sourceware.org Git - newlib-cygwin.git/commit

]> sourceware.org Git - newlib-cygwin.git/commit

git://sourceware.org / newlib-cygwin.git / commit

author	Szabolcs Nagy <szabolcs.nagy@arm.com>
	Tue, 26 Jun 2018 15:25:12 +0000 (16:25 +0100)
committer	Corinna Vinschen <corinna@vinschen.de>
	Wed, 27 Jun 2018 13:40:49 +0000 (15:40 +0200)
commit	b99d49e506f595c4be91072fe0169a405ae16007
tree	57ff07b7cd9110b22e64526ea5dc93ae734a0be5	tree
parent	07e2c328285e63c78d23ded96fd333afad35961d	commit \| diff

New pow implementation

The new implementation is provided under !__OBSOLETE_MATH, it uses
ISO C99 code. With default settings the worst case error in nearest
rounding mode is 0.54 ULP with inlined fma and fma contraction. It uses
a 4 KB lookup table in addition to the table in exp_data.c, on aarch64
.text+.rodata size of libm.a is increased by 2295 bytes.

Improvements on Cortex-A72:
latency: 3.3x
thruput: 4.9x

newlib/libm/common/Makefile.am		diff \| blob \| blame \| history
newlib/libm/common/Makefile.in		diff \| blob \| blame \| history
newlib/libm/common/fdlibm.h		diff \| blob \| blame \| history
newlib/libm/common/math_config.h		diff \| blob \| blame \| history
newlib/libm/common/pow.c	[new file with mode: 0644]	blob
newlib/libm/common/pow_log_data.c	[new file with mode: 0644]	blob
newlib/libm/math/e_pow.c		diff \| blob \| blame \| history
newlib/libm/math/w_pow.c		diff \| blob \| blame \| history

newlib and cygwin

RSS Atom

This page took 0.03217 seconds and 5 git commands to generate.