From: Qing Zhao <qing.zhao@oracle.com>
To: Andrew Pinski <pinskia@gmail.com>
Cc: "richard.earnshaw@arm.com" <richard.earnshaw@arm.com>,
gcc Patches <gcc-patches@gcc.gnu.org>,
Richard Sandiford <richard.sandiford@arm.com>
Subject: Re: Should -ffp-contract=off the default on GCC?
Date: Thu, 16 Mar 2023 16:38:41 +0000 [thread overview]
Message-ID: <A886C65B-073E-4015-958E-33F08E50BB70@oracle.com> (raw)
In-Reply-To: <CA+=Sn1kEO8JmKy8a0njfjkQn68C9ds+GasjncwGRGQ=Nh2aMjA@mail.gmail.com>
> On Mar 16, 2023, at 12:31 PM, Andrew Pinski <pinskia@gmail.com> wrote:
>
> On Thu, Mar 16, 2023 at 9:25 AM Qing Zhao via Gcc-patches
> <gcc-patches@gcc.gnu.org> wrote:
>>
>> Hi,
>>
>> Recently, we discovered some floating point precision diffs when using GCC8 to build our
>> application on arm64: After some investigation, it turns out that this is due to the
>> -ffp-contract=fast option that is on by default. Therefore, we have to explicitly add
>> -ffp-contract=off and do a full-rebuild.
>>
>> GCC by default turns -ffp-contract=fast on.
>> https://gcc.gnu.org/onlinedocs/gcc-8.5.0/gcc/Optimize-Options.html#Optimize-Options
>> https://gcc.gnu.org/onlinedocs/gcc-12.2.0/gcc/Optimize-Options.html#Optimize-Options
>>
>> "
>> -ffp-contract=style
>> -ffp-contract=off disables floating-point expression contraction. -ffp-contract=fast enables
>> floating-point expression contraction such as forming of fused multiply-add operations if
>> the target has native support for them. -ffp-contract=on enables floating-point expression
>> contraction if allowed by the language standard. This is currently not implemented and
>> treated equal to -ffp-contract=off.
>>
>> The default is -ffp-contract=fast.
>> "
>>
>> This can be shown by a small example for arm64 with gcc8.5 in https://godbolt.org/z/MxYfnG8TE.
>> Only when adding -std=c89 explicitly, this transformaton is off.
>>
>> another exmaple also shows that Clang and MSVC only allow this transformation when speifiying
>> ffast-math and fp:fast: https://godbolt.org/z/o54bYfPbP
>>
>> When searching online, we found that there were similar discussions recently on the exact same issue:
>> https://github.com/dotnet/runtime/issues/64604
>> https://github.com/dotnet/runtime/issues/64591
>>
>> a summary of these discussions is:
>>
>> 1. "fmadd" is a fused operation and will return a different result for many inputs;
>> 2. therefore, -ffp-contract=fast is not a safe optimization to be on by default;
>> 3. Clang and MSVC only allow this when specifying ffast-math and fp:fast since this is not an
>> IEEE754 compliant optimization;
>> 4. The reasons why GCC turns on this option by default are:
>> A. GNU C language spec allows such transformation.
>> B. this did not expose real problem for most X86/X64 apps previously since FMA instructions
>> didn't exist until 2013 when the FMA3 instruction set was added, and also these instructions
>> were not always available..
>> 5. Arm64 has fused multiply-add instructions as "baseline" and are always available. therefore
>> -ffp-contract=fast exposed more serious problems on Arm64 platforms.
>
> This summary ignores x87 and even ignores PowerPC in GCC having FMA
> even before clang/LLVM was around.
Okay.
>
>>
>> our major question:
>>
>> Should GCC turn off -ffp-contract=fast by default since it's not IEEE754 compliant and more
>> modern processors have the FMA instructions available by default?
>
>
> NO. We have this debate every few years and such.
So, what’s the major reason we keep the default that is not IEEE754 compliant from the beginning?
thanks.
Qing
>
> Thanks,
> Andrew Pinski
>
>>
>> Thanks.
>>
>> Qing
next prev parent reply other threads:[~2023-03-16 16:38 UTC|newest]
Thread overview: 32+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-03-16 16:24 Qing Zhao
2023-03-16 16:31 ` Andrew Pinski
2023-03-16 16:38 ` Qing Zhao [this message]
2023-03-16 16:53 ` Jakub Jelinek
2023-03-16 18:40 ` Qing Zhao
2023-03-20 22:05 ` Qing Zhao
2023-03-20 22:25 ` Jakub Jelinek
2023-03-21 12:49 ` Qing Zhao
2023-03-21 15:01 ` Qing Zhao
2023-03-21 16:56 ` Paul Koning
2023-03-21 17:00 ` Qing Zhao
2023-03-21 17:59 ` Jeff Law
2023-03-21 18:03 ` Paul Koning
2023-03-21 18:55 ` Toon Moene
2023-03-21 18:12 ` Alexander Monakov
2023-03-21 18:18 ` Jeff Law
2023-03-22 10:13 ` Richard Biener
2023-03-22 12:26 ` Alexander Monakov
2023-03-22 13:57 ` Richard Biener
2023-03-22 15:52 ` Qing Zhao
2023-03-24 7:12 ` Fangrui Song
2023-03-24 19:42 ` Andrew Pinski
2023-03-31 16:31 ` Qing Zhao
2023-03-21 19:01 ` Qing Zhao
2023-03-21 19:12 ` Jakub Jelinek
2023-03-21 23:28 ` Jeff Law
2023-03-21 23:31 ` Jakub Jelinek
2023-03-21 19:51 ` Jeff Law
2023-03-21 21:08 ` Qing Zhao
2023-03-22 12:33 ` Alexander Monakov
2023-03-22 14:33 ` Qing Zhao
2023-03-27 9:03 Zeson
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=A886C65B-073E-4015-958E-33F08E50BB70@oracle.com \
--to=qing.zhao@oracle.com \
--cc=gcc-patches@gcc.gnu.org \
--cc=pinskia@gmail.com \
--cc=richard.earnshaw@arm.com \
--cc=richard.sandiford@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).