From: Richard Biener <rguenther@suse.de>
To: Tamar Christina <Tamar.Christina@arm.com>
Cc: GCC Patches <gcc-patches@gcc.gnu.org>, nd <nd@arm.com>,
Richard Earnshaw <Richard.Earnshaw@foss.arm.com>
Subject: RE: [PATCH]middle-end convert negate + right shift into compare greater.
Date: Thu, 4 Nov 2021 14:06:02 +0100 (CET) [thread overview]
Message-ID: <q9171p92-snq4-qo8p-80n5-p855sq36qs82@fhfr.qr> (raw)
In-Reply-To: <VI1PR08MB5325EC1BF5CEBD010702FF54FF8C9@VI1PR08MB5325.eurprd08.prod.outlook.com>
On Wed, 3 Nov 2021, Tamar Christina wrote:
> Hi,
>
> I have addressed all the feedback and updated patch attached:
>
> Bootstrapped Regtested on aarch64-none-linux-gnu,
> x86_64-pc-linux-gnu and no regressions.
>
> Ok for master?
>
> Thanks,
> Tamar
>
> gcc/ChangeLog:
>
> * match.pd: New negate+shift pattern.
>
> gcc/testsuite/ChangeLog:
>
> * gcc.dg/signbit-2.c: New test.
> * gcc.dg/signbit-3.c: New test.
> * gcc.dg/signbit-4.c: New test.
> * gcc.dg/signbit-5.c: New test.
> * gcc.dg/signbit-6.c: New test.
> * gcc.target/aarch64/signbit-1.c: New test.
>
> --- inline copy of patch ---
>
> diff --git a/gcc/match.pd b/gcc/match.pd
> index 65a6591f75c03333602147bbdf6d59f9ccd4b1e5..fe93500d22e2388889c8c9faf4c58cee95dec7f9 100644
> --- a/gcc/match.pd
> +++ b/gcc/match.pd
> @@ -38,7 +38,8 @@ along with GCC; see the file COPYING3. If not see
> uniform_integer_cst_p
> HONOR_NANS
> uniform_vector_p
> - bitmask_inv_cst_vector_p)
> + bitmask_inv_cst_vector_p
> + expand_vec_cmp_expr_p)
>
> /* Operator lists. */
> (define_operator_list tcc_comparison
> @@ -832,6 +833,38 @@ DEFINE_INT_AND_FLOAT_ROUND_FN (RINT)
> { tree utype = unsigned_type_for (type); }
> (convert (rshift (lshift (convert:utype @0) @2) @3))))))
>
> +/* Fold (-x >> C) into -(x > 0) where C = precision(type) - 1. */
> +(for cst (INTEGER_CST VECTOR_CST)
> + (simplify
> + (rshift (negate:s @0) cst@1)
> + (if (!TYPE_UNSIGNED (type)
> + && TYPE_OVERFLOW_UNDEFINED (type))
> + (with { tree stype = TREE_TYPE (@1);
> + tree bt = truth_type_for (type);
> + tree zeros = build_zero_cst (type); }
> + (switch
> + /* Handle scalar case. */
> + (if (INTEGRAL_TYPE_P (type)
> + /* If we apply the rule to the scalar type before vectorization
> + we will enforce the result of the comparison being a bool
> + which will require an extra AND on the result that will be
> + indistinguishable from when the user did actually want 0
> + or 1 as the result so it can't be removed. */
> + && canonicalize_math_after_vectorization_p ()
> + && wi::eq_p (wi::to_wide (@1), TYPE_PRECISION (type) - 1))
> + (negate (convert (gt @0 { zeros; }))))
> + /* Handle vector case. */
> + (if (VECTOR_INTEGER_TYPE_P (type)
> + /* First check whether the target has the same mode for vector
> + comparison results as it's operands do. */
> + && TYPE_MODE (bt) == TYPE_MODE (type)
> + /* Then check to see if the target is able to expand the comparison
> + with the given type later on, otherwise we may ICE. */
> + && expand_vec_cmp_expr_p (type, bt, { GT_EXPR }))
No need to wrap GT_EXPR in { }
> + (with { tree cst = uniform_integer_cst_p (@1); }
if you declare 'cst' above where you declare 'bt' you can do
&& (cst = uniform_integer_cst_p (@1)))
combining it with the if above, and the one below, simplifying indents
and flow.
