public inbox for gcc-cvs@sourceware.org
help / color / mirror / Atom feed
* [gcc/devel/autopar_devel] vect: Rename fully_masked_p to using_partial_vectors_p
@ 2020-08-22 22:23 Giuliano Belinassi
0 siblings, 0 replies; only message in thread
From: Giuliano Belinassi @ 2020-08-22 22:23 UTC (permalink / raw)
To: gcc-cvs
https://gcc.gnu.org/g:42114f2bed540909672369cc08633158f79f7b49
commit 42114f2bed540909672369cc08633158f79f7b49
Author: Kewen Lin <linkw@gcc.gnu.org>
Date: Thu Jun 4 10:09:01 2020 +0800
vect: Rename fully_masked_p to using_partial_vectors_p
Power supports vector memory access with length (in bytes) instructions.
Like existing fully masking for SVE, it is another approach to vectorize
the loop using partially-populated vectors.
As Richard Sandiford suggested, this patch is to update the existing
fully_masked_p field to using_partial_vectors_p. Introduce one macro
LOOP_VINFO_USING_PARTIAL_VECTORS_P for partial vectorization checking
usage, update the LOOP_VINFO_FULLY_MASKED_P with
LOOP_VINFO_USING_PARTIAL_VECTORS_P && !masks.is_empty() and still use
it for mask-based partial vectors approach specific checks.
Bootstrapped/regtested on aarch64-linux-gnu.
gcc/ChangeLog:
* tree-vect-loop-manip.c (vect_set_loop_condition): Rename
LOOP_VINFO_FULLY_MASKED_P to LOOP_VINFO_USING_PARTIAL_VECTORS_P.
(vect_gen_vector_loop_niters): Likewise.
(vect_do_peeling): Likewise.
* tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Rename
fully_masked_p to using_partial_vectors_p.
(vect_analyze_loop_costing): Rename LOOP_VINFO_FULLY_MASKED_P to
LOOP_VINFO_USING_PARTIAL_VECTORS_P.
(determine_peel_for_niter): Likewise.
(vect_estimate_min_profitable_iters): Likewise.
(vect_transform_loop): Likewise.
* tree-vectorizer.h (LOOP_VINFO_FULLY_MASKED_P): Updated.
(LOOP_VINFO_USING_PARTIAL_VECTORS_P): New macro.
Diff:
---
gcc/tree-vect-loop-manip.c | 6 +++---
gcc/tree-vect-loop.c | 32 +++++++++++++++++---------------
gcc/tree-vectorizer.h | 11 ++++++++---
3 files changed, 28 insertions(+), 21 deletions(-)
diff --git a/gcc/tree-vect-loop-manip.c b/gcc/tree-vect-loop-manip.c
index 8c5e696b995..4bb1c19e8d8 100644
--- a/gcc/tree-vect-loop-manip.c
+++ b/gcc/tree-vect-loop-manip.c
@@ -912,7 +912,7 @@ vect_set_loop_condition (class loop *loop, loop_vec_info loop_vinfo,
gcond *orig_cond = get_loop_exit_condition (loop);
gimple_stmt_iterator loop_cond_gsi = gsi_for_stmt (orig_cond);
- if (loop_vinfo && LOOP_VINFO_FULLY_MASKED_P (loop_vinfo))
+ if (loop_vinfo && LOOP_VINFO_USING_PARTIAL_VECTORS_P (loop_vinfo))
cond_stmt = vect_set_loop_condition_masked (loop, loop_vinfo, niters,
final_iv, niters_maybe_zero,
loop_cond_gsi);
@@ -1939,7 +1939,7 @@ vect_gen_vector_loop_niters (loop_vec_info loop_vinfo, tree niters,
unsigned HOST_WIDE_INT const_vf;
if (vf.is_constant (&const_vf)
- && !LOOP_VINFO_FULLY_MASKED_P (loop_vinfo))
+ && !LOOP_VINFO_USING_PARTIAL_VECTORS_P (loop_vinfo))
{
/* Create: niters >> log2(vf) */
/* If it's known that niters == number of latch executions + 1 doesn't
@@ -2471,7 +2471,7 @@ vect_do_peeling (loop_vec_info loop_vinfo, tree niters, tree nitersm1,
poly_uint64 vf = LOOP_VINFO_VECT_FACTOR (loop_vinfo);
poly_uint64 bound_epilog = 0;
- if (!LOOP_VINFO_FULLY_MASKED_P (loop_vinfo)
+ if (!LOOP_VINFO_USING_PARTIAL_VECTORS_P (loop_vinfo)
&& LOOP_VINFO_PEELING_FOR_NITER (loop_vinfo))
bound_epilog += vf - 1;
if (LOOP_VINFO_PEELING_FOR_GAPS (loop_vinfo))
