public inbox for gcc-cvs@sourceware.org
help / color / mirror / Atom feed
* [gcc(refs/users/rguenth/heads/vect-force-slp)] Allow bigger SLP graphs
@ 2023-10-19 13:29 Richard Biener
0 siblings, 0 replies; 6+ messages in thread
From: Richard Biener @ 2023-10-19 13:29 UTC (permalink / raw)
To: gcc-cvs
https://gcc.gnu.org/g:46e34eee40aea108f3db710382545895834f0000
commit 46e34eee40aea108f3db710382545895834f0000
Author: Richard Biener <rguenther@suse.de>
Date: Fri Sep 29 13:05:01 2023 +0200
Allow bigger SLP graphs
When doing single-lane SLP discovery only we're easily running into
the SLP graph size limit when patterns are involved. The following
ups the limit from the number of scalar stmts to the number of
scalar or pattern stmts by using the number of stmt_vec_infos created.
* tree-vect-loop.cc (vect_analyze_loop_2): Use the number
of stmt_vec_infos created to limit the SLP graph size.
Diff:
---
gcc/tree-vect-loop.cc | 6 ++++--
1 file changed, 4 insertions(+), 2 deletions(-)
diff --git a/gcc/tree-vect-loop.cc b/gcc/tree-vect-loop.cc
index 8877ebde2466..0de43168460a 100644
--- a/gcc/tree-vect-loop.cc
+++ b/gcc/tree-vect-loop.cc
@@ -2813,8 +2813,10 @@ vect_analyze_loop_2 (loop_vec_info loop_vinfo, bool &fatal,
if (slp)
{
/* Check the SLP opportunities in the loop, analyze and build
- SLP trees. */
- ok = vect_analyze_slp (loop_vinfo, LOOP_VINFO_N_STMTS (loop_vinfo));
+ SLP trees. Use the number of stmt_vec_infos as graph limit
+ since that also includes pattern stmts which LOOP_VINFO_N_STMTS
+ does not. */
+ ok = vect_analyze_slp (loop_vinfo, loop_vinfo->stmt_vec_infos.length ());
if (!ok)
return ok;
^ permalink raw reply [flat|nested] 6+ messages in thread
* [gcc(refs/users/rguenth/heads/vect-force-slp)] Allow bigger SLP graphs
@ 2024-02-23 7:31 Richard Biener
0 siblings, 0 replies; 6+ messages in thread
From: Richard Biener @ 2024-02-23 7:31 UTC (permalink / raw)
To: gcc-cvs
https://gcc.gnu.org/g:305009b272b20c08eff8573445ccf70198170def
commit 305009b272b20c08eff8573445ccf70198170def
Author: Richard Biener <rguenther@suse.de>
Date: Fri Sep 29 13:05:01 2023 +0200
Allow bigger SLP graphs
When doing single-lane SLP discovery only we're easily running into
the SLP graph size limit when patterns are involved. The following
ups the limit from the number of scalar stmts to the number of
scalar or pattern stmts by using the number of stmt_vec_infos created.
* tree-vect-loop.cc (vect_analyze_loop_2): Use the number
of stmt_vec_infos created to limit the SLP graph size.
Diff:
---
gcc/tree-vect-loop.cc | 6 ++++--
1 file changed, 4 insertions(+), 2 deletions(-)
diff --git a/gcc/tree-vect-loop.cc b/gcc/tree-vect-loop.cc
index 35f1f8c7d424..405c7fb16180 100644
--- a/gcc/tree-vect-loop.cc
+++ b/gcc/tree-vect-loop.cc
@@ -2855,8 +2855,10 @@ vect_analyze_loop_2 (loop_vec_info loop_vinfo, bool &fatal,
if (slp)
{
/* Check the SLP opportunities in the loop, analyze and build
- SLP trees. */
- ok = vect_analyze_slp (loop_vinfo, LOOP_VINFO_N_STMTS (loop_vinfo));
+ SLP trees. Use the number of stmt_vec_infos as graph limit
+ since that also includes pattern stmts which LOOP_VINFO_N_STMTS
+ does not. */
+ ok = vect_analyze_slp (loop_vinfo, loop_vinfo->stmt_vec_infos.length ());
if (!ok)
return ok;
^ permalink raw reply [flat|nested] 6+ messages in thread
* [gcc(refs/users/rguenth/heads/vect-force-slp)] Allow bigger SLP graphs
@ 2023-11-09 13:03 Richard Biener
0 siblings, 0 replies; 6+ messages in thread
From: Richard Biener @ 2023-11-09 13:03 UTC (permalink / raw)
To: gcc-cvs
https://gcc.gnu.org/g:74ef772d49727647f66fffe866e76c7f861a1889
commit 74ef772d49727647f66fffe866e76c7f861a1889
Author: Richard Biener <rguenther@suse.de>
Date: Fri Sep 29 13:05:01 2023 +0200
Allow bigger SLP graphs
When doing single-lane SLP discovery only we're easily running into
the SLP graph size limit when patterns are involved. The following
ups the limit from the number of scalar stmts to the number of
scalar or pattern stmts by using the number of stmt_vec_infos created.
