (bsi
);
584 if (gimple_code (stmt
) != GIMPLE_ASSIGN
)
586 if (gimple_vuse (stmt
)
587 || (is_gimple_call (stmt
)
588 && !(gimple_call_flags (stmt
) & ECF_CONST
)))
589 *no_other_refs
= false;
593 lhs
= gimple_assign_lhs (stmt
);
594 rhs
= gimple_assign_rhs1 (stmt
);
596 if (REFERENCE_CLASS_P (rhs
))
598 *no_other_refs
&= gather_memory_references_ref (loop
, &refs
,
602 if (REFERENCE_CLASS_P (lhs
))
604 *no_other_refs
&= gather_memory_references_ref (loop
, &refs
,
615 /* Prune the prefetch candidate REF using the self-reuse. */
618 prune_ref_by_self_reuse (struct mem_ref
*ref
)
623 /* If the step size is non constant, we cannot calculate prefetch_mod. */
624 if (!cst_and_fits_in_hwi (ref
->group
->step
))
627 step
= int_cst_value (ref
->group
->step
);
633 /* Prefetch references to invariant address just once. */
634 ref
->prefetch_before
= 1;
641 if (step
> PREFETCH_BLOCK
)
644 if ((backward
&& HAVE_BACKWARD_PREFETCH
)
645 || (!backward
&& HAVE_FORWARD_PREFETCH
))
647 ref
->prefetch_before
= 1;
651 ref
->prefetch_mod
= PREFETCH_BLOCK
/ step
;
654 /* Divides X by BY, rounding down. */
657 ddown (HOST_WIDE_INT x
, unsigned HOST_WIDE_INT by
)
664 return (x
+ by
- 1) / by
;
667 /* Given a CACHE_LINE_SIZE and two inductive memory references
668 with a common STEP greater than CACHE_LINE_SIZE and an address
669 difference DELTA, compute the probability that they will fall
670 in different cache lines. Return true if the computed miss rate
671 is not greater than the ACCEPTABLE_MISS_RATE. DISTINCT_ITERS is the
672 number of distinct iterations after which the pattern repeats itself.
673 ALIGN_UNIT is the unit of alignment in bytes. */
676 is_miss_rate_acceptable (unsigned HOST_WIDE_INT cache_line_size
,
677 HOST_WIDE_INT step
, HOST_WIDE_INT delta
,
678 unsigned HOST_WIDE_INT distinct_iters
,
681 unsigned align
, iter
;
682 int total_positions
, miss_positions
, max_allowed_miss_positions
;
683 int address1
, address2
, cache_line1
, cache_line2
;
685 /* It always misses if delta is greater than or equal to the cache
687 if (delta
>= (HOST_WIDE_INT
) cache_line_size
)
691 total_positions
= (cache_line_size
/ align_unit
) * distinct_iters
;
692 max_allowed_miss_positions
= (ACCEPTABLE_MISS_RATE
* total_positions
) / 1000;
694 /* Iterate through all possible alignments of the first
695 memory reference within its cache line. */
696 for (align
= 0; align
< cache_line_size
; align
+= align_unit
)
698 /* Iterate through all distinct iterations. */
699 for (iter
= 0; iter
< distinct_iters
; iter
++)
701 address1
= align
+ step
* iter
;
702 address2
= address1
+ delta
;
703 cache_line1
= address1
/ cache_line_size
;
704 cache_line2
= address2
/ cache_line_size
;
705 if (cache_line1
!= cache_line2
)
708 if (miss_positions
> max_allowed_miss_positions
)
715 /* Prune the prefetch candidate REF using the reuse with BY.
716 If BY_IS_BEFORE is true, BY is before REF in the loop. */
719 prune_ref_by_group_reuse (struct mem_ref
*ref
, struct mem_ref
*by
,
724 HOST_WIDE_INT delta_r
= ref
->delta
, delta_b
= by
->delta
;
725 HOST_WIDE_INT delta
= delta_b
- delta_r
;
726 HOST_WIDE_INT hit_from
;
727 unsigned HOST_WIDE_INT prefetch_before
, prefetch_block
;
728 HOST_WIDE_INT reduced_step
;
729 unsigned HOST_WIDE_INT reduced_prefetch_block
;
733 /* If the step is non constant we cannot calculate prefetch_before. */
734 if (!cst_and_fits_in_hwi (ref
->group
->step
)) {
738 step
= int_cst_value (ref
->group
->step
);
745 /* If the references has the same address, only prefetch the
748 ref
->prefetch_before
= 0;
755 /* If the reference addresses are invariant and fall into the
756 same cache line, prefetch just the first one. */
760 if (ddown (ref
->delta
, PREFETCH_BLOCK
)
761 != ddown (by
->delta
, PREFETCH_BLOCK
))
764 ref
->prefetch_before
= 0;
768 /* Only prune the reference that is behind in the array. */
774 /* Transform the data so that we may assume that the accesses
778 delta_r
= PREFETCH_BLOCK
- 1 - delta_r
;
779 delta_b
= PREFETCH_BLOCK
- 1 - delta_b
;
787 /* Check whether the two references are likely to hit the same cache
788 line, and how distant the iterations in that it occurs are from
791 if (step
<= PREFETCH_BLOCK
)
793 /* The accesses are sure to meet. Let us check when. */
