public inbox for libc-alpha@sourceware.org
 help / color / mirror / Atom feed
From: Adhemerval Zanella <adhemerval.zanella@linaro.org>
To: Paul Zimmermann <Paul.Zimmermann@inria.fr>
Cc: libc-alpha@sourceware.org
Subject: Re: Fix the inaccuracy of j0f/y0f/j1f/y1f
Date: Thu, 25 Feb 2021 08:24:12 -0300	[thread overview]
Message-ID: <ee2b8e3b-aaaa-214b-dc8a-8ac620abc843@linaro.org> (raw)
In-Reply-To: <mwzgzsfnml.fsf@salade.loria.fr>



On 25/02/2021 07:27, Paul Zimmermann wrote:
>        Dear Adhemerval,
> 
>> I have checked your set on different architectures [...]
> 
> thank you, I can reproduce the failures on the GCC compile farm.
> So far it is just due to some ranges around zeroes that are too
> narrow. I guess it is due to the fact that computations on x86_64
> are done using double extended precision. Btw is there a way to
> disable that when building glibc?

Afaik -mfpmath=sse is the default for x86_64, so I am not sure the
extension precision on x86_64 s the culprit here.

I noted that cosf shows a better precision on x86_64 compared to
other architectures (based on glibc own ulp files), so it might
be the origin of the difference.

  reply	other threads:[~2021-02-25 11:24 UTC|newest]

Thread overview: 10+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-02-19  9:02 Paul Zimmermann
2021-02-19 16:44 ` Michael Morrell
2021-02-19 21:15   ` Joseph Myers
2021-02-21  7:37   ` Paul Zimmermann
2021-02-23 19:07     ` Michael Morrell
2021-02-23 19:34       ` Joseph Myers
2021-02-24 18:59 ` Adhemerval Zanella
2021-02-25 10:27   ` Paul Zimmermann
2021-02-25 11:24     ` Adhemerval Zanella [this message]
2021-02-26  6:40       ` 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=ee2b8e3b-aaaa-214b-dc8a-8ac620abc843@linaro.org \
    --to=adhemerval.zanella@linaro.org \
    --cc=Paul.Zimmermann@inria.fr \
    --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).