From: Robin Dapp <rdapp@linux.vnet.ibm.com>
To: Richard Biener <richard.guenther@gmail.com>
Cc: GCC Patches <gcc-patches@gcc.gnu.org>,
"Bin.Cheng" <amker.cheng@gmail.com>
Subject: [PATCH 4/5] Vect peeling cost model
Date: Thu, 11 May 2017 11:20:00 -0000 [thread overview]
Message-ID: <15b3df5a-f6a1-23f0-57ff-1f065420df19@linux.vnet.ibm.com> (raw)
In-Reply-To: <CAFiYyc1LFtniQv9FNXJW7g8uByb8H_4vnrx52oS0kzBaKQedEw@mail.gmail.com>
[-- Attachment #1: Type: text/plain, Size: 289 bytes --]
gcc/ChangeLog:
2017-05-11 Robin Dapp <rdapp@linux.vnet.ibm.com>
* tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
Remove check for supportable_dr_alignment, compute costs for
doing no peeling at all, compare to the best peeling costs so
far and do no peeling if cheaper.
[-- Attachment #2: gcc-peeling-p4.diff --]
[-- Type: text/x-patch, Size: 9291 bytes --]
diff --git a/gcc/tree-vect-data-refs.c b/gcc/tree-vect-data-refs.c
index 1bba2b9..00c4d5d 100644
--- a/gcc/tree-vect-data-refs.c
+++ b/gcc/tree-vect-data-refs.c
@@ -1517,10 +1517,8 @@ vect_enhance_data_refs_alignment (loop_vec_info loop_vinfo)
{
vec<data_reference_p> datarefs = LOOP_VINFO_DATAREFS (loop_vinfo);
struct loop *loop = LOOP_VINFO_LOOP (loop_vinfo);
- enum dr_alignment_support supportable_dr_alignment;
struct data_reference *dr0 = NULL, *first_store = NULL;
struct data_reference *dr;
- struct data_reference *dr0_known_align = NULL;
unsigned int i, j;
bool do_peeling = false;
bool do_versioning = false;
@@ -1601,7 +1599,6 @@ vect_enhance_data_refs_alignment (loop_vec_info loop_vinfo)
&& !STMT_VINFO_GROUPED_ACCESS (stmt_info))
continue;
- supportable_dr_alignment = vect_supportable_dr_alignment (dr, true);
do_peeling = vector_alignment_reachable_p (dr);
if (do_peeling)
{
@@ -1690,16 +1687,6 @@ vect_enhance_data_refs_alignment (loop_vec_info loop_vinfo)
if (!first_store && DR_IS_WRITE (dr))
first_store = dr;
- /* If there are both known and unknown misaligned accesses in the
- loop, we choose peeling amount according to the known
- accesses. */
- if (!supportable_dr_alignment)
- {
- dr0 = dr;
- if (!first_store && DR_IS_WRITE (dr))
- first_store = dr;
- }
-
one_misalignment_unknown = true;
}
}
@@ -1721,81 +1708,85 @@ vect_enhance_data_refs_alignment (loop_vec_info loop_vinfo)
|| loop->inner)
do_peeling = false;
- unsigned int unknown_align_inside_cost = INT_MAX;
- unsigned int unknown_align_outside_cost = INT_MAX;
+ struct _vect_peel_extended_info peel_for_known_alignment;
+ struct _vect_peel_extended_info peel_for_unknown_alignment;
+ struct _vect_peel_extended_info best_peel;
+
+ peel_for_unknown_alignment.inside_cost = INT_MAX;
+ peel_for_unknown_alignment.outside_cost = INT_MAX;
+ peel_for_unknown_alignment.peel_info.count = 0;
if (do_peeling
- && one_misalignment_unknown
- && vect_supportable_dr_alignment (dr0, false))
+ && one_misalignment_unknown)
{
/* Check if the target requires to prefer stores over loads, i.e., if