794 hit_from
= ddown (delta_b
, PREFETCH_BLOCK
) * PREFETCH_BLOCK
;
795 prefetch_before
= (hit_from
- delta_r
+ step
- 1) / step
;
797 /* Do not reduce prefetch_before if we meet beyond cache size. */
798 if (prefetch_before
> absu_hwi (L2_CACHE_SIZE_BYTES
/ step
))
799 prefetch_before
= PREFETCH_ALL
;
800 if (prefetch_before
< ref
->prefetch_before
)
801 ref
->prefetch_before
= prefetch_before
;
806 /* A more complicated case with step > prefetch_block. First reduce
807 the ratio between the step and the cache line size to its simplest
808 terms. The resulting denominator will then represent the number of
809 distinct iterations after which each address will go back to its
810 initial location within the cache line. This computation assumes
811 that PREFETCH_BLOCK is a power of two. */
812 prefetch_block
= PREFETCH_BLOCK
;
813 reduced_prefetch_block
= prefetch_block
;
815 while ((reduced_step
& 1) == 0
816 && reduced_prefetch_block
> 1)
819 reduced_prefetch_block
>>= 1;
822 prefetch_before
= delta
/ step
;
824 ref_type
= TREE_TYPE (ref
->mem
);
825 align_unit
= TYPE_ALIGN (ref_type
) / 8;
826 if (is_miss_rate_acceptable (prefetch_block
, step
, delta
,
827 reduced_prefetch_block
, align_unit
))
829 /* Do not reduce prefetch_before if we meet beyond cache size. */
830 if (prefetch_before
> L2_CACHE_SIZE_BYTES
/ PREFETCH_BLOCK
)
831 prefetch_before
= PREFETCH_ALL
;
832 if (prefetch_before
< ref
->prefetch_before
)
833 ref
->prefetch_before
= prefetch_before
;
838 /* Try also the following iteration. */
840 delta
= step
- delta
;
841 if (is_miss_rate_acceptable (prefetch_block
, step
, delta
,
842 reduced_prefetch_block
, align_unit
))
844 if (prefetch_before
< ref
->prefetch_before
)
845 ref
->prefetch_before
= prefetch_before
;
850 /* The ref probably does not reuse by. */
854 /* Prune the prefetch candidate REF using the reuses with other references
858 prune_ref_by_reuse (struct mem_ref
*ref
, struct mem_ref
*refs
)
860 struct mem_ref
*prune_by
;
863 prune_ref_by_self_reuse (ref
);
865 for (prune_by
= refs
; prune_by
; prune_by
= prune_by
->next
)
873 if (!WRITE_CAN_USE_READ_PREFETCH
875 && !prune_by
->write_p
)
877 if (!READ_CAN_USE_WRITE_PREFETCH
879 && prune_by
->write_p
)
882 prune_ref_by_group_reuse (ref
, prune_by
, before
);
886 /* Prune the prefetch candidates in GROUP using the reuse analysis. */
889 prune_group_by_reuse (struct mem_ref_group
*group
)
891 struct mem_ref
*ref_pruned
;
893 for (ref_pruned
= group
->refs
; ref_pruned
; ref_pruned
= ref_pruned
->next
)
895 prune_ref_by_reuse (ref_pruned
, group
->refs
);
897 if (dump_file
&& (dump_flags
& TDF_DETAILS
))
899 fprintf (dump_file
, "Reference %p:", (void *) ref_pruned
);
901 if (ref_pruned
->prefetch_before
== PREFETCH_ALL
902 && ref_pruned
->prefetch_mod
== 1)
903 fprintf (dump_file
, " no restrictions");
904 else if (ref_pruned
->prefetch_before
== 0)
905 fprintf (dump_file
, " do not prefetch");
906 else if (ref_pruned
->prefetch_before
<= ref_pruned
->prefetch_mod
)
907 fprintf (dump_file
, " prefetch once");
910 if (ref_pruned
->prefetch_before
!= PREFETCH_ALL
)
912 fprintf (dump_file
, " prefetch before ");
913 fprintf (dump_file
, HOST_WIDE_INT_PRINT_DEC
,
914 ref_pruned
->prefetch_before
);
916 if (ref_pruned
->prefetch_mod
!= 1)
918 fprintf (dump_file
, " prefetch mod ");
919 fprintf (dump_file
, HOST_WIDE_INT_PRINT_DEC
,
920 ref_pruned
->prefetch_mod
);
923 fprintf (dump_file
, "\n");
928 /* Prune the list of prefetch candidates GROUPS using the reuse analysis. */
931 prune_by_reuse (struct mem_ref_group
*groups
)
933 for (; groups
; groups
= groups
->next
)
934 prune_group_by_reuse (groups
);
937 /* Returns true if we should issue prefetch for REF. */
940 should_issue_prefetch_p (struct mem_ref
*ref
)
942 /* For now do not issue prefetches for only first few of the
944 if (ref
->prefetch_before
!= PREFETCH_ALL
)
946 if (dump_file
&& (dump_flags
& TDF_DETAILS
))
947 fprintf (dump_file
, "Ignoring %p due to prefetch_before\n",
952 /* Do not prefetch nontemporal stores. */
955 if (dump_file
&& (dump_flags
& TDF_DETAILS
))
956 fprintf (dump_file
, "Ignoring nontemporal store %p\n", (void *) ref
);
963 /* Decide which of the prefetch candidates in GROUPS to prefetch.
