From: Ken Matsui <kmatsui@cs.washington.edu>
To: Jonathan Wakely <jwakely@redhat.com>
Cc: "François Dumont" <frs.dumont@gmail.com>,
"Ken Matsui" <kmatsui@gcc.gnu.org>,
gcc-patches@gcc.gnu.org, libstdc++@gcc.gnu.org
Subject: Re: [PATCH v2 3/3] libstdc++: Optimize is_fundamental performance by __is_arithmetic built-in
Date: Thu, 31 Aug 2023 06:47:27 -0700 [thread overview]
Message-ID: <CAML+3pXV9DYkKORuYGkBnDYW1yY+f6t_2sOgB+MOHVZ588OgKQ@mail.gmail.com> (raw)
In-Reply-To: <CACb0b4mvWvHyrW2F6HRbTLt=UeK6Koxa1hwfKrCk3R-mgdmueQ@mail.gmail.com>
On Tue, Aug 8, 2023 at 1:14 PM Jonathan Wakely <jwakely@redhat.com> wrote:
>
>
>
> On Tue, 18 Jul 2023 at 07:28, Ken Matsui via Libstdc++ <libstdc++@gcc.gnu.org> wrote:
>>
>> I will eventually work on disjunction to somehow optimize, but in the
>> meantime, this might be a better implementation. Of course, my
>> benchmark could be wrong.
>
>
> You should use __or_ internally in libstdc++ code, not std::disjunction.
>
> Patrick already optimized both of those, and __or_ is slightly faster (because it doesn't have to conform to the full requirements of std::disjunction).
>
I will! Thank you!
> A compiler built-in for __or_ / __disjunction might perform better. But eventually if we're going to have built-ins for all of __is_arithmetic, __is_void, and __is_null_pointer, then we would want simply:
>
> __is_arithmetic(T) || __is_void(T) || __is_null_pointer(T)
>
> and so we wouldn't need to avoid instantiating any class templates at all.
>
I think we are not going to define built-ins for is_void and
is_null_pointer as their type trait implementations are already
optimal.
>
>>
>>
>> On Mon, Jul 17, 2023 at 11:24 PM Ken Matsui <kmatsui@cs.washington.edu> wrote:
>> >
>> > Hi,
>> >
>> > I took a benchmark for this.
>> >
>> > https://github.com/ken-matsui/gcc-benches/blob/main/is_fundamental-disjunction.md#mon-jul-17-105937-pm-pdt-2023
>> >
>> > template<typename _Tp>
>> > struct is_fundamental
>> > : public std::bool_constant<__is_arithmetic(_Tp)
>> > || std::is_void<_Tp>::value
>> > || std::is_null_pointer<_Tp>::value>
>> > { };
>> >
>> > is faster than:
>> >
>> > template<typename _Tp>
>> > struct is_fundamental
>> > : public std::bool_constant<__is_arithmetic(_Tp)
>> > || std::disjunction<std::is_void<_Tp>,
>> > std::is_null_pointer<_Tp>
>> > >::value>
>> > { };
>> >
>> > Time: -32.2871%
>> > Peak Memory: -18.5071%
>> > Total Memory: -20.1991%
>> >
>> > Sincerely,
>> > Ken Matsui
>> >
>> > On Sun, Jul 16, 2023 at 9:49 PM Ken Matsui <kmatsui@cs.washington.edu> wrote:
>> > >
>> > > On Sun, Jul 16, 2023 at 5:41 AM François Dumont <frs.dumont@gmail.com> wrote:
>> > > >
>> > > >
>> > > > On 15/07/2023 06:55, Ken Matsui via Libstdc++ wrote:
>> > > > > This patch optimizes the performance of the is_fundamental trait by
>> > > > > dispatching to the new __is_arithmetic built-in trait.
>> > > > >
>> > > > > libstdc++-v3/ChangeLog:
>> > > > >
>> > > > > * include/std/type_traits (is_fundamental_v): Use __is_arithmetic
>> > > > > built-in trait.
>> > > > > (is_fundamental): Likewise. Optimize the original implementation.
>> > > > >
>> > > > > Signed-off-by: Ken Matsui <kmatsui@gcc.gnu.org>
>> > > > > ---
>> > > > > libstdc++-v3/include/std/type_traits | 21 +++++++++++++++++----
>> > > > > 1 file changed, 17 insertions(+), 4 deletions(-)
>> > > > >
>> > > > > diff --git a/libstdc++-v3/include/std/type_traits b/libstdc++-v3/include/std/type_traits
>> > > > > index 7ebbe04c77b..cf24de2fcac 100644
>> > > > > --- a/libstdc++-v3/include/std/type_traits
>> > > > > +++ b/libstdc++-v3/include/std/type_traits
>> > > > > @@ -668,11 +668,21 @@ _GLIBCXX_BEGIN_NAMESPACE_VERSION
>> > > > > #endif
>> > > > >
>> > > > > /// is_fundamental
>> > > > > +#if __has_builtin(__is_arithmetic)
>> > > > > + template<typename _Tp>
>> > > > > + struct is_fundamental
>> > > > > + : public __bool_constant<__is_arithmetic(_Tp)
>> > > > > + || is_void<_Tp>::value
>> > > > > + || is_null_pointer<_Tp>::value>
>> > > > > + { };
>> > > >
>> > > > What about doing this ?