misaligned stores are more expensive than misaligned loads (taking
drs with same alignment into account). */
- if (first_store && DR_IS_READ (dr0))
- {
- unsigned int load_inside_cost = 0;
- unsigned int load_outside_cost = 0;
- unsigned int store_inside_cost = 0;
- unsigned int store_outside_cost = 0;
- stmt_vector_for_cost dummy;
- dummy.create (2);
- vect_get_peeling_costs_all_drs (dr0,
- &load_inside_cost,
- &load_outside_cost,
- &dummy, vf / 2, vf);
- dummy.release ();
-
+ unsigned int load_inside_cost = 0;
+ unsigned int load_outside_cost = 0;
+ unsigned int store_inside_cost = 0;
+ unsigned int store_outside_cost = 0;
+
+ stmt_vector_for_cost dummy;
+ dummy.create (2);
+ vect_get_peeling_costs_all_drs (dr0,
+ &load_inside_cost,
+ &load_outside_cost,
+ &dummy, vf / 2, vf);
+ dummy.release ();
+
+ if (first_store)
+ {
dummy.create (2);
vect_get_peeling_costs_all_drs (first_store,
&store_inside_cost,
&store_outside_cost,
&dummy, vf / 2, vf);
dummy.release ();
+ }
+ else
+ {
+ store_inside_cost = INT_MAX;
+ store_outside_cost = INT_MAX;
+ }
- if (load_inside_cost > store_inside_cost
- || (load_inside_cost == store_inside_cost
- && load_outside_cost > store_outside_cost))
- {
- dr0 = first_store;
- unknown_align_inside_cost = store_inside_cost;
- unknown_align_outside_cost = store_outside_cost;
- }
- else
- {
- unknown_align_inside_cost = load_inside_cost;
- unknown_align_outside_cost = load_outside_cost;
- }
+ if (load_inside_cost > store_inside_cost
+ || (load_inside_cost == store_inside_cost
+ && load_outside_cost > store_outside_cost))
+ {
+ dr0 = first_store;
+ peel_for_unknown_alignment.inside_cost = store_inside_cost;
+ peel_for_unknown_alignment.outside_cost = store_outside_cost;
+ }
+ else
+ {
+ peel_for_unknown_alignment.inside_cost = load_inside_cost;
+ peel_for_unknown_alignment.outside_cost = load_outside_cost;
+ }
- stmt_vector_for_cost prologue_cost_vec, epilogue_cost_vec;
- prologue_cost_vec.create (2);
- epilogue_cost_vec.create (2);
+ stmt_vector_for_cost prologue_cost_vec, epilogue_cost_vec;
+ prologue_cost_vec.create (2);
+ epilogue_cost_vec.create (2);
- int dummy2;
- unknown_align_outside_cost += vect_get_known_peeling_cost
- (loop_vinfo, vf / 2, &dummy2,
- &LOOP_VINFO_SCALAR_ITERATION_COST (loop_vinfo),
- &prologue_cost_vec, &epilogue_cost_vec);
+ int dummy2;
+ peel_for_unknown_alignment.outside_cost += vect_get_known_peeling_cost
+ (loop_vinfo, vf / 2, &dummy2,
+ &LOOP_VINFO_SCALAR_ITERATION_COST (loop_vinfo),
+ &prologue_cost_vec, &epilogue_cost_vec);
- prologue_cost_vec.release ();
- epilogue_cost_vec.release ();
- }
+ prologue_cost_vec.release ();
+ epilogue_cost_vec.release ();
- /* In case there are only loads with different unknown misalignments, use
- peeling only if it may help to align other accesses in the loop or
- if it may help improving load bandwith when we'd end up using
- unaligned loads. */