964 AHEAD is the number of iterations to prefetch ahead (which corresponds
965 to the number of simultaneous instances of one prefetch running at a
966 time). UNROLL_FACTOR is the factor by that the loop is going to be
967 unrolled. Returns true if there is anything to prefetch. */
970 schedule_prefetches (struct mem_ref_group
*groups
, unsigned unroll_factor
,
973 unsigned remaining_prefetch_slots
, n_prefetches
, prefetch_slots
;
974 unsigned slots_per_prefetch
;
978 /* At most SIMULTANEOUS_PREFETCHES should be running at the same time. */
979 remaining_prefetch_slots
= SIMULTANEOUS_PREFETCHES
;
981 /* The prefetch will run for AHEAD iterations of the original loop, i.e.,
982 AHEAD / UNROLL_FACTOR iterations of the unrolled loop. In each iteration,
983 it will need a prefetch slot. */
984 slots_per_prefetch
= (ahead
+ unroll_factor
/ 2) / unroll_factor
;
985 if (dump_file
&& (dump_flags
& TDF_DETAILS
))
986 fprintf (dump_file
, "Each prefetch instruction takes %u prefetch slots.\n",
989 /* For now we just take memory references one by one and issue
990 prefetches for as many as possible. The groups are sorted
991 starting with the largest step, since the references with
992 large step are more likely to cause many cache misses. */
994 for (; groups
; groups
= groups
->next
)
995 for (ref
= groups
->refs
; ref
; ref
= ref
->next
)
997 if (!should_issue_prefetch_p (ref
))
1000 /* The loop is far from being sufficiently unrolled for this
1001 prefetch. Do not generate prefetch to avoid many redudant
1003 if (ref
->prefetch_mod
/ unroll_factor
> PREFETCH_MOD_TO_UNROLL_FACTOR_RATIO
)
1006 /* If we need to prefetch the reference each PREFETCH_MOD iterations,
1007 and we unroll the loop UNROLL_FACTOR times, we need to insert
1008 ceil (UNROLL_FACTOR / PREFETCH_MOD) instructions in each
1010 n_prefetches
= ((unroll_factor
+ ref
->prefetch_mod
- 1)
1011 / ref
->prefetch_mod
);
1012 prefetch_slots
= n_prefetches
* slots_per_prefetch
;
1014 /* If more than half of the prefetches would be lost anyway, do not
1015 issue the prefetch. */
1016 if (2 * remaining_prefetch_slots
< prefetch_slots
)
1019 ref
->issue_prefetch_p
= true;
1021 if (remaining_prefetch_slots
<= prefetch_slots
)
1023 remaining_prefetch_slots
-= prefetch_slots
;
1030 /* Return TRUE if no prefetch is going to be generated in the given
1034 nothing_to_prefetch_p (struct mem_ref_group
*groups
)
1036 struct mem_ref
*ref
;
1038 for (; groups
; groups
= groups
->next
)
1039 for (ref
= groups
->refs
; ref
; ref
= ref
->next
)
1040 if (should_issue_prefetch_p (ref
))
1046 /* Estimate the number of prefetches in the given GROUPS.
1047 UNROLL_FACTOR is the factor by which LOOP was unrolled. */
1050 estimate_prefetch_count (struct mem_ref_group
*groups
, unsigned unroll_factor
)
1052 struct mem_ref
*ref
;
1053 unsigned n_prefetches
;
1054 int prefetch_count
= 0;
1056 for (; groups
; groups
= groups
->next
)
1057 for (ref
= groups
->refs
; ref
; ref
= ref
->next
)
1058 if (should_issue_prefetch_p (ref
))
1060 n_prefetches
= ((unroll_factor
+ ref
->prefetch_mod
- 1)
1061 / ref
->prefetch_mod
);
1062 prefetch_count
+= n_prefetches
;
1065 return prefetch_count
;
1068 /* Issue prefetches for the reference REF into loop as decided before.
1069 HEAD is the number of iterations to prefetch ahead. UNROLL_FACTOR
1070 is the factor by which LOOP was unrolled. */
1073 issue_prefetch_ref (struct mem_ref
*ref
, unsigned unroll_factor
, unsigned ahead
)
1075 HOST_WIDE_INT delta
;
1076 tree addr
, addr_base
, write_p
, local
, forward
;
1078 gimple_stmt_iterator bsi
;
1079 unsigned n_prefetches
, ap
;
1080 bool nontemporal
= ref
->reuse_distance
>= L2_CACHE_SIZE_BYTES
;
1082 if (dump_file
&& (dump_flags
& TDF_DETAILS
))
1083 fprintf (dump_file
, "Issued%s prefetch for %p.\n",
1084 nontemporal
? " nontemporal" : "",
1087 bsi
= gsi_for_stmt (ref
->stmt
);
1089 n_prefetches
= ((unroll_factor
+ ref
->prefetch_mod
- 1)
1090 / ref
->prefetch_mod
);
1091 addr_base
= build_fold_addr_expr_with_type (ref
->mem
, ptr_type_node
);
1092 addr_base
= force_gimple_operand_gsi (&bsi
, unshare_expr (addr_base
),
1093 true, NULL
, true, GSI_SAME_STMT
);
1094 write_p
= ref
->write_p
? integer_one_node
: integer_zero_node
;
1095 local
= nontemporal
? integer_zero_node
: integer_three_node
;
1097 for (ap
= 0; ap
< n_prefetches
; ap
++)
1099 if (cst_and_fits_in_hwi (ref
->group
->step
))
1101 /* Determine the address to prefetch. */
1102 delta
= (ahead
+ ap
* ref
->prefetch_mod
) *
1103 int_cst_value (ref
->group
->step
);
1104 addr
= fold_build_pointer_plus_hwi (addr_base
, delta
);
1105 addr
= force_gimple_operand_gsi (&bsi
, unshare_expr (addr
), true, NULL
,
1106 true, GSI_SAME_STMT
);
1110 /* The step size is non-constant but loop-invariant. We use the
1111 heuristic to simply prefetch ahead iterations ahead. */
1112 forward
= fold_build2 (MULT_EXPR
, sizetype
,
1113 fold_convert (sizetype
, ref
->group
->step
),
1114 fold_convert (sizetype
, size_int (ahead
)));
1115 addr
= fold_build_pointer_plus (addr_base
, forward
);
1116 addr
= force_gimple_operand_gsi (&bsi
, unshare_expr (addr
), true,
1117 NULL
, true, GSI_SAME_STMT
);
1119 /* Create the prefetch instruction. */
1120 prefetch
= gimple_build_call (builtin_decl_explicit (BUILT_IN_PREFETCH
),
1121 3, addr
, write_p
, local
);
1122 gsi_insert_before (&bsi
, prefetch
, GSI_SAME_STMT
);
1126 /* Issue prefetches for the references in GROUPS into loop as decided before.