* tree-vect-loop.cc (vect_analyze_loop_2): Use the number
of stmt_vec_infos created to limit the SLP graph size.
Diff:
---
gcc/tree-vect-loop.cc | 6 ++++--
1 file changed, 4 insertions(+), 2 deletions(-)
diff --git a/gcc/tree-vect-loop.cc b/gcc/tree-vect-loop.cc
index 8abc1937d74..7ce2d7411e3 100644
--- a/gcc/tree-vect-loop.cc
+++ b/gcc/tree-vect-loop.cc
@@ -2816,8 +2816,10 @@ vect_analyze_loop_2 (loop_vec_info loop_vinfo, bool &fatal,
if (slp)
{
/* Check the SLP opportunities in the loop, analyze and build
- SLP trees. */
- ok = vect_analyze_slp (loop_vinfo, LOOP_VINFO_N_STMTS (loop_vinfo));
+ SLP trees. Use the number of stmt_vec_infos as graph limit
+ since that also includes pattern stmts which LOOP_VINFO_N_STMTS
+ does not. */
+ ok = vect_analyze_slp (loop_vinfo, loop_vinfo->stmt_vec_infos.length ());
if (!ok)
return ok;
^ permalink raw reply [flat|nested] 6+ messages in thread
* [gcc(refs/users/rguenth/heads/vect-force-slp)] Allow bigger SLP graphs
@ 2023-11-02 13:59 Richard Biener
0 siblings, 0 replies; 6+ messages in thread
From: Richard Biener @ 2023-11-02 13:59 UTC (permalink / raw)
To: gcc-cvs
https://gcc.gnu.org/g:4a2309e2599ee0dce49187ad89d90bf373950ee3
commit 4a2309e2599ee0dce49187ad89d90bf373950ee3
Author: Richard Biener <rguenther@suse.de>
Date: Fri Sep 29 13:05:01 2023 +0200
Allow bigger SLP graphs
When doing single-lane SLP discovery only we're easily running into
the SLP graph size limit when patterns are involved. The following
ups the limit from the number of scalar stmts to the number of
scalar or pattern stmts by using the number of stmt_vec_infos created.
* tree-vect-loop.cc (vect_analyze_loop_2): Use the number
of stmt_vec_infos created to limit the SLP graph size.
Diff:
---
gcc/tree-vect-loop.cc | 6 ++++--
1 file changed, 4 insertions(+), 2 deletions(-)
diff --git a/gcc/tree-vect-loop.cc b/gcc/tree-vect-loop.cc
index 3b28c826b3b9..77215a175a3e 100644
--- a/gcc/tree-vect-loop.cc
+++ b/gcc/tree-vect-loop.cc
@@ -2813,8 +2813,10 @@ vect_analyze_loop_2 (loop_vec_info loop_vinfo, bool &fatal,
if (slp)
{
/* Check the SLP opportunities in the loop, analyze and build
- SLP trees. */
- ok = vect_analyze_slp (loop_vinfo, LOOP_VINFO_N_STMTS (loop_vinfo));
+ SLP trees. Use the number of stmt_vec_infos as graph limit
+ since that also includes pattern stmts which LOOP_VINFO_N_STMTS
+ does not. */
+ ok = vect_analyze_slp (loop_vinfo, loop_vinfo->stmt_vec_infos.length ());
if (!ok)
return ok;
^ permalink raw reply [flat|nested] 6+ messages in thread
* [gcc(refs/users/rguenth/heads/vect-force-slp)] Allow bigger SLP graphs
@ 2023-10-16 12:49 Richard Biener
0 siblings, 0 replies; 6+ messages in thread
From: Richard Biener @ 2023-10-16 12:49 UTC (permalink / raw)
To: gcc-cvs
https://gcc.gnu.org/g:64a5842cd8c64508c96e33d9f97118d64e096740
commit 64a5842cd8c64508c96e33d9f97118d64e096740
Author: Richard Biener <rguenther@suse.de>
Date: Fri Sep 29 13:05:01 2023 +0200
Allow bigger SLP graphs
When doing single-lane SLP discovery only we're easily running into
the SLP graph size limit when patterns are involved. The following
ups the limit from the number of scalar stmts to the number of
scalar or pattern stmts by using the number of stmt_vec_infos created.
