This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
Re: [PATCH 5/9] New generic log2f
- From: Szabolcs Nagy <szabolcs dot nagy at arm dot com>
- To: GNU C Library <libc-alpha at sourceware dot org>
- Cc: nd at arm dot com
- Date: Wed, 20 Sep 2017 14:20:31 +0100
- Subject: Re: [PATCH 5/9] New generic log2f
- Authentication-results: sourceware.org; auth=none
- Authentication-results: spf=none (sender IP is ) smtp.mailfrom=Szabolcs dot Nagy at arm dot com;
- Nodisclaimer: True
- References: <59C1123F.9080003@arm.com> <59C114D5.5090009@arm.com>
- Spamdiagnosticmetadata: NSPM
- Spamdiagnosticoutput: 1:99
On 19/09/17 14:00, Szabolcs Nagy wrote:
> Similar to the new logf: double precision arithmetics
> and a small lookup table is used. The argument reduction
> step is the same as for the new logf.
>
> with naive ubenchmark (on aarch64):
> reciprocal-throughput: 2.7x faster
> latency: 2x faster
with the committed log2f benchmark it's
reciprocal-throughput: 2.3x faster
latency: 2.1x faster
> old worst case error: 1.72 ulp
> new worst case error: 0.75 ulp
> aarch64 .text size: -252 bytes
> aarch64 .rodata size: +244 bytes
>
> 2017-09-19 Szabolcs Nagy <szabolcs.nagy@arm.com>
>
> * math/Makefile (type-float-routines): Add e_log2f_data.
> * sysdeps/ieee754/flt-32/e_log2f.c: New implementation.
> * sysdeps/ieee754/flt-32/e_log2f_data.c: New file.
> * sysdeps/ieee754/flt-32/math_config.h (__log2f_data): Define.
> (LOG2F_TABLE_BITS, LOG2F_POLY_ORDER): Define.
> * sysdeps/i386/fpu/e_log2f_data.c: New file.
> * sysdeps/ia64/fpu/e_log2f_data.c: New file.
> * sysdeps/m68k/m680x0/fpu/e_log2f_data.c: New file.
>