public inbox for libc-alpha@sourceware.org
 help / color / mirror / Atom feed
From: Wilco Dijkstra <Wilco.Dijkstra@arm.com>
To: Adhemerval Zanella <adhemerval.zanella@linaro.org>
Cc: 'GNU C Library' <libc-alpha@sourceware.org>
Subject: [PATCH v2 3/9] math: Simplify hypotf implementation
Date: Thu, 28 Oct 2021 14:27:09 +0000	[thread overview]
Message-ID: <VE1PR08MB559962EBD2B3517D4C54515783869@VE1PR08MB5599.eurprd08.prod.outlook.com> (raw)

Hi Adhemerval,

+  if ((isinf (x) || isinf (y))
+      && !issignaling (x) && !issignaling (y))
+    return INFINITY;
+  if (isnan (x) || isnan (y))
+    return x + y;

These checks are inefficient. It's partly because GCC uses FP comparisons for
isinf/isnan when integer arithmetic is faster on almost all targets and more
correct for signaling NaNs. However GCC doesn't seem to treat these as
unlikely and so there are 3 comparisons and 3 taken branches before entering
the main code. If you change it to:

if (!isfinite (x) || !isfinite (y))
{
  if ((isinf (x) || isinf (y))
      && !issignaling (x) && !issignaling (y))
    return INFINITY;
 return x + y;
}

then it uses only 2 comparisons and we fallthrough into the common case.
This gives a speedup of 10%.

Wilco

             reply	other threads:[~2021-10-28 14:27 UTC|newest]

Thread overview: 3+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-10-28 14:27 Wilco Dijkstra [this message]
  -- strict thread matches above, loose matches on Subject: below --
2021-10-25 11:57 [PATCH v2 0/9] Improve hypot() Adhemerval Zanella
2021-10-25 11:57 ` [PATCH v2 3/9] math: Simplify hypotf implementation Adhemerval Zanella
2021-10-22 20:48 [PATCH v2 0/9] Improve hypot() Adhemerval Zanella
2021-10-22 20:48 ` [PATCH v2 3/9] math: Simplify hypotf implementation Adhemerval Zanella

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=VE1PR08MB559962EBD2B3517D4C54515783869@VE1PR08MB5599.eurprd08.prod.outlook.com \
    --to=wilco.dijkstra@arm.com \
    --cc=adhemerval.zanella@linaro.org \
    --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).