1127 HEAD is the number of iterations to prefetch ahead. UNROLL_FACTOR is the
1128 factor by that LOOP was unrolled. */
1131 issue_prefetches (struct mem_ref_group
*groups
,
1132 unsigned unroll_factor
, unsigned ahead
)
1134 struct mem_ref
*ref
;
1136 for (; groups
; groups
= groups
->next
)
1137 for (ref
= groups
->refs
; ref
; ref
= ref
->next
)
1138 if (ref
->issue_prefetch_p
)
1139 issue_prefetch_ref (ref
, unroll_factor
, ahead
);
1142 /* Returns true if REF is a memory write for that a nontemporal store insn
1146 nontemporal_store_p (struct mem_ref
*ref
)
1148 enum machine_mode mode
;
1149 enum insn_code code
;
1151 /* REF must be a write that is not reused. We require it to be independent
1152 on all other memory references in the loop, as the nontemporal stores may
1153 be reordered with respect to other memory references. */
1155 || !ref
->independent_p
1156 || ref
->reuse_distance
< L2_CACHE_SIZE_BYTES
)
1159 /* Check that we have the storent instruction for the mode. */
1160 mode
= TYPE_MODE (TREE_TYPE (ref
->mem
));
1161 if (mode
== BLKmode
)
1164 code
= optab_handler (storent_optab
, mode
);
1165 return code
!= CODE_FOR_nothing
;
1168 /* If REF is a nontemporal store, we mark the corresponding modify statement
1169 and return true. Otherwise, we return false. */
1172 mark_nontemporal_store (struct mem_ref
*ref
)
1174 if (!nontemporal_store_p (ref
))
1177 if (dump_file
&& (dump_flags
& TDF_DETAILS
))
1178 fprintf (dump_file
, "Marked reference %p as a nontemporal store.\n",
1181 gimple_assign_set_nontemporal_move (ref
->stmt
, true);
1182 ref
->storent_p
= true;
1187 /* Issue a memory fence instruction after LOOP. */
1190 emit_mfence_after_loop (struct loop
*loop
)
1192 VEC (edge
, heap
) *exits
= get_loop_exit_edges (loop
);
1195 gimple_stmt_iterator bsi
;
1198 FOR_EACH_VEC_ELT (edge
, exits
, i
, exit
)
1200 call
= gimple_build_call (FENCE_FOLLOWING_MOVNT
, 0);
1202 if (!single_pred_p (exit
->dest
)
1203 /* If possible, we prefer not to insert the fence on other paths
1205 && !(exit
->flags
& EDGE_ABNORMAL
))
1206 split_loop_exit_edge (exit
);
1207 bsi
= gsi_after_labels (exit
->dest
);
1209 gsi_insert_before (&bsi
, call
, GSI_NEW_STMT
);
1210 mark_virtual_ops_for_renaming (call
);
1213 VEC_free (edge
, heap
, exits
);
1214 update_ssa (TODO_update_ssa_only_virtuals
);
1217 /* Returns true if we can use storent in loop, false otherwise. */
1220 may_use_storent_in_loop_p (struct loop
*loop
)
1224 if (loop
->inner
!= NULL
)
1227 /* If we must issue a mfence insn after using storent, check that there
1228 is a suitable place for it at each of the loop exits. */
1229 if (FENCE_FOLLOWING_MOVNT
!= NULL_TREE
)
1231 VEC (edge
, heap
) *exits
= get_loop_exit_edges (loop
);
1235 FOR_EACH_VEC_ELT (edge
, exits
, i
, exit
)
1236 if ((exit
->flags
& EDGE_ABNORMAL
)
1237 && exit
->dest
== EXIT_BLOCK_PTR
)
1240 VEC_free (edge
, heap
, exits
);
1246 /* Marks nontemporal stores in LOOP. GROUPS contains the description of memory
1247 references in the loop. */
1250 mark_nontemporal_stores (struct loop
*loop
, struct mem_ref_group
*groups
)
1252 struct mem_ref
*ref
;
1255 if (!may_use_storent_in_loop_p (loop
))
1258 for (; groups
; groups
= groups
->next
)
1259 for (ref
= groups
->refs
; ref
; ref
= ref
->next
)
1260 any
|= mark_nontemporal_store (ref
);
1262 if (any
&& FENCE_FOLLOWING_MOVNT
!= NULL_TREE
)
1263 emit_mfence_after_loop (loop
);
1266 /* Determines whether we can profitably unroll LOOP FACTOR times, and if
1267 this is the case, fill in DESC by the description of number of
1271 should_unroll_loop_p (struct loop
*loop
, struct tree_niter_desc
*desc
,
1274 if (!can_unroll_loop_p (loop
, factor
, desc
))
1277 /* We only consider loops without control flow for unrolling. This is not
1278 a hard restriction -- tree_unroll_loop works with arbitrary loops
1279 as well; but the unrolling/prefetching is usually more profitable for
1280 loops consisting of a single basic block, and we want to limit the
1282 if (loop
->num_nodes
> 2)
1288 /* Determine the coefficient by that unroll LOOP, from the information
1289 contained in the list of memory references REFS. Description of
1290 umber of iterations of LOOP is stored to DESC. NINSNS is the number of
1291 insns of the LOOP. EST_NITER is the estimated number of iterations of
1292 the loop, or -1 if no estimate is available. */
1295 determine_unroll_factor (struct loop
*loop
, struct mem_ref_group
*refs
,
1296 unsigned ninsns
, struct tree_niter_desc
*desc
,
1297 HOST_WIDE_INT est_niter
)
1299 unsigned upper_bound
;
