[2/5] Avoid bogus SLP outer loop vectorization

Message ID	20240524134255.23E1A13A3D@imap1.dmz-prg2.suse.org
State	New
Headers	show Return-Path: <gcc-patches-bounces+incoming=patchwork.ozlabs.org@gcc.gnu.org> DMARC-Filter: OpenDMARC Filter v1.4.2 sourceware.org 369E03865C10 Date: Fri, 24 May 2024 15:42:50 +0200 (CEST) From: Richard Biener <rguenther@suse.de> To: gcc-patches@gcc.gnu.org Subject: [PATCH 2/5] Avoid bogus SLP outer loop vectorization MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Message-Id: <20240524134255.23E1A13A3D@imap1.dmz-prg2.suse.org> default: False [-4.20 / 50.00]; BAYES_HAM(-2.90)[99.56%]; NEURAL_HAM_LONG(-1.00)[-1.000]; NEURAL_HAM_SHORT(-0.20)[-1.000]; MIME_GOOD(-0.10)[text/plain]; RCVD_VIA_SMTP_AUTH(0.00)[]; ARC_NA(0.00)[]; RCPT_COUNT_ONE(0.00)[1]; MIME_TRACE(0.00)[0:+]; MISSING_XM_UA(0.00)[]; RCVD_TLS_ALL(0.00)[]; DKIM_SIGNED(0.00)[suse.de:s=susede2_rsa,suse.de:s=susede2_ed25519]; PREVIOUSLY_DELIVERED(0.00)[gcc-patches@gcc.gnu.org]; FROM_EQ_ENVFROM(0.00)[]; FROM_HAS_DN(0.00)[]; FUZZY_BLOCKED(0.00)[rspamd.com]; RCVD_COUNT_TWO(0.00)[2]; TO_MATCH_ENVRCPT_ALL(0.00)[]; TO_DN_NONE(0.00)[]; DBL_BLOCKED_OPENRESOLVER(0.00)[imap1.dmz-prg2.suse.org:helo] Precedence: list Errors-To: gcc-patches-bounces+incoming=patchwork.ozlabs.org@gcc.gnu.org
Series	[1/5] Do single-lane SLP discovery for reductions \| expand [1/5] Do single-lane SLP discovery for reductions [2/5] Avoid bogus SLP outer loop vectorization [3/5] Reduce single-lane SLP testresult noise [4/5] Allow optimized SLP reduction epilog with single-lane reductions [5/5] RISC-V: tree-optimization/65518 - extend fix to SLP

Message ID

20240524134255.23E1A13A3D@imap1.dmz-prg2.suse.org

State

New

Headers

DMARC-Filter: OpenDMARC Filter v1.4.2 sourceware.org 369E03865C10
Date: Fri, 24 May 2024 15:42:50 +0200 (CEST)
From: Richard Biener <rguenther@suse.de>
To: gcc-patches@gcc.gnu.org
Subject: [PATCH 2/5] Avoid bogus SLP outer loop vectorization
MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Message-Id: <20240524134255.23E1A13A3D@imap1.dmz-prg2.suse.org>
Precedence: list
Errors-To: gcc-patches-bounces+incoming=patchwork.ozlabs.org@gcc.gnu.org

Series

[1/5] Do single-lane SLP discovery for reductions | expand

Commit Message

Richard Biener May 24, 2024, 1:42 p.m. UTC

This fixes the check for multiple types which go wrong I think
because of bogus pointer IV increments when there are multiple
copies of vector stmts in the inner loop.

	* tree-vect-stmts.cc (vectorizable_load): Avoid outer loop
	SLP vectorization with multi-copy vector stmts in the inner
	loop.
	(vectorizable_store): Likewise.
---
 gcc/tree-vect-stmts.cc | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)

diff --git a/gcc/tree-vect-stmts.cc b/gcc/tree-vect-stmts.cc
index 4219ad832db..1c30a0388ca 100644
--- a/gcc/tree-vect-stmts.cc
+++ b/gcc/tree-vect-stmts.cc
@@ -8196,7 +8196,9 @@  vectorizable_store (vec_info *vinfo,
   gcc_assert (ncopies >= 1);
 
   /* FORNOW.  This restriction should be relaxed.  */
-  if (loop && nested_in_vect_loop_p (loop, stmt_info) && ncopies > 1)
+  if (loop
+      && nested_in_vect_loop_p (loop, stmt_info)
+      && (ncopies > 1 || (slp && SLP_TREE_NUMBER_OF_VEC_STMTS (slp_node) > 1)))
     {
       if (dump_enabled_p ())
 	dump_printf_loc (MSG_MISSED_OPTIMIZATION, vect_location,
@@ -9939,7 +9941,8 @@  vectorizable_load (vec_info *vinfo,
   gcc_assert (ncopies >= 1);
 
   /* FORNOW. This restriction should be relaxed.  */
-  if (nested_in_vect_loop && ncopies > 1)
+  if (nested_in_vect_loop
+      && (ncopies > 1 || (slp && SLP_TREE_NUMBER_OF_VEC_STMTS (slp_node) > 1)))
     {
       if (dump_enabled_p ())
         dump_printf_loc (MSG_MISSED_OPTIMIZATION, vect_location,

[2/5] Avoid bogus SLP outer loop vectorization

Commit Message

Patch