- tree dr0_vt = STMT_VINFO_VECTYPE (vinfo_for_stmt (DR_STMT (dr0)));
- if (!first_store
- && !STMT_VINFO_SAME_ALIGN_REFS (
- vinfo_for_stmt (DR_STMT (dr0))).length ()
- && (vect_supportable_dr_alignment (dr0, false)
- != dr_unaligned_supported
- || (builtin_vectorization_cost (vector_load, dr0_vt, 0)
- == builtin_vectorization_cost (unaligned_load, dr0_vt, -1))))
- do_peeling = false;
+ peel_for_unknown_alignment.peel_info.count = 1
+ + STMT_VINFO_SAME_ALIGN_REFS
+ (vinfo_for_stmt (DR_STMT (dr0))).length ();
}
- struct _vect_peel_extended_info peel_for_known_alignment;
+ peel_for_unknown_alignment.peel_info.npeel = 0;
+ peel_for_unknown_alignment.peel_info.dr = dr0;
+
+ best_peel = peel_for_unknown_alignment;
+
peel_for_known_alignment.inside_cost = INT_MAX;
peel_for_known_alignment.outside_cost = INT_MAX;
peel_for_known_alignment.peel_info.count = 0;
@@ -1804,24 +1795,52 @@ vect_enhance_data_refs_alignment (loop_vec_info loop_vinfo)
if (do_peeling && one_misalignment_known)
{
/* Peeling is possible, but there is no data access that is not supported
- unless aligned. So we try to choose the best possible peeling. */
-
- /* Choose the best peeling from the hash table. */
+ unless aligned. So we try to choose the best possible peeling from
+ the hash table. */
peel_for_known_alignment = vect_peeling_hash_choose_best_peeling
(&peeling_htab, loop_vinfo, &npeel, &body_cost_vec);
- dr0_known_align = peel_for_known_alignment.peel_info.dr;
}
/* Compare costs of peeling for known and unknown alignment. */
- if (dr0_known_align != NULL
- && unknown_align_inside_cost >= peel_for_known_alignment.inside_cost)
- {
- dr0 = dr0_known_align;
- if (!npeel)
- do_peeling = false;
- }
+ if (peel_for_known_alignment.peel_info.dr != NULL
+ && peel_for_unknown_alignment.inside_cost
+ >= peel_for_known_alignment.inside_cost)
+ best_peel = peel_for_known_alignment;
+
+ /* Calculate the penalty for no peeling, i.e. leaving everything
+ unaligned.
+ TODO: use something like an adapted vect_get_peeling_costs_all_drs. */
+ unsigned nopeel_inside_cost = 0;
+ unsigned nopeel_outside_cost = 0;
+
+ stmt_vector_for_cost dummy;
+ dummy.create (2);
+ FOR_EACH_VEC_ELT (datarefs, i, dr)
+ vect_get_data_access_cost (dr, &nopeel_inside_cost,
+ &nopeel_outside_cost, &dummy);
+ dummy.release ();
+
+ /* Add epilogue costs. As we do not peel for alignment here, no prologue
+ costs will be recorded. */
+ stmt_vector_for_cost prologue_cost_vec, epilogue_cost_vec;
+ prologue_cost_vec.create (2);
+ epilogue_cost_vec.create (2);
+
+ int dummy2;
+ nopeel_outside_cost += vect_get_known_peeling_cost
+ (loop_vinfo, 0, &dummy2,
+ &LOOP_VINFO_SCALAR_ITERATION_COST (loop_vinfo),
+ &prologue_cost_vec, &epilogue_cost_vec);
+
+ prologue_cost_vec.release ();
+ epilogue_cost_vec.release ();
+
+ npeel = best_peel.peel_info.npeel;
+ dr0 = best_peel.peel_info.dr;
- if (dr0 == NULL)
+ /* If no peeling is not more expensive than the best peeling we