1300 unsigned nfactor
, factor
, mod_constraint
;
1301 struct mem_ref_group
*agp
;
1302 struct mem_ref
*ref
;
1304 /* First check whether the loop is not too large to unroll. We ignore
1305 PARAM_MAX_UNROLL_TIMES, because for small loops, it prevented us
1306 from unrolling them enough to make exactly one cache line covered by each
1307 iteration. Also, the goal of PARAM_MAX_UNROLL_TIMES is to prevent
1308 us from unrolling the loops too many times in cases where we only expect
1309 gains from better scheduling and decreasing loop overhead, which is not
1311 upper_bound
= PARAM_VALUE (PARAM_MAX_UNROLLED_INSNS
) / ninsns
;
1313 /* If we unrolled the loop more times than it iterates, the unrolled version
1314 of the loop would be never entered. */
1315 if (est_niter
>= 0 && est_niter
< (HOST_WIDE_INT
) upper_bound
)
1316 upper_bound
= est_niter
;
1318 if (upper_bound
<= 1)
1321 /* Choose the factor so that we may prefetch each cache just once,
1322 but bound the unrolling by UPPER_BOUND. */
1324 for (agp
= refs
; agp
; agp
= agp
->next
)
1325 for (ref
= agp
->refs
; ref
; ref
= ref
->next
)
1326 if (should_issue_prefetch_p (ref
))
1328 mod_constraint
= ref
->prefetch_mod
;
1329 nfactor
= least_common_multiple (mod_constraint
, factor
);
1330 if (nfactor
<= upper_bound
)
1334 if (!should_unroll_loop_p (loop
, desc
, factor
))
1340 /* Returns the total volume of the memory references REFS, taking into account
1341 reuses in the innermost loop and cache line size. TODO -- we should also
1342 take into account reuses across the iterations of the loops in the loop
1346 volume_of_references (struct mem_ref_group
*refs
)
1348 unsigned volume
= 0;
1349 struct mem_ref_group
*gr
;
1350 struct mem_ref
*ref
;
1352 for (gr
= refs
; gr
; gr
= gr
->next
)
1353 for (ref
= gr
->refs
; ref
; ref
= ref
->next
)
1355 /* Almost always reuses another value? */
1356 if (ref
->prefetch_before
!= PREFETCH_ALL
)
1359 /* If several iterations access the same cache line, use the size of
1360 the line divided by this number. Otherwise, a cache line is
1361 accessed in each iteration. TODO -- in the latter case, we should
1362 take the size of the reference into account, rounding it up on cache
1363 line size multiple. */
1364 volume
+= L1_CACHE_LINE_SIZE
/ ref
->prefetch_mod
;
1369 /* Returns the volume of memory references accessed across VEC iterations of
1370 loops, whose sizes are described in the LOOP_SIZES array. N is the number
1371 of the loops in the nest (length of VEC and LOOP_SIZES vectors). */
1374 volume_of_dist_vector (lambda_vector vec
, unsigned *loop_sizes
, unsigned n
)
1378 for (i
= 0; i
< n
; i
++)
1385 gcc_assert (vec
[i
] > 0);
1387 /* We ignore the parts of the distance vector in subloops, since usually
1388 the numbers of iterations are much smaller. */
1389 return loop_sizes
[i
] * vec
[i
];
1392 /* Add the steps of ACCESS_FN multiplied by STRIDE to the array STRIDE
1393 at the position corresponding to the loop of the step. N is the depth
1394 of the considered loop nest, and, LOOP is its innermost loop. */
1397 add_subscript_strides (tree access_fn
, unsigned stride
,
1398 HOST_WIDE_INT
*strides
, unsigned n
, struct loop
*loop
)
1402 HOST_WIDE_INT astep
;
1403 unsigned min_depth
= loop_depth (loop
) - n
;
1405 while (TREE_CODE (access_fn
) == POLYNOMIAL_CHREC
)
1407 aloop
= get_chrec_loop (access_fn
);
1408 step
= CHREC_RIGHT (access_fn
);
1409 access_fn
= CHREC_LEFT (access_fn
);
1411 if ((unsigned) loop_depth (aloop
) <= min_depth
)
1414 if (host_integerp (step
, 0))
1415 astep
= tree_low_cst (step
, 0);
1417 astep
= L1_CACHE_LINE_SIZE
;
1419 strides
[n
- 1 - loop_depth (loop
) + loop_depth (aloop
)] += astep
* stride
;
1424 /* Returns the volume of memory references accessed between two consecutive
1425 self-reuses of the reference DR. We consider the subscripts of DR in N
1426 loops, and LOOP_SIZES contains the volumes of accesses in each of the
1427 loops. LOOP is the innermost loop of the current loop nest. */
1430 self_reuse_distance (data_reference_p dr
, unsigned *loop_sizes
, unsigned n
,
1433 tree stride
, access_fn
;
1434 HOST_WIDE_INT
*strides
, astride
;
1435 VEC (tree
, heap
) *access_fns
;
1436 tree ref
= DR_REF (dr
);
1437 unsigned i
, ret
= ~0u;
1439 /* In the following example:
1441 for (i = 0; i < N; i++)
1442 for (j = 0; j < N; j++)
1444 the same cache line is accessed each N steps (except if the change from
1445 i to i + 1 crosses the boundary of the cache line). Thus, for self-reuse,
1446 we cannot rely purely on the results of the data dependence analysis.
