From: Noah Goldstein <goldstein.w.n@gmail.com>
To: "H.J. Lu" <hjl.tools@gmail.com>
Cc: libc-alpha@sourceware.org
Subject: Re: [PATCH] x86_64: Add log2 with FMA
Date: Thu, 10 Aug 2023 16:39:18 -0500 [thread overview]
Message-ID: <CAFUsyfK1MHFd6tpQ5DUYJtYe7nUV=SdNezQdx_BDPANiKG0J9g@mail.gmail.com> (raw)
In-Reply-To: <20230810182430.6859-1-hjl.tools@gmail.com>
On Thu, Aug 10, 2023 at 1:24 PM H.J. Lu via Libc-alpha
<libc-alpha@sourceware.org> wrote:
>
> On Skylake, it improves log2 bench performance by:
>
> Before After Improvement
> max 208.779 63.827 69%
> min 9.977 6.55 34%
> mean 10.366 6.8191 34%
> ---
> sysdeps/x86_64/fpu/multiarch/Makefile | 2 ++
> sysdeps/x86_64/fpu/multiarch/e_log2-fma.c | 3 ++
> sysdeps/x86_64/fpu/multiarch/e_log2.c | 43 +++++++++++++++++++++++
> 3 files changed, 48 insertions(+)
> create mode 100644 sysdeps/x86_64/fpu/multiarch/e_log2-fma.c
> create mode 100644 sysdeps/x86_64/fpu/multiarch/e_log2.c
>
> diff --git a/sysdeps/x86_64/fpu/multiarch/Makefile b/sysdeps/x86_64/fpu/multiarch/Makefile
> index e37e488c37..f773255721 100644
> --- a/sysdeps/x86_64/fpu/multiarch/Makefile
> +++ b/sysdeps/x86_64/fpu/multiarch/Makefile
> @@ -34,6 +34,7 @@ libm-sysdep_routines += \
> e_atan2-fma \
> e_exp-fma \
> e_log-fma \
> + e_log2-fma \
> e_pow-fma \
> s_atan-fma \
> s_sin-fma \
> @@ -45,6 +46,7 @@ CFLAGS-e_asin-fma.c = -mfma -mavx2
> CFLAGS-e_atan2-fma.c = -mfma -mavx2
> CFLAGS-e_exp-fma.c = -mfma -mavx2
> CFLAGS-e_log-fma.c = -mfma -mavx2
> +CFLAGS-e_log2-fma.c = -mfma -mavx2
> CFLAGS-e_pow-fma.c = -mfma -mavx2
> CFLAGS-s_atan-fma.c = -mfma -mavx2
> CFLAGS-s_sin-fma.c = -mfma -mavx2
> diff --git a/sysdeps/x86_64/fpu/multiarch/e_log2-fma.c b/sysdeps/x86_64/fpu/multiarch/e_log2-fma.c
> new file mode 100644
> index 0000000000..9fbebc1b47
> --- /dev/null
> +++ b/sysdeps/x86_64/fpu/multiarch/e_log2-fma.c
> @@ -0,0 +1,3 @@
> +#define __log2 __log2_fma
> +
> +#include <sysdeps/ieee754/dbl-64/e_log2.c>
> diff --git a/sysdeps/x86_64/fpu/multiarch/e_log2.c b/sysdeps/x86_64/fpu/multiarch/e_log2.c
> new file mode 100644
> index 0000000000..c0320caf36
> --- /dev/null
> +++ b/sysdeps/x86_64/fpu/multiarch/e_log2.c
> @@ -0,0 +1,43 @@
> +/* Multiple versions of log2.
> + Copyright (C) 2023 Free Software Foundation, Inc.
> + This file is part of the GNU C Library.
> +
> + The GNU C Library is free software; you can redistribute it and/or
> + modify it under the terms of the GNU Lesser General Public
> + License as published by the Free Software Foundation; either
> + version 2.1 of the License, or (at your option) any later version.
> +
> + The GNU C Library is distributed in the hope that it will be useful,
> + but WITHOUT ANY WARRANTY; without even the implied warranty of
> + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
> + Lesser General Public License for more details.
> +
> + You should have received a copy of the GNU Lesser General Public
> + License along with the GNU C Library; if not, see
> + <https://www.gnu.org/licenses/>. */
> +
> +#include <libm-alias-double.h>
> +#include <libm-alias-finite.h>
> +
> +extern double __redirect_log2 (double);
> +
> +#define SYMBOL_NAME log2
> +#include "ifunc-fma.h"
> +
> +libc_ifunc_redirected (__redirect_log2, __log2, IFUNC_SELECTOR ());
> +
> +#ifdef SHARED
> +__hidden_ver1 (__log2, __GI___log2, __redirect_log2)
> + __attribute__ ((visibility ("hidden")));
> +
> +versioned_symbol (libm, __ieee754_log2, log2, GLIBC_2_29);
> +libm_alias_double_other (__log2, log2)
> +#else
> +libm_alias_double (__log2, log2)
> +#endif
> +
> +strong_alias (__log2, __ieee754_log2)
> +libm_alias_finite (__log2, __log2)
> +
> +#define __log2 __log2_sse2
> +#include <sysdeps/ieee754/dbl-64/e_log2.c>
> --
> 2.41.0
>
LGTM
next prev parent reply other threads:[~2023-08-10 21:39 UTC|newest]
Thread overview: 3+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-08-10 18:24 H.J. Lu
2023-08-10 21:39 ` Noah Goldstein [this message]
2023-08-11 9:25 ` Paul Zimmermann
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='CAFUsyfK1MHFd6tpQ5DUYJtYe7nUV=SdNezQdx_BDPANiKG0J9g@mail.gmail.com' \
--to=goldstein.w.n@gmail.com \
--cc=hjl.tools@gmail.com \
--cc=libc-alpha@sourceware.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).