From: Zhongyunde <zhongyunde@huawei.com>
To: Andrew Pinski <pinskia@gcc.gnu.org>
Cc: "hongtao.liu@intel.com" <hongtao.liu@intel.com>,
"gcc-patches@gcc.gnu.org" <gcc-patches@gcc.gnu.org>,
"Zhangwen(Esan)" <zwzhangwen.zhang@huawei.com>,
"Weiwei (weiwei, Compiler)" <weiwei64@huawei.com>,
"zhong_1985624@163.com" <zhong_1985624@163.com>
Subject: RE: [PATCH] [PHIOPT] Add A ? B + CST : B match and simplify optimizations
Date: Sat, 5 Nov 2022 09:03:10 +0000 [thread overview]
Message-ID: <7b6d19fcef784a019c89e77c9917c5a8@huawei.com> (raw)
In-Reply-To: <CA+=Sn1kwT-KE5Q53xk8ZEGLnaVk_vwD7iaCtAEZSac63Z=-SOA@mail.gmail.com>
[-- Attachment #1: Type: text/plain, Size: 1343 bytes --]
> -----Original Message-----
> From: Andrew Pinski [mailto:pinskia@gcc.gnu.org]
> Sent: Saturday, November 5, 2022 2:34 PM
> To: Zhongyunde <zhongyunde@huawei.com>
> Cc: hongtao.liu@intel.com; gcc-patches@gcc.gnu.org; Zhangwen(Esan)
> <zwzhangwen.zhang@huawei.com>; Weiwei (weiwei, Compiler)
> <weiwei64@huawei.com>; zhong_1985624@163.com
> Subject: Re: [PATCH] [PHIOPT] Add A ? B + CST : B match and simplify
> optimizations
>
> On Fri, Nov 4, 2022 at 11:17 PM Zhongyunde <zhongyunde@huawei.com>
> wrote:
> >
> > hi,
> > This patch is try to fix the issue
> > https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107190,
> > would you like to give me some suggestion, thanks.
>
> This seems like a "simplified" version of
> https://gcc.gnu.org/pipermail/gcc-patches/2021-November/584411.html
> which just handles power of 2 constants where we know the cond will be
> removed.
> We could do even more "simplified" of 1 if needed really.
> What is the IR before PHI-OPT? Is it just + 1?
Thanks for your attention. It is + 4294967296 before PHI-OPT (See detail https://gcc.godbolt.org/z/6zEc6ja1z)
So we should keep matching the power of 2 constants ?
> Also your pattern can be simplified to use integer_pow2p in the match part
> instead of INTEGER_CST.
>
Apply your comment, thanks
> Thanks,
> Andrew
[-- Attachment #2: 0001-PHIOPT-Add-A-B-op-CST-B-match-and-simplify-optimizat.patch --]
[-- Type: application/octet-stream, Size: 3072 bytes --]
From 1dfbda734390d8398a12a455028149b058076a51 Mon Sep 17 00:00:00 2001
From: zhongyunde <zhongyunde@huawei.com>
Date: Sat, 5 Nov 2022 13:22:33 +0800
Subject: [PATCH] [PHIOPT] Add A ? B op CST : B match and simplify
optimizations
Refer to commit b6bdd7a4, use pattern match to simple
A ? B op CST : B (where CST is power of 2) simplifications.
Fixes https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107190
gcc/
* match.pd (A ? B op CST : B): Add simplifcations for A ? B op POW2 : B
gcc/testsuite/
* gcc.dg/pr107190.c: New test.
