math: add some assembly implementations on ARM64
Also add GP<->FP move addressing mode to FMOVS, FMOVD instructions. Ceil-8 37.1ns ± 0% 7.9ns ± 0% -78.64% (p=0.000 n=4+5) Dim-8 20.9ns ± 1% 11.3ns ± 0% -45.93% (p=0.008 n=5+5) Floor-8 22.9ns ± 0% 7.9ns ± 0% -65.41% (p=0.029 n=4+4) Gamma-8 117ns ± 0% 94ns ± 1% -19.50% (p=0.016 n=4+5) PowInt-8 121ns ± 0% 108ns ± 1% -11.07% (p=0.008 n=5+5) PowFrac-8 331ns ± 0% 318ns ± 0% -3.93% (p=0.000 n=5+4) Trunc-8 18.8ns ± 0% 7.9ns ± 0% -57.83% (p=0.016 n=4+5) Change-Id: I709b7f1a914b28adc27414522db551e2630cfb92 Reviewed-on: https://go-review.googlesource.com/29734 Run-TryBot: Cherry Zhang <cherryyz@google.com> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: David Chase <drchase@google.com>
Showing
src/math/dim_arm64.s
0 → 100644
src/math/floor_arm64.s
0 → 100644
src/math/modf_arm64.s
0 → 100644
Please register or sign in to comment