This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
Re: [PATCH] Reduce the maximum precision for exp and log
- From: "Joseph S. Myers" <joseph at codesourcery dot com>
- To: Siddhesh Poyarekar <siddhesh at redhat dot com>
- Cc: <libc-alpha at sourceware dot org>
- Date: Thu, 28 Feb 2013 17:08:18 +0000
- Subject: Re: [PATCH] Reduce the maximum precision for exp and log
- References: <20130228160427.GE2358@spoyarek.pnq.redhat.com>
On Thu, 28 Feb 2013, Siddhesh Poyarekar wrote:
> Hi,
>
> The worst case precision required for exp() and log() are known[1] to
> be 158 bits and 118 bits respectively. This patch uses this knowledge
> to limit the mp computations of both these functions to minimum
> precision larger than these values. No regressions were seen in the
> testsuite due to this. I could not find a slow path-invoking input
> for log, so I don't have measurements for it. For exp however, I have
> stats on x86_64:
Please give details of your error analysis for the maximum inaccuracy in
these implementations in glibc that shows that, together with the above
worst-case figures, the error in the glibc implementation cannot exceed
the distance from a half-way value.
--
Joseph S. Myers
joseph@codesourcery.com