+ have so far, don't perform any peeling. */
+ if (nopeel_inside_cost <= best_peel.inside_cost)
do_peeling = false;
if (do_peeling)
@@ -2000,7 +2019,8 @@ vect_enhance_data_refs_alignment (loop_vec_info loop_vinfo)
break;
}
- supportable_dr_alignment = vect_supportable_dr_alignment (dr, false);
+ enum dr_alignment_support supportable_dr_alignment =
+ vect_supportable_dr_alignment (dr, false);
if (!supportable_dr_alignment)
{
next prev parent reply other threads:[~2017-05-11 11:19 UTC|newest]
Thread overview: 51+ messages / expand[flat|nested] mbox.gz Atom feed top
2017-04-11 14:38 [RFC] S/390: Alignment peeling prolog generation Robin Dapp
2017-04-11 14:57 ` Bin.Cheng
2017-04-11 15:03 ` Robin Dapp
2017-04-11 15:07 ` Bin.Cheng
2017-04-11 16:25 ` Richard Biener
2017-04-12 7:51 ` Robin Dapp
2017-04-12 7:58 ` Richard Biener
2017-05-04 9:04 ` [PATCH 1/3] Vect peeling cost model Robin Dapp
2017-05-05 10:32 ` Richard Biener
2017-05-04 9:04 ` [RFC] S/390: Alignment peeling prolog generation Robin Dapp
2017-05-05 11:04 ` Richard Biener
2017-05-08 16:12 ` Robin Dapp
2017-05-09 10:38 ` Richard Biener
2017-05-11 11:17 ` Robin Dapp
2017-05-11 12:15 ` Richard Biener
2017-05-11 12:16 ` Richard Biener
2017-05-11 12:48 ` Richard Biener
2017-05-11 11:17 ` [PATCH 1/5] Vect peeling cost model Robin Dapp
2017-05-11 11:18 ` [PATCH 2/5] " Robin Dapp
2017-05-11 11:19 ` [PATCH 3/5] " Robin Dapp
2017-05-11 11:20 ` Robin Dapp [this message]
2017-05-11 15:30 ` [PATCH 4/5 v2] " Robin Dapp
2017-05-12 9:36 ` Richard Biener
2017-05-23 15:58 ` [PATCH 1/5 v3] " Robin Dapp
2017-05-23 15:58 ` [PATCH 0/5 " Robin Dapp
2017-05-24 7:51 ` Richard Biener
2017-05-24 11:57 ` Robin Dapp
2017-05-24 13:56 ` Richard Biener
2017-06-03 17:12 ` Andreas Schwab
2017-06-06 7:13 ` Robin Dapp
2017-06-06 17:26 ` Andreas Schwab
2017-06-07 10:50 ` Robin Dapp
2017-06-07 11:43 ` Andreas Schwab
2017-05-23 15:58 ` [PATCH 2/5 " Robin Dapp
2017-05-23 19:25 ` Richard Sandiford
2017-05-24 7:37 ` Robin Dapp
2017-05-24 7:53 ` Richard Sandiford
2017-05-23 15:59 ` [PATCH 4/5 " Robin Dapp
2017-05-31 13:56 ` Christophe Lyon
2017-05-31 14:37 ` Robin Dapp
2017-05-31 14:49 ` Christophe Lyon
2017-05-23 15:59 ` [PATCH 5/5 " Robin Dapp
2017-05-23 16:02 ` [PATCH 3/5 " Robin Dapp
2017-05-11 11:59 ` [PATCH 5/5] " Robin Dapp
2017-05-08 16:13 ` [PATCH 3/4] " Robin Dapp
2017-05-09 10:41 ` Richard Biener
2017-05-08 16:27 ` [PATCH 4/4] " Robin Dapp
2017-05-09 10:55 ` Richard Biener
2017-05-04 9:07 ` [PATCH 2/3] " Robin Dapp
2017-05-05 10:37 ` Richard Biener
2017-05-04 9:14 ` [PATCH 3/3] " Robin Dapp
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=15b3df5a-f6a1-23f0-57ff-1f065420df19@linux.vnet.ibm.com \
--to=rdapp@linux.vnet.ibm.com \
--cc=amker.cheng@gmail.com \
--cc=gcc-patches@gcc.gnu.org \
--cc=richard.guenther@gmail.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
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).