OK with that change.
I guess it might happen that the scalar transform expands badly
on some targets? Please have an eye on problems that come up.
Thanks,
Richard.
> + (if (cst && wi::eq_p (wi::to_wide (cst), element_precision (type) - 1))
> + (view_convert (gt:bt @0 { zeros; }))))))))))
> +
> /* Fold (C1/X)*C2 into (C1*C2)/X. */
> (simplify
> (mult (rdiv@3 REAL_CST@0 @1) REAL_CST@2)
> diff --git a/gcc/testsuite/gcc.dg/signbit-2.c b/gcc/testsuite/gcc.dg/signbit-2.c
> new file mode 100644
> index 0000000000000000000000000000000000000000..fc0157cbc5c7996b481f2998bc30176c96a669bb
> --- /dev/null
> +++ b/gcc/testsuite/gcc.dg/signbit-2.c
> @@ -0,0 +1,19 @@
> +/* { dg-do assemble } */
> +/* { dg-options "-O3 --save-temps -fdump-tree-optimized" } */
> +
> +#include <stdint.h>
> +
> +void fun1(int32_t *x, int n)
> +{
> + for (int i = 0; i < (n & -16); i++)
> + x[i] = (-x[i]) >> 31;
> +}
> +
> +void fun2(int32_t *x, int n)
> +{
> + for (int i = 0; i < (n & -16); i++)
> + x[i] = (-x[i]) >> 30;
> +}
> +
> +/* { dg-final { scan-tree-dump-times {\s+>\s+\{ 0, 0, 0, 0 \}} 1 optimized } } */
> +/* { dg-final { scan-tree-dump-not {\s+>>\s+31} optimized } } */
> diff --git a/gcc/testsuite/gcc.dg/signbit-3.c b/gcc/testsuite/gcc.dg/signbit-3.c
> new file mode 100644
> index 0000000000000000000000000000000000000000..19e9c06c349b3287610f817628f00938ece60bf7
> --- /dev/null
> +++ b/gcc/testsuite/gcc.dg/signbit-3.c
> @@ -0,0 +1,13 @@
> +/* { dg-do assemble } */
> +/* { dg-options "-O1 --save-temps -fdump-tree-optimized" } */
> +
> +#include <stdint.h>
> +
> +void fun1(int32_t *x, int n)
> +{
> + for (int i = 0; i < (n & -16); i++)
> + x[i] = (-x[i]) >> 31;
> +}
> +
> +/* { dg-final { scan-tree-dump-times {\s+>\s+0;} 1 optimized } } */
> +/* { dg-final { scan-tree-dump-not {\s+>>\s+31} optimized } } */
> diff --git a/gcc/testsuite/gcc.dg/signbit-4.c b/gcc/testsuite/gcc.dg/signbit-4.c
> new file mode 100644
> index 0000000000000000000000000000000000000000..bc459ba60a760bdf49e94dbec762f378c24fe9b5
> --- /dev/null
> +++ b/gcc/testsuite/gcc.dg/signbit-4.c
> @@ -0,0 +1,65 @@
> +/* { dg-do run } */
> +/* { dg-options "-O1 -fwrapv" } */
> +
> +#include <stdint.h>
> +#include <limits.h>
> +#include <stdio.h>
> +
> +#ifndef N
> +#define N 65
> +#endif
> +
> +#ifndef TYPE
> +#define TYPE int32_t
> +#endif
> +
> +#ifndef DEBUG
> +#define DEBUG 1
> +#endif
> +
> +#define BASE ((TYPE) -1 < 0 ? -126 : 4)
> +
> +__attribute__ ((noinline, noipa))
> +void fun1(TYPE *x, int n)
> +{
> + for (int i = 0; i < n; i++)
> + x[i] = (-x[i]) >> 31;
> +}
> +
> +__attribute__ ((noinline, noipa, optimize("O0")))
> +void fun2(TYPE *x, int n)
> +{
> + for (int i = 0; i < n; i++)
> + x[i] = (-x[i]) >> 31;
