This is the mail archive of the libc-alpha@sourceware.org mailing list for the glibc project.

Index Nav:	[Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav:	[Date Prev] [Date Next]	[Thread Prev] [Thread Next]
Other format:	[Raw text]

Re: [PATCH] Benchtests for sinf, cosf and sincosf

From: Siddhesh Poyarekar <siddhesh at gotplt dot org>
To: Joseph Myers <joseph at codesourcery dot com>
Cc: Rajalakshmi Srinivasaraghavan <raji at linux dot vnet dot ibm dot com>, libc-alpha at sourceware dot org
Date: Thu, 12 Oct 2017 07:27:02 +0530
Subject: Re: [PATCH] Benchtests for sinf, cosf and sincosf
Authentication-results: sourceware.org; auth=none
References: <1507651615-2814-1-git-send-email-raji@linux.vnet.ibm.com> <487c8841-0c32-ac4d-f6c5-1b5e1de3dd78@gotplt.org> <alpine.DEB.2.20.1710111302310.4866@digraph.polyomino.org.uk>

On Wednesday 11 October 2017 06:35 PM, Joseph Myers wrote:
> I don't think that's appropriate for sharing between different types.  The 
> inputs for sin include many outside the range of float (which probably 
> illustrates how unrepresentative they are of typical usage of sin), and we 
> don't want the float test to have lots of infinities tested.

OK, as for larger values, they were there to exercise the range
reduction paths, which can be done by smaller values as well so we could
just drop those inputs.

> Sharing inputs between sinf, cosf and sincosf may make much more sense.

Agreed.

Siddhesh

Follow-Ups:
- Re: [PATCH] Benchtests for sinf, cosf and sincosf
  - From: Rajalakshmi Srinivasaraghavan

References:
- [PATCH] Benchtests for sinf, cosf and sincosf
  - From: Rajalakshmi Srinivasaraghavan
- Re: [PATCH] Benchtests for sinf, cosf and sincosf
  - From: Siddhesh Poyarekar
- Re: [PATCH] Benchtests for sinf, cosf and sincosf
  - From: Joseph Myers

Index Nav:	[Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav:	[Date Prev] [Date Next]	[Thread Prev] [Thread Next]