1448 Instead, we compute the stride of the reference in each loop, and consider
1449 the innermost loop in that the stride is less than cache size. */
1451 strides
= XCNEWVEC (HOST_WIDE_INT
, n
);
1452 access_fns
= DR_ACCESS_FNS (dr
);
1454 FOR_EACH_VEC_ELT (tree
, access_fns
, i
, access_fn
)
1456 /* Keep track of the reference corresponding to the subscript, so that we
1458 while (handled_component_p (ref
) && TREE_CODE (ref
) != ARRAY_REF
)
1459 ref
= TREE_OPERAND (ref
, 0);
1461 if (TREE_CODE (ref
) == ARRAY_REF
)
1463 stride
= TYPE_SIZE_UNIT (TREE_TYPE (ref
));
1464 if (host_integerp (stride
, 1))
1465 astride
= tree_low_cst (stride
, 1);
1467 astride
= L1_CACHE_LINE_SIZE
;
1469 ref
= TREE_OPERAND (ref
, 0);
1474 add_subscript_strides (access_fn
, astride
, strides
, n
, loop
);
1477 for (i
= n
; i
-- > 0; )
1479 unsigned HOST_WIDE_INT s
;
1481 s
= strides
[i
] < 0 ? -strides
[i
] : strides
[i
];
1483 if (s
< (unsigned) L1_CACHE_LINE_SIZE
1485 > (unsigned) (L1_CACHE_SIZE_BYTES
/ NONTEMPORAL_FRACTION
)))
1487 ret
= loop_sizes
[i
];
1496 /* Determines the distance till the first reuse of each reference in REFS
1497 in the loop nest of LOOP. NO_OTHER_REFS is true if there are no other
1498 memory references in the loop. */
1501 determine_loop_nest_reuse (struct loop
*loop
, struct mem_ref_group
*refs
,
1504 struct loop
*nest
, *aloop
;
1505 VEC (data_reference_p
, heap
) *datarefs
= NULL
;
1506 VEC (ddr_p
, heap
) *dependences
= NULL
;
1507 struct mem_ref_group
*gr
;
1508 struct mem_ref
*ref
, *refb
;
1509 VEC (loop_p
, heap
) *vloops
= NULL
;
1510 unsigned *loop_data_size
;
1512 unsigned volume
, dist
, adist
;
1514 data_reference_p dr
;
1520 /* Find the outermost loop of the loop nest of loop (we require that
1521 there are no sibling loops inside the nest). */
1525 aloop
= loop_outer (nest
);
1527 if (aloop
== current_loops
->tree_root
1528 || aloop
->inner
->next
)
1534 /* For each loop, determine the amount of data accessed in each iteration.
1535 We use this to estimate whether the reference is evicted from the
1536 cache before its reuse. */
1537 find_loop_nest (nest
, &vloops
);
1538 n
= VEC_length (loop_p
, vloops
);
1539 loop_data_size
= XNEWVEC (unsigned, n
);
1540 volume
= volume_of_references (refs
);
1544 loop_data_size
[i
] = volume
;
1545 /* Bound the volume by the L2 cache size, since above this bound,
1546 all dependence distances are equivalent. */
1547 if (volume
> L2_CACHE_SIZE_BYTES
)
1550 aloop
= VEC_index (loop_p
, vloops
, i
);
1551 vol
= estimated_stmt_executions_int (aloop
);
1553 vol
= expected_loop_iterations (aloop
);
1557 /* Prepare the references in the form suitable for data dependence
1558 analysis. We ignore unanalyzable data references (the results
1559 are used just as a heuristics to estimate temporality of the
1560 references, hence we do not need to worry about correctness). */
1561 for (gr
= refs
; gr
; gr
= gr
->next
)
1562 for (ref
= gr
->refs
; ref
; ref
= ref
->next
)
1564 dr
= create_data_ref (nest
, loop_containing_stmt (ref
->stmt
),
1565 ref
->mem
, ref
->stmt
, !ref
->write_p
);
1569 ref
->reuse_distance
= volume
;
1571 VEC_safe_push (data_reference_p
, heap
, datarefs
, dr
);
1574 no_other_refs
= false;
1577 FOR_EACH_VEC_ELT (data_reference_p
, datarefs
, i
, dr
)
1579 dist
= self_reuse_distance (dr
, loop_data_size
, n
, loop
);
1580 ref
= (struct mem_ref
*) dr
->aux
;
1581 if (ref
->reuse_distance
> dist
)
1582 ref
->reuse_distance
= dist
;
1585 ref
->independent_p
= true;
1588 compute_all_dependences (datarefs
, &dependences
, vloops
, true);
1590 FOR_EACH_VEC_ELT (ddr_p
, dependences
, i
, dep
)
1592 if (DDR_ARE_DEPENDENT (dep
) == chrec_known
)
1595 ref
= (struct mem_ref
*) DDR_A (dep
)->aux
;
1596 refb
= (struct mem_ref
*) DDR_B (dep
)->aux
;
1598 if (DDR_ARE_DEPENDENT (dep
) == chrec_dont_know
1599 || DDR_NUM_DIST_VECTS (dep
) == 0)
1601 /* If the dependence cannot be analyzed, assume that there might be
1605 ref
->independent_p
= false;
1606 refb
->independent_p
= false;
1610 /* The distance vectors are normalized to be always lexicographically
1611 positive, hence we cannot tell just from them whether DDR_A comes
1612 before DDR_B or vice versa. However, it is not important,
1613 anyway -- if DDR_A is close to DDR_B, then it is either reused in
1614 DDR_B (and it is not nontemporal), or it reuses the value of DDR_B
1615 in cache (and marking it as nontemporal would not affect
1619 for (j
= 0; j
< DDR_NUM_DIST_VECTS (dep
); j
++)
1621 adist
= volume_of_dist_vector (DDR_DIST_VECT (dep
, j
),
1624 /* If this is a dependence in the innermost loop (i.e., the
1625 distances in all superloops are zero) and it is not
1626 the trivial self-dependence with distance zero, record that
1627 the references are not completely independent. */
1628 if (lambda_vector_zerop (DDR_DIST_VECT (dep
, j
), n
- 1)
1630 || DDR_DIST_VECT (dep
, j
)[n
-1] != 0))
1632 ref
->independent_p
= false;
1633 refb
->independent_p
= false;
1636 /* Ignore accesses closer than
1637 L1_CACHE_SIZE_BYTES / NONTEMPORAL_FRACTION,
1638 so that we use nontemporal prefetches e.g. if single memory
1639 location is accessed several times in a single iteration of
1641 if (adist
< L1_CACHE_SIZE_BYTES
/ NONTEMPORAL_FRACTION
)
1649 if (ref
->reuse_distance
> dist
)
1650 ref
->reuse_distance
= dist
;
1651 if (refb
->reuse_distance
> dist
)
1652 refb
->reuse_distance
= dist
;
1655 free_dependence_relations (dependences
);
1656 free_data_refs (datarefs
);
1657 free (loop_data_size
);
1659 if (dump_file
&& (dump_flags
& TDF_DETAILS
))
1661 fprintf (dump_file
, "Reuse distances:\n");
1662 for (gr
= refs
; gr
; gr
= gr
->next
)
1663 for (ref
= gr
->refs
; ref
; ref
= ref
->next
)
1664 fprintf (dump_file
, " ref %p distance %u\n",
1665 (void *) ref
, ref
->reuse_distance
);
1669 /* Determine whether or not the trip count to ahead ratio is too small based
1670 on prefitablility consideration.