* tree-vect-loop.cc (vect_analyze_loop_2): Use the number
of stmt_vec_infos created to limit the SLP graph size.
Diff:
---
gcc/tree-vect-loop.cc | 6 ++++--
1 file changed, 4 insertions(+), 2 deletions(-)
diff --git a/gcc/tree-vect-loop.cc b/gcc/tree-vect-loop.cc
index 89bdcaa09104..7a912716c2d1 100644
--- a/gcc/tree-vect-loop.cc
+++ b/gcc/tree-vect-loop.cc
@@ -2723,8 +2723,10 @@ vect_analyze_loop_2 (loop_vec_info loop_vinfo, bool &fatal,
if (slp)
{
/* Check the SLP opportunities in the loop, analyze and build
- SLP trees. */
- ok = vect_analyze_slp (loop_vinfo, LOOP_VINFO_N_STMTS (loop_vinfo));
+ SLP trees. Use the number of stmt_vec_infos as graph limit
+ since that also includes pattern stmts which LOOP_VINFO_N_STMTS
+ does not. */
+ ok = vect_analyze_slp (loop_vinfo, loop_vinfo->stmt_vec_infos.length ());
if (!ok)
return ok;
^ permalink raw reply [flat|nested] 6+ messages in thread
* [gcc(refs/users/rguenth/heads/vect-force-slp)] Allow bigger SLP graphs
@ 2023-10-06 7:07 Richard Biener
0 siblings, 0 replies; 6+ messages in thread
From: Richard Biener @ 2023-10-06 7:07 UTC (permalink / raw)
To: gcc-cvs
https://gcc.gnu.org/g:8520f385b0f424d66d20db40ff1bb54cec063802
commit 8520f385b0f424d66d20db40ff1bb54cec063802
Author: Richard Biener <rguenther@suse.de>
Date: Fri Sep 29 13:05:01 2023 +0200
Allow bigger SLP graphs
When doing single-lane SLP discovery only we're easily running into
the SLP graph size limit when patterns are involved. The following
ups the limit from the number of scalar stmts to the number of
scalar or pattern stmts by using the number of stmt_vec_infos created.
* tree-vect-loop.cc (vect_analyze_loop_2): Use the number
of stmt_vec_infos created to limit the SLP graph size.
Diff:
---
gcc/tree-vect-loop.cc | 6 ++++--
1 file changed, 4 insertions(+), 2 deletions(-)
diff --git a/gcc/tree-vect-loop.cc b/gcc/tree-vect-loop.cc
index 23c6e8259e7..490d4282ac8 100644
--- a/gcc/tree-vect-loop.cc
+++ b/gcc/tree-vect-loop.cc
@@ -2723,8 +2723,10 @@ vect_analyze_loop_2 (loop_vec_info loop_vinfo, bool &fatal,
if (slp)
{
/* Check the SLP opportunities in the loop, analyze and build
- SLP trees. */
- ok = vect_analyze_slp (loop_vinfo, LOOP_VINFO_N_STMTS (loop_vinfo));
+ SLP trees. Use the number of stmt_vec_infos as graph limit
+ since that also includes pattern stmts which LOOP_VINFO_N_STMTS
+ does not. */
+ ok = vect_analyze_slp (loop_vinfo, loop_vinfo->stmt_vec_infos.length ());
if (!ok)
return ok;
^ permalink raw reply [flat|nested] 6+ messages in thread
end of thread, other threads:[~2024-02-23 7:31 UTC | newest]
Thread overview: 6+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2023-10-19 13:29 [gcc(refs/users/rguenth/heads/vect-force-slp)] Allow bigger SLP graphs Richard Biener
-- strict thread matches above, loose matches on Subject: below --
2024-02-23 7:31 Richard Biener
2023-11-09 13:03 Richard Biener
2023-11-02 13:59 Richard Biener
2023-10-16 12:49 Richard Biener
2023-10-06 7:07 Richard Biener
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).