]> git.feebdaed.xyz Git - 0xmirror/glibc.git/commit
aarch64: Optimise AdvSIMD log10f
authorJames Chesterman <James.Chesterman@arm.com>
Wed, 19 Nov 2025 14:11:40 +0000 (14:11 +0000)
committerAdhemerval Zanella <adhemerval.zanella@linaro.org>
Thu, 4 Dec 2025 11:31:19 +0000 (08:31 -0300)
commit57215df30e4978c6eb3b6b78dee8230ff575acbb
treed003c542f55aa3c288e3fff506b482d48845696e
parentfe83660a7eaf29144c517fff4950030056b78078
aarch64: Optimise AdvSIMD log10f

Optimise AdvSIMD log10f by vectorising the special case.
Use scaling technique on subnormal values, then check for inf and
nan values.
The scaling technique will sqrt the input then multiply the output
by 2 because:
log(sqrt(x)) = 1/2(log(x)), so log(x) = 2log(sqrt(x))

Reviewed-by: Adhemerval Zanella <adhemerval.zanella@linaro.org>
sysdeps/aarch64/fpu/log10f_advsimd.c