1671 AHEAD: the iteration ahead distance,
1672 EST_NITER: the estimated trip count. */
1675 trip_count_to_ahead_ratio_too_small_p (unsigned ahead
, HOST_WIDE_INT est_niter
)
1677 /* Assume trip count to ahead ratio is big enough if the trip count could not
1678 be estimated at compile time. */
1682 if (est_niter
< (HOST_WIDE_INT
) (TRIP_COUNT_TO_AHEAD_RATIO
* ahead
))
1684 if (dump_file
&& (dump_flags
& TDF_DETAILS
))
1686 "Not prefetching -- loop estimated to roll only %d times\n",
1694 /* Determine whether or not the number of memory references in the loop is
1695 reasonable based on the profitablity and compilation time considerations.
1696 NINSNS: estimated number of instructions in the loop,
1697 MEM_REF_COUNT: total number of memory references in the loop. */
1700 mem_ref_count_reasonable_p (unsigned ninsns
, unsigned mem_ref_count
)
1702 int insn_to_mem_ratio
;
1704 if (mem_ref_count
== 0)
1707 /* Miss rate computation (is_miss_rate_acceptable) and dependence analysis
1708 (compute_all_dependences) have high costs based on quadratic complexity.
1709 To avoid huge compilation time, we give up prefetching if mem_ref_count
1711 if (mem_ref_count
> PREFETCH_MAX_MEM_REFS_PER_LOOP
)
1714 /* Prefetching improves performance by overlapping cache missing
1715 memory accesses with CPU operations. If the loop does not have
1716 enough CPU operations to overlap with memory operations, prefetching
1717 won't give a significant benefit. One approximate way of checking
1718 this is to require the ratio of instructions to memory references to
1719 be above a certain limit. This approximation works well in practice.
1720 TODO: Implement a more precise computation by estimating the time
1721 for each CPU or memory op in the loop. Time estimates for memory ops
1722 should account for cache misses. */
1723 insn_to_mem_ratio
= ninsns
/ mem_ref_count
;
1725 if (insn_to_mem_ratio
< PREFETCH_MIN_INSN_TO_MEM_RATIO
)
1727 if (dump_file
&& (dump_flags
& TDF_DETAILS
))
1729 "Not prefetching -- instruction to memory reference ratio (%d) too small\n",
1737 /* Determine whether or not the instruction to prefetch ratio in the loop is
1738 too small based on the profitablity consideration.
1739 NINSNS: estimated number of instructions in the loop,
1740 PREFETCH_COUNT: an estimate of the number of prefetches,
1741 UNROLL_FACTOR: the factor to unroll the loop if prefetching. */
1744 insn_to_prefetch_ratio_too_small_p (unsigned ninsns
, unsigned prefetch_count
,
1745 unsigned unroll_factor
)
1747 int insn_to_prefetch_ratio
;
1749 /* Prefetching most likely causes performance degradation when the instruction
1750 to prefetch ratio is too small. Too many prefetch instructions in a loop
1751 may reduce the I-cache performance.
1752 (unroll_factor * ninsns) is used to estimate the number of instructions in
1753 the unrolled loop. This implementation is a bit simplistic -- the number
1754 of issued prefetch instructions is also affected by unrolling. So,
1755 prefetch_mod and the unroll factor should be taken into account when
1756 determining prefetch_count. Also, the number of insns of the unrolled
1757 loop will usually be significantly smaller than the number of insns of the
1758 original loop * unroll_factor (at least the induction variable increases
1759 and the exit branches will get eliminated), so it might be better to use
1760 tree_estimate_loop_size + estimated_unrolled_size. */
1761 insn_to_prefetch_ratio
= (unroll_factor
* ninsns
) / prefetch_count
;
1762 if (insn_to_prefetch_ratio
< MIN_INSN_TO_PREFETCH_RATIO
)
1764 if (dump_file
&& (dump_flags
& TDF_DETAILS
))
1766 "Not prefetching -- instruction to prefetch ratio (%d) too small\n",
1767 insn_to_prefetch_ratio
);
1775 /* Issue prefetch instructions for array references in LOOP. Returns
1776 true if the LOOP was unrolled. */
1779 loop_prefetch_arrays (struct loop
*loop
)
1781 struct mem_ref_group
*refs
;
1782 unsigned ahead
, ninsns
, time
, unroll_factor
;
1783 HOST_WIDE_INT est_niter
;
1784 struct tree_niter_desc desc
;
1785 bool unrolled
= false, no_other_refs
;
1786 unsigned prefetch_count
;
1787 unsigned mem_ref_count
;
1789 if (optimize_loop_nest_for_size_p (loop
))
1791 if (dump_file
&& (dump_flags
& TDF_DETAILS
))
1792 fprintf (dump_file
, " ignored (cold area)\n");
1796 /* FIXME: the time should be weighted by the probabilities of the blocks in
1798 time