> +}
> +
> +int main ()
> +{
> + TYPE a[N];
> + TYPE b[N];
> +
> + a[0] = INT_MIN;
> + b[0] = INT_MIN;
> +
> + for (int i = 1; i < N; ++i)
> + {
> + a[i] = BASE + i * 13;
> + b[i] = BASE + i * 13;
> + if (DEBUG)
> + printf ("%d: 0x%x\n", i, a[i]);
> + }
> +
> + fun1 (a, N);
> + fun2 (b, N);
> +
> + for (int i = 0; i < N; ++i)
> + {
> + if (DEBUG)
> + printf ("%d = 0x%x == 0x%x\n", i, a[i], b[i]);
> +
> + if (a[i] != b[i])
> + __builtin_abort ();
> + }
> + return 0;
> +}
> +
> diff --git a/gcc/testsuite/gcc.dg/signbit-5.c b/gcc/testsuite/gcc.dg/signbit-5.c
> new file mode 100644
> index 0000000000000000000000000000000000000000..22a92704773e3282759524b74d35196a477d43dd
> --- /dev/null
> +++ b/gcc/testsuite/gcc.dg/signbit-5.c
> @@ -0,0 +1,65 @@
> +/* { dg-do run } */
> +/* { dg-options "-O3" } */
> +
> +#include <stdint.h>
> +#include <limits.h>
> +#include <stdio.h>
> +
> +#ifndef N
> +#define N 65
> +#endif
> +
> +#ifndef TYPE
> +#define TYPE int32_t
> +#endif
> +
> +#ifndef DEBUG
> +#define DEBUG 1
> +#endif
> +
> +#define BASE ((TYPE) -1 < 0 ? -126 : 4)
> +
> +__attribute__ ((noinline, noipa))
> +void fun1(TYPE *x, int n)
> +{
> + for (int i = 0; i < n; i++)
> + x[i] = (-x[i]) >> 31;
> +}
> +
> +__attribute__ ((noinline, noipa, optimize("O1")))
> +void fun2(TYPE *x, int n)
> +{
> + for (int i = 0; i < n; i++)
> + x[i] = (-x[i]) >> 31;
> +}
> +
> +int main ()
> +{
> + TYPE a[N];
> + TYPE b[N];
> +
> + a[0] = INT_MIN;
> + b[0] = INT_MIN;
> +
> + for (int i = 1; i < N; ++i)
> + {
> + a[i] = BASE + i * 13;
> + b[i] = BASE + i * 13;
> + if (DEBUG)
> + printf ("%d: 0x%x\n", i, a[i]);
> + }
> +
> + fun1 (a, N);
> + fun2 (b, N);
> +
> + for (int i = 0; i < N; ++i)
> + {
> + if (DEBUG)
> + printf ("%d = 0x%x == 0x%x\n", i, a[i], b[i]);
> +
> + if (a[i] != b[i])
> + __builtin_abort ();
> + }
> + return 0;
> +}
> +
> diff --git a/gcc/testsuite/gcc.dg/signbit-6.c b/gcc/testsuite/gcc.dg/signbit-6.c
> new file mode 100644
> index 0000000000000000000000000000000000000000..da186624cfa057dfc3780c8af4f6b1335ba07e7e
> --- /dev/null
> +++ b/gcc/testsuite/gcc.dg/signbit-6.c
> @@ -0,0 +1,72 @@
> +/* { dg-do run } */
> +/* { dg-options "-O1" } */
> +
> +#include <stdint.h>
> +#include <limits.h>
> +#include <stdio.h>
> +
> +#ifndef N
> +#define N 65
> +#endif
> +
> +#ifndef TYPE
> +#define TYPE int32_t
> +#endif
> +
> +#ifndef DEBUG
> +#define DEBUG 1
> +#endif
> +
> +#define BASE ((TYPE) -1 < 0 ? -126 : 4)
> +
> +__attribute__ ((noinline, noipa))
> +void fun1(TYPE *x, int n)
> +{
> + for (int i = 0; i < n; i++)
> + x[i] = (-x[i]) >> 31;
> +}
> +
> +__attribute__ ((noinline, noipa, optimize("O0")))
> +void fun2(TYPE *x, int n)
> +{
> + for (int i = 0; i < n; i++)