>> > > >
>> > > > template<typename _Tp>
>> > > > struct is_fundamental
>> > > > : public __bool_constant<__is_arithmetic(_Tp)
>> > > > || __or_<is_void<_Tp>,
>> > > > is_null_pointer<_Tp>>::value>
>> > > > { };
>> > > >
>> > > > Based on your benches it seems that builtin __is_arithmetic is much better that std::is_arithmetic. But __or_ could still avoid instantiation of is_null_pointer.
>> > > >
>> > > Let me take a benchmark for this later.
>>
next prev parent reply other threads:[~2023-08-31 13:47 UTC|newest]
Thread overview: 41+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-07-09 12:57 [PATCH 1/2] c++, libstdc++: implement __is_arithmetic built-in trait Ken Matsui
2023-07-09 12:57 ` [PATCH 2/2] libstdc++: use new built-in trait __is_arithmetic Ken Matsui
2023-07-09 13:03 ` [PATCH 1/2] c++, libstdc++: implement __is_arithmetic built-in trait Ken Matsui
2023-07-15 4:55 ` [PATCH v2 1/3] c++, libstdc++: Implement " Ken Matsui
2023-07-15 4:55 ` [PATCH v2 2/3] libstdc++: Optimize is_arithmetic performance by __is_arithmetic built-in Ken Matsui
2023-07-16 12:32 ` François Dumont
2023-07-17 4:48 ` Ken Matsui
2023-07-22 13:39 ` François Dumont
2023-08-08 20:09 ` Jonathan Wakely
2023-08-08 20:28 ` Jonathan Wakely
2023-07-15 4:55 ` [PATCH v2 3/3] libstdc++: Optimize is_fundamental " Ken Matsui
2023-07-15 5:07 ` Ken Matsui
2023-07-16 12:41 ` François Dumont
2023-07-17 4:49 ` Ken Matsui
2023-07-18 6:24 ` Ken Matsui
2023-07-18 6:26 ` Ken Matsui
2023-08-08 20:14 ` Jonathan Wakely
2023-08-31 13:47 ` Ken Matsui [this message]
2023-07-22 13:45 ` François Dumont
2023-08-08 20:18 ` Jonathan Wakely
2023-08-31 13:57 ` Ken Matsui
2023-08-31 15:01 ` Ken Matsui
2023-07-16 12:27 ` [PATCH v2 1/3] c++, libstdc++: Implement __is_arithmetic built-in trait François Dumont
2023-07-17 4:46 ` Ken Matsui
2023-07-18 6:27 ` [PATCH v3 " Ken Matsui
2023-07-18 6:27 ` [PATCH v3 2/3] libstdc++: Optimize is_arithmetic performance by __is_arithmetic built-in Ken Matsui
2023-07-18 6:27 ` [PATCH v3 3/3] libstdc++: Optimize is_fundamental " Ken Matsui
2023-07-18 7:40 ` [PATCH v4 1/4] c++, libstdc++: Implement __is_arithmetic built-in trait Ken Matsui
2023-07-18 7:40 ` [PATCH v4 2/4] libstdc++: Optimize is_arithmetic trait performance Ken Matsui
2023-07-18 7:40 ` [PATCH v4 3/4] libstdc++: Optimize is_fundamental " Ken Matsui
2023-07-18 7:40 ` [PATCH v4 4/4] libstdc++: Optimize is_compound " Ken Matsui
2023-08-08 20:33 ` Jonathan Wakely
2023-09-01 10:40 ` Ken Matsui
2023-07-22 13:50 ` [PATCH v3 1/3] c++, libstdc++: Implement __is_arithmetic built-in trait François Dumont
2023-08-08 20:04 ` Jonathan Wakely
2023-09-01 11:25 ` [PATCH v5 1/4] " Ken Matsui
2023-09-01 11:25 ` [PATCH v5 2/4] libstdc++: Optimize is_arithmetic trait performance Ken Matsui
2023-09-01 11:25 ` [PATCH v5 3/4] libstdc++: Optimize is_fundamental " Ken Matsui
2023-09-01 11:25 ` [PATCH v5 4/4] libstdc++: Optimize is_compound " Ken Matsui
2023-09-04 10:27 ` [PATCH v5 1/4] c++, libstdc++: Implement __is_arithmetic built-in trait Jonathan Wakely
2023-09-04 15:00 ` Ken Matsui
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=CAML+3pXV9DYkKORuYGkBnDYW1yY+f6t_2sOgB+MOHVZ588OgKQ@mail.gmail.com \
--to=kmatsui@cs.washington.edu \
--cc=frs.dumont@gmail.com \
--cc=gcc-patches@gcc.gnu.org \
--cc=jwakely@redhat.com \
--cc=kmatsui@gcc.gnu.org \
--cc=libstdc++@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).