= tree_num_loop_insns (loop
, &eni_time_weights
);
1802 ahead
= (PREFETCH_LATENCY
+ time
- 1) / time
;
1803 est_niter
= estimated_stmt_executions_int (loop
);
1805 /* Prefetching is not likely to be profitable if the trip count to ahead
1806 ratio is too small. */
1807 if (trip_count_to_ahead_ratio_too_small_p (ahead
, est_niter
))
1810 ninsns
= tree_num_loop_insns (loop
, &eni_size_weights
);
1812 /* Step 1: gather the memory references. */
1813 refs
= gather_memory_references (loop
, &no_other_refs
, &mem_ref_count
);
1815 /* Give up prefetching if the number of memory references in the
1816 loop is not reasonable based on profitablity and compilation time
1818 if (!mem_ref_count_reasonable_p (ninsns
, mem_ref_count
))
1821 /* Step 2: estimate the reuse effects. */
1822 prune_by_reuse (refs
);
1824 if (nothing_to_prefetch_p (refs
))
1827 determine_loop_nest_reuse (loop
, refs
, no_other_refs
);
1829 /* Step 3: determine unroll factor. */
1830 unroll_factor
= determine_unroll_factor (loop
, refs
, ninsns
, &desc
,
1833 /* Estimate prefetch count for the unrolled loop. */
1834 prefetch_count
= estimate_prefetch_count (refs
, unroll_factor
);
1835 if (prefetch_count
== 0)
1838 if (dump_file
&& (dump_flags
& TDF_DETAILS
))
1839 fprintf (dump_file
, "Ahead %d, unroll factor %d, trip count "
1840 HOST_WIDE_INT_PRINT_DEC
"\n"
1841 "insn count %d, mem ref count %d, prefetch count %d\n",
1842 ahead
, unroll_factor
, est_niter
,
1843 ninsns
, mem_ref_count
, prefetch_count
);
1845 /* Prefetching is not likely to be profitable if the instruction to prefetch
1846 ratio is too small. */
1847 if (insn_to_prefetch_ratio_too_small_p (ninsns
, prefetch_count
,
1851 mark_nontemporal_stores (loop
, refs
);
1853 /* Step 4: what to prefetch? */
1854 if (!schedule_prefetches (refs
, unroll_factor
, ahead
))
1857 /* Step 5: unroll the loop. TODO -- peeling of first and last few
1858 iterations so that we do not issue superfluous prefetches. */
1859 if (unroll_factor
!= 1)
1861 tree_unroll_loop (loop
, unroll_factor
,
1862 single_dom_exit (loop
), &desc
);
1866 /* Step 6: issue the prefetches. */
1867 issue_prefetches (refs
, unroll_factor
, ahead
);
1870 release_mem_refs (refs
);
1874 /* Issue prefetch instructions for array references in loops. */
1877 tree_ssa_prefetch_arrays (void)
1881 bool unrolled
= false;
1885 /* It is possible to ask compiler for say -mtune=i486 -march=pentium4.
1886 -mtune=i486 causes us having PREFETCH_BLOCK 0, since this is part
1887 of processor costs and i486 does not have prefetch, but
1888 -march=pentium4 causes HAVE_prefetch to be true. Ugh. */
1889 || PREFETCH_BLOCK
== 0)
1892 if (dump_file
&& (dump_flags
& TDF_DETAILS
))
1894 fprintf (dump_file
, "Prefetching parameters:\n");
1895 fprintf (dump_file
, " simultaneous prefetches: %d\n",
1896 SIMULTANEOUS_PREFETCHES
);
1897 fprintf (dump_file
, " prefetch latency: %d\n", PREFETCH_LATENCY
);
1898 fprintf (dump_file
, " prefetch block size: %d\n", PREFETCH_BLOCK
);
1899 fprintf (dump_file
, " L1 cache size: %d lines, %d kB\n",
1900 L1_CACHE_SIZE_BYTES
/ L1_CACHE_LINE_SIZE
, L1_CACHE_SIZE
);
1901 fprintf (dump_file
, " L1 cache line size: %d\n", L1_CACHE_LINE_SIZE
);
1902 fprintf (dump_file
, " L2 cache size: %d kB\n", L2_CACHE_SIZE
);
1903 fprintf (dump_file
, " min insn-to-prefetch ratio: %d \n",
1904 MIN_INSN_TO_PREFETCH_RATIO
);
1905 fprintf (dump_file
, " min insn-to-mem ratio: %d \n",
1906 PREFETCH_MIN_INSN_TO_MEM_RATIO
);
1907 fprintf (dump_file
, "\n");
1910 initialize_original_copy_tables ();
1912 if (!builtin_decl_explicit_p (BUILT_IN_PREFETCH
))
1914 tree type
= build_function_type_list (void_type_node
,
1915 const_ptr_type_node
, NULL_TREE
);
1916 tree decl
= add_builtin_function ("__builtin_prefetch", type
,
1917 BUILT_IN_PREFETCH
, BUILT_IN_NORMAL
,
1919 DECL_IS_NOVOPS (decl
) = true;
1920 set_builtin_decl (BUILT_IN_PREFETCH
, decl
, false);
1923 /* We assume that size of cache line is a power of two, so verify this
1925 gcc_assert ((PREFETCH_BLOCK
& (PREFETCH_BLOCK
- 1)) == 0);
1927 FOR_EACH_LOOP (li
, loop
, LI_FROM_INNERMOST
)
1929 if (dump_file
&& (dump_flags
& TDF_DETAILS
))
1930 fprintf (dump_file
, "Processing loop %d:\n", loop
->num
);
1932 unrolled
|= loop_prefetch_arrays (loop
);
1934 if (dump_file
&& (dump_flags
& TDF_DETAILS
))
1935 fprintf (dump_file
, "\n\n");
1941 todo_flags
|= TODO_cleanup_cfg
;
1944 free_original_copy_tables ();