public inbox for gcc-patches@gcc.gnu.org
 help / color / mirror / Atom feed
From: "Roger Sayle" <roger@nextmovesoftware.com>
To: "'GCC Patches'" <gcc-patches@gcc.gnu.org>
Subject: [PATCH] Optimize (X<<C)+(Y<<C) as (X+Y)<<C for signed addition.
Date: Tue, 13 Sep 2022 19:37:20 +0100	[thread overview]
Message-ID: <002d01d8c79f$dc5fe830$951fb890$@nextmovesoftware.com> (raw)

[-- Attachment #1: Type: text/plain, Size: 1077 bytes --]


This patch tweaks the match.pd transformation previously added to fold
(X<<C)+(Y<<C) as (X+Y)<<C that was previously restricted to unsigned
(wrapping) types, to also allow signed integer types provided that they
don't trap and the overflow needn't be preserved for sanitization.
i.e. this should now apply (by default) for "int x,y;", but is disabled
with -ftrapv.

This unsigned transformation has baked on mainline without problems, so it's
time to turn up the heat...  LLVM, icc and MSVC perform this optimization.


This patch has been tested on x86_64-pc-linux-gnu with make bootstrap and
make -k check, both with and without --target_board=unix{-m32}, with no
new failures.  Ok for mainline?


2022-09-13  Roger Sayle  <roger@nextmovesoftware.com>

gcc/ChangeLog
        * match.pd (op (lshift @0 @1) (lshift @2 @1)): Optimize the
        expression (X<<C) + (Y<<C) to (X+Y)<<C for signed types, that
        may have undefined overflow, provided that they don't trap.

gcc/testsuite/ChangeLog
        * gcc.dg/pr71343-3.c: New test case.


Thanks in advance,
Roger
--


[-- Attachment #2: patchfd3.txt --]
[-- Type: text/plain, Size: 1011 bytes --]

diff --git a/gcc/match.pd b/gcc/match.pd
index 17318f52..e1e6af9 100644
--- a/gcc/match.pd
+++ b/gcc/match.pd
@@ -988,7 +988,8 @@ DEFINE_INT_AND_FLOAT_ROUND_FN (RINT)
   (simplify
     (op (lshift:s @0 @1) (lshift:s @2 @1))
     (if (INTEGRAL_TYPE_P (type)
-	 && TYPE_OVERFLOW_WRAPS (type)
+	 && !TYPE_OVERFLOW_SANITIZED (type)
+	 && !TYPE_OVERFLOW_TRAPS (type)
 	 && !TYPE_SATURATING (type))
       (lshift (op @0 @2) @1))))
 
diff --git a/gcc/testsuite/gcc.dg/pr71343-3.c b/gcc/testsuite/gcc.dg/pr71343-3.c
new file mode 100644
index 0000000..d0bdbae
--- /dev/null
+++ b/gcc/testsuite/gcc.dg/pr71343-3.c
@@ -0,0 +1,15 @@
+/* { dg-do compile } */
+/* { dg-options "-O2 -fdump-tree-optimized" } */
+
+int foo(int a, int b)
+{
+  return (a << 2) + (b << 2);
+}
+
+int bar(int a, int b)
+{
+  return (a << 2) + (b << 2) == (a + b) << 2;
+}
+
+/* { dg-final { scan-tree-dump-times " << " 1 "optimized" } } */
+/* { dg-final { scan-tree-dump-times "return 1" 1 "optimized" } } */

             reply	other threads:[~2022-09-13 18:37 UTC|newest]

Thread overview: 4+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2022-09-13 18:37 Roger Sayle [this message]
2022-09-14  8:28 ` Richard Biener
2022-09-14 21:42 ` Marc Glisse
2022-09-15  6:39   ` Richard Biener

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='002d01d8c79f$dc5fe830$951fb890$@nextmovesoftware.com' \
    --to=roger@nextmovesoftware.com \
    --cc=gcc-patches@gcc.gnu.org \
    /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).