diff --git a/gcc/tree-vect-loop.c b/gcc/tree-vect-loop.c
index cec903de047..a9a32e55063 100644
--- a/gcc/tree-vect-loop.c
+++ b/gcc/tree-vect-loop.c
@@ -815,7 +815,7 @@ _loop_vec_info::_loop_vec_info (class loop *loop_in, vec_info_shared *shared)
vec_inside_cost (0),
vectorizable (false),
can_use_partial_vectors_p (true),
- fully_masked_p (false),
+ using_partial_vectors_p (false),
peeling_for_gaps (false),
peeling_for_niter (false),
no_data_dependencies (false),
@@ -1634,9 +1634,9 @@ vect_analyze_loop_costing (loop_vec_info loop_vinfo)
class loop *loop = LOOP_VINFO_LOOP (loop_vinfo);
unsigned int assumed_vf = vect_vf_for_cost (loop_vinfo);
- /* Only fully-masked loops can have iteration counts less than the
- vectorization factor. */
- if (!LOOP_VINFO_FULLY_MASKED_P (loop_vinfo))
+ /* Only loops that can handle partially-populated vectors can have iteration
+ counts less than the vectorization factor. */
+ if (!LOOP_VINFO_USING_PARTIAL_VECTORS_P (loop_vinfo))
{
HOST_WIDE_INT max_niter;
@@ -1872,7 +1872,7 @@ determine_peel_for_niter (loop_vec_info loop_vinfo)
th = LOOP_VINFO_COST_MODEL_THRESHOLD (LOOP_VINFO_ORIG_LOOP_INFO
(loop_vinfo));
- if (LOOP_VINFO_FULLY_MASKED_P (loop_vinfo))
+ if (LOOP_VINFO_USING_PARTIAL_VECTORS_P (loop_vinfo))
/* The main loop handles all iterations. */
LOOP_VINFO_PEELING_FOR_NITER (loop_vinfo) = false;
else if (LOOP_VINFO_NITERS_KNOWN_P (loop_vinfo)
@@ -2146,7 +2146,7 @@ start_over:
/* Decide whether to use a fully-masked loop for this vectorization
factor. */
- LOOP_VINFO_FULLY_MASKED_P (loop_vinfo)
+ LOOP_VINFO_USING_PARTIAL_VECTORS_P (loop_vinfo)
= (LOOP_VINFO_CAN_USE_PARTIAL_VECTORS_P (loop_vinfo)
&& vect_verify_full_masking (loop_vinfo));
if (dump_enabled_p ())
@@ -2164,7 +2164,7 @@ start_over:
enough iterations for vectorization. */
if (LOOP_VINFO_PEELING_FOR_GAPS (loop_vinfo)
&& LOOP_VINFO_NITERS_KNOWN_P (loop_vinfo)
- && !LOOP_VINFO_FULLY_MASKED_P (loop_vinfo))
+ && !LOOP_VINFO_USING_PARTIAL_VECTORS_P (loop_vinfo))
{
poly_uint64 vf = LOOP_VINFO_VECT_FACTOR (loop_vinfo);
tree scalar_niters = LOOP_VINFO_NITERSM1 (loop_vinfo);
@@ -2175,10 +2175,11 @@ start_over:
" support peeling for gaps.\n");
}
- /* If we're vectorizing an epilogue loop, we either need a fully-masked
- loop or a loop that has a lower VF than the main loop. */
+ /* If we're vectorizing an epilogue loop, the vectorized loop either needs
+ to be able to handle fewer than VF scalars, or needs to have a lower VF
+ than the main loop. */
if (LOOP_VINFO_EPILOGUE_P (loop_vinfo)
- && !LOOP_VINFO_FULLY_MASKED_P (loop_vinfo)
+ && !LOOP_VINFO_USING_PARTIAL_VECTORS_P (loop_vinfo)
&& maybe_ge (LOOP_VINFO_VECT_FACTOR (loop_vinfo),
LOOP_VINFO_VECT_FACTOR (orig_loop_vinfo)))
return opt_result::failure_at (vect_location,
@@ -2249,7 +2250,7 @@ start_over:
}
/* Niters for at least one iteration of vectorized loop. */
- if (!LOOP_VINFO_FULLY_MASKED_P (loop_vinfo))
+ if (!LOOP_VINFO_USING_PARTIAL_VECTORS_P (loop_vinfo))
niters_th += LOOP_VINFO_VECT_FACTOR (loop_vinfo);
/* One additional iteration because of peeling for gap. */
if (LOOP_VINFO_PEELING_FOR_GAPS (loop_vinfo))
@@ -3826,7 +3827,7 @@ vect_estimate_min_profitable_iters (loop_vec_info loop_vinfo,
" Calculated minimum iters for profitability: %d\n",
min_profitable_iters);
- if (!LOOP_VINFO_FULLY_MASKED_P (loop_vinfo)
+ if (!LOOP_VINFO_USING_PARTIAL_VECTORS_P (loop_vinfo)
&& min_profitable_iters < (assumed_vf + peel_iters_prologue))
/* We want the vectorized loop to execute at least once. */
min_profitable_iters = assumed_vf + peel_iters_prologue;
@@ -8582,7 +8583,7 @@ vect_transform_loop (loop_vec_info loop_vinfo, gimple *loop_vectorized_call)
if (niters_vector == NULL_TREE)
{
if (LOOP_VINFO_NITERS_KNOWN_P (loop_vinfo)
- && !LOOP_VINFO_FULLY_MASKED_P (loop_vinfo)
+ && !LOOP_VINFO_USING_PARTIAL_VECTORS_P (loop_vinfo)
&& known_eq (lowest_vf, vf))
{
niters_vector
@@ -8751,7 +8752,8 @@ vect_transform_loop (loop_vec_info loop_vinfo, gimple *loop_vectorized_call)
/* True if the final iteration might not handle a full vector's
worth of scalar iterations. */
- bool final_iter_may_be_partial = LOOP_VINFO_FULLY_MASKED_P (loop_vinfo);
+ bool final_iter_may_be_partial
+ = LOOP_VINFO_USING_PARTIAL_VECTORS_P (loop_vinfo);
/* The minimum number of iterations performed by the epilogue. This
is 1 when peeling for gaps because we always need a final scalar
iteration. */
@@ -8762,7 +8764,7 @@ vect_transform_loop (loop_vec_info loop_vinfo, gimple *loop_vectorized_call)
int bias_for_lowest = 1 - min_epilogue_iters;
int bias_for_assumed = bias_for_lowest;
int alignment_npeels = LOOP_VINFO_PEELING_FOR_ALIGNMENT (loop_vinfo);
- if (alignment_npeels && LOOP_VINFO_FULLY_MASKED_P (loop_vinfo))
+ if (alignment_npeels && LOOP_VINFO_USING_PARTIAL_VECTORS_P (loop_vinfo))
{
/* When the amount of peeling is known at compile time, the first
iteration will have exactly alignment_npeels active elements.
diff --git a/gcc/tree-vectorizer.h b/gcc/tree-vectorizer.h
index c459280c051..9be4d1599ee 100644
--- a/gcc/tree-vectorizer.h
+++ b/gcc/tree-vectorizer.h
@@ -637,8 +637,9 @@ public:
fewer than VF scalars. */
bool can_use_partial_vectors_p;
- /* True if have decided to use a fully-masked loop. */
- bool fully_masked_p;
+ /* True if we've decided to use partially-populated vectors, so that
+ the vector loop can handle fewer than VF scalars. */
+ bool using_partial_vectors_p;
/* When we have grouped data accesses with gaps, we may introduce invalid
memory accesses. We peel the last iteration of the loop to prevent
@@ -702,7 +703,7 @@ public:
#define LOOP_VINFO_VERSIONING_THRESHOLD(L) (L)->versioning_threshold
#define LOOP_VINFO_VECTORIZABLE_P(L) (L)->vectorizable
#define LOOP_VINFO_CAN_USE_PARTIAL_VECTORS_P(L) (L)->can_use_partial_vectors_p
-#define LOOP_VINFO_FULLY_MASKED_P(L) (L)->fully_masked_p
+#define LOOP_VINFO_USING_PARTIAL_VECTORS_P(L) (L)->using_partial_vectors_p
#define LOOP_VINFO_VECT_FACTOR(L) (L)->vectorization_factor
#define LOOP_VINFO_MAX_VECT_FACTOR(L) (L)->max_vectorization_factor
#define LOOP_VINFO_MASKS(L) (L)->masks
@@ -739,6 +740,10 @@ public:
#define LOOP_VINFO_ORIG_LOOP_INFO(L) (L)->orig_loop_info
#define LOOP_VINFO_SIMD_IF_COND(L) (L)->simd_if_cond
+#define LOOP_VINFO_FULLY_MASKED_P(L) \
+ (LOOP_VINFO_USING_PARTIAL_VECTORS_P (L) \
+ && !LOOP_VINFO_MASKS (L).is_empty ())
+
#define LOOP_REQUIRES_VERSIONING_FOR_ALIGNMENT(L) \
((L)->may_misalign_stmts.length () > 0)
#define LOOP_REQUIRES_VERSIONING_FOR_ALIAS(L) \
^ permalink raw reply [flat|nested] only message in thread
only message in thread, other threads:[~2020-08-22 22:23 UTC | newest]
Thread overview: (only message) (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2020-08-22 22:23 [gcc/devel/autopar_devel] vect: Rename fully_masked_p to using_partial_vectors_p Giuliano Belinassi
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).