diff options
| author | Adhemerval Zanella <adhemerval.zanella@linaro.org> | 2021-04-03 23:52:45 -0300 |
|---|---|---|
| committer | Adhemerval Zanella <adhemerval.zanella@linaro.org> | 2021-12-13 09:08:07 -0300 |
| commit | 2eb1cd2f47fe6568c539fa105551bb73df8368ec (patch) | |
| tree | 8404bf6b82969a864d916c1bdbda844fad7231e2 /sysdeps/powerpc/powerpc32/power4/fpu/multiarch/Makefile | |
| parent | a1d3c9b64266d84736ba8eac96aa864165905b79 (diff) | |
| download | glibc-2eb1cd2f47fe6568c539fa105551bb73df8368ec.tar.gz glibc-2eb1cd2f47fe6568c539fa105551bb73df8368ec.tar.xz glibc-2eb1cd2f47fe6568c539fa105551bb73df8368ec.zip | |
math: Remove powerpc e_hypot
The generic implementation is shows only slight worse performance: POWER10 reciprocal-throughput latency master 8.28478 13.7253 new hypot 7.21945 13.1933 POWER9 reciprocal-throughput latency master 13.4024 14.0967 new hypot 14.8479 15.