diff options
author | Pierre Blanchard <pierre.blanchard@arm.com> | 2022-09-21 10:29:29 +0100 |
---|---|---|
committer | Pierre Blanchard <pierre.blanchard@arm.com> | 2022-09-21 10:30:48 +0100 |
commit | 52558f2f6ac3a22b2e0382f10caa87f10fb0b23d (patch) | |
tree | 76bf6d9c7cc5ffb77763e5f201aa054778cd803e /pl/math/test/ulp_wrappers.h | |
parent | b56e103d1afaa625557f1f275fa2b6a0606f6456 (diff) | |
download | arm-optimized-routines-52558f2f6ac3a22b2e0382f10caa87f10fb0b23d.tar.gz |
pl/math: Add Vector/Neon tanf
The Neon implementation is accurate up to 3.2ulps.
It relies on range reduction to [0, pi/4] and late reciprocal
depending on quadrant, as opposed to the early reciprocal used
in the scalar routine.
Diffstat (limited to 'pl/math/test/ulp_wrappers.h')
-rw-r--r-- | pl/math/test/ulp_wrappers.h | 2 |
1 files changed, 2 insertions, 0 deletions
diff --git a/pl/math/test/ulp_wrappers.h b/pl/math/test/ulp_wrappers.h index 0ceef13..2ab1af2 100644 --- a/pl/math/test/ulp_wrappers.h +++ b/pl/math/test/ulp_wrappers.h @@ -62,6 +62,7 @@ VF1_WRAP(erfc) VF1_WRAP(log10) VF1_WRAP(log1p) VF1_WRAP(log2) +VF1_WRAP(tan) VD1_WRAP(atan) VD2_WRAP(atan2) VD1_WRAP(erf) @@ -77,6 +78,7 @@ ZVNF1_WRAP(erfc) ZVNF1_WRAP(log10) ZVNF1_WRAP(log1p) ZVNF1_WRAP(log2) +ZVNF1_WRAP(tan) ZVND1_WRAP(atan) ZVND2_WRAP(atan2) ZVND1_WRAP(erf) |