> + x[i] = (-x[i]) >> 31;
> +}
> +
> +int main ()
> +{
> + TYPE a[N];
> + TYPE b[N];
> +
> + a[0] = INT_MIN;
> + b[0] = INT_MIN;
> +
> + for (int i = 1; i < N; ++i)
> + {
> + a[i] = BASE + i * 13;
> + b[i] = BASE + i * 13;
> + if (DEBUG)
> + printf ("%d: 0x%x\n", i, a[i]);
> + }
> +
> + fun1 (a, N);
> + fun2 (b, N);
> +
> + if (DEBUG)
> + printf ("%d = 0x%x == 0x%x\n", 0, a[0], b[0]);
> +
> + if (a[0] != 0x0 || b[0] != -1)
> + __builtin_abort ();
> +
> +
> + for (int i = 1; i < N; ++i)
> + {
> + if (DEBUG)
> + printf ("%d = 0x%x == 0x%x\n", i, a[i], b[i]);
> +
> + if (a[i] != b[i])
> + __builtin_abort ();
> + }
> + return 0;
> +}
> +
> diff --git a/gcc/testsuite/gcc.target/aarch64/signbit-1.c b/gcc/testsuite/gcc.target/aarch64/signbit-1.c
> new file mode 100644
> index 0000000000000000000000000000000000000000..3ebfb0586f37de29cf58635b27fe48503714447e
> --- /dev/null
> +++ b/gcc/testsuite/gcc.target/aarch64/signbit-1.c
> @@ -0,0 +1,18 @@
> +/* { dg-do assemble } */
> +/* { dg-options "-O3 --save-temps" } */
> +
> +#include <stdint.h>
> +
> +void fun1(int32_t *x, int n)
> +{
> + for (int i = 0; i < (n & -16); i++)
> + x[i] = (-x[i]) >> 31;
> +}
> +
> +void fun2(int32_t *x, int n)
> +{
> + for (int i = 0; i < (n & -16); i++)
> + x[i] = (-x[i]) >> 30;
> +}
> +
> +/* { dg-final { scan-assembler-times {\tcmgt\t} 1 } } */
>
--
Richard Biener <rguenther@suse.de>
SUSE Software Solutions Germany GmbH, Maxfeldstrasse 5, 90409 Nuernberg,
Germany; GF: Ivo Totev; HRB 36809 (AG Nuernberg)
prev parent reply other threads:[~2021-11-04 13:06 UTC|newest]
Thread overview: 17+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-10-05 12:50 Tamar Christina
2021-10-05 12:56 ` Richard Earnshaw
2021-10-05 13:30 ` Tamar Christina
2021-10-05 13:34 ` Richard Earnshaw
2021-10-05 13:49 ` Tamar Christina
2021-10-05 13:51 ` Richard Earnshaw
2021-10-05 13:56 ` Tamar Christina
2021-10-07 12:46 ` Richard Earnshaw
2021-10-11 11:36 ` Tamar Christina
2021-10-13 12:11 ` Richard Biener
2021-10-15 7:48 ` Tamar Christina
2021-10-15 9:06 ` Richard Biener
2021-10-15 10:36 ` Richard Earnshaw
2021-10-15 10:57 ` Richard Biener
2021-10-15 11:55 ` Tamar Christina
[not found] ` <VI1PR08MB5325D87574F2D09568EF5C40FF849@VI1PR08MB5325.eurprd08.prod.outlook.com>
[not found] ` <34p8433-751p-2n5s-qp50-r8rss490npop@fhfr.qr>
2021-11-03 13:21 ` Tamar Christina
2021-11-04 13:06 ` Richard Biener [this message]
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=q9171p92-snq4-qo8p-80n5-p855sq36qs82@fhfr.qr \
--to=rguenther@suse.de \
--cc=Richard.Earnshaw@foss.arm.com \
--cc=Tamar.Christina@arm.com \
--cc=gcc-patches@gcc.gnu.org \
--cc=nd@arm.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).