---
gcc/match.pd | 21 +++++++++++++++++++++
gcc/testsuite/gcc.dg/pr107190.c | 27 +++++++++++++++++++++++++++
2 files changed, 48 insertions(+)
create mode 100644 gcc/testsuite/gcc.dg/pr107190.c
diff --git a/gcc/match.pd b/gcc/match.pd
index 194ba8f5188..7ff393f371f 100644
--- a/gcc/match.pd
+++ b/gcc/match.pd
@@ -4503,6 +4503,27 @@ DEFINE_INT_AND_FLOAT_ROUND_FN (RINT)
&& INTEGRAL_TYPE_P (TREE_TYPE (@0)))
(cond @1 (convert @2) (convert @3))))
+#if GIMPLE
+(if (canonicalize_math_p ())
+/* These patterns are mostly used by PHIOPT to move some operations outside of
+ the if statements. They should be done late because it gives jump threading
+ and few other passes to reduce what is going on. */
+/* a ? x op C : x -> x op << log2(C) when C is power of 2. */
+ (for op (plus minus bit_ior bit_xor lshift rshift lrotate rrotate)
+ (simplify
+ (cond @0 (op:s @1 integer_pow2p@2) @1)
+ /* powerof2cst */
+ (if (INTEGRAL_TYPE_P (type))
+ (with {
+ tree shift = build_int_cst (integer_type_node, tree_log2 (@2));
+ }
+ (op @1 (lshift (convert (convert:boolean_type_node @0)) { shift; })))
+ )
+ )
+ )
+)
+#endif
+
/* Simplification moved from fold_cond_expr_with_comparison. It may also
be extended. */
/* This pattern implements two kinds simplification:
diff --git a/gcc/testsuite/gcc.dg/pr107190.c b/gcc/testsuite/gcc.dg/pr107190.c
new file mode 100644
index 00000000000..235b2761a02
--- /dev/null
+++ b/gcc/testsuite/gcc.dg/pr107190.c
@@ -0,0 +1,27 @@
+/* { dg-do compile } */
+/* { dg-options "-O2 -fexpensive-optimizations -fdump-tree-phiopt2-details" } */
+
+# define BN_BITS4 32
+# define BN_MASK2 (0xffffffffffffffffL)
+# define BN_MASK2l (0xffffffffL)
+# define BN_MASK2h (0xffffffff00000000L)
+# define BN_MASK2h1 (0xffffffff80000000L)
+# define LBITS(a) ((a)&BN_MASK2l)
+# define HBITS(a) (((a)>>BN_BITS4)&BN_MASK2l)
+# define L2HBITS(a) (((a)<<BN_BITS4)&BN_MASK2)
+
+unsigned int test_m(unsigned long in0, unsigned long in1) {
+ unsigned long m, m1, lt, ht, bl, bh;
+ lt = LBITS(in0);
+ ht = HBITS(in0);
+ bl = LBITS(in1);
+ bh = HBITS(in1);
+ m = bh * lt;
+ m1 = bl * ht;
+ ht = bh * ht;
+ m = (m + m1) & BN_MASK2;
+ if (m < m1) ht += L2HBITS((unsigned long)1);
+ return ht + m;
+}
+
+/* { dg-final { scan-tree-dump "COND_EXPR in block 2 and PHI in block 4 converted to straightline code" "phiopt2" } } */
--
2.19.1
next prev parent reply other threads:[~2022-11-05 9:03 UTC|newest]
Thread overview: 18+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-09-05 21:44 [PATCH] riscv: implement TARGET_MODE_REP_EXTENDED Philipp Tomsich
2022-09-06 11:39 ` Alexander Monakov
2022-09-16 23:48 ` Jeff Law
2022-09-17 7:59 ` Palmer Dabbelt
2022-11-04 23:00 ` Philipp Tomsich
2022-11-05 6:16 ` [PATCH] [PHIOPT] Add A ? B + CST : B match and simplify optimizations Zhongyunde
2022-11-05 6:34 ` Andrew Pinski
2022-11-05 9:03 ` Zhongyunde [this message]
2022-11-08 14:58 ` Richard Biener
2022-11-08 15:51 ` 钟云德
2022-11-09 8:00 ` Richard Biener
2022-11-07 13:55 ` [PATCH] riscv: implement TARGET_MODE_REP_EXTENDED Alexander Monakov
2022-11-08 23:45 ` Philipp Tomsich
2022-11-09 17:21 ` Alexander Monakov
2022-11-20 16:09 ` Jeff Law
2022-11-21 13:49 ` Alexander Monakov
2022-11-21 14:56 ` Jeff Law
2022-11-21 15:33 ` Alexander Monakov
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=7b6d19fcef784a019c89e77c9917c5a8@huawei.com \
--to=zhongyunde@huawei.com \
--cc=gcc-patches@gcc.gnu.org \
--cc=hongtao.liu@intel.com \
--cc=pinskia@gcc.gnu.org \
--cc=weiwei64@huawei.com \
--cc=zhong_1985624@163.com \
--cc=zwzhangwen.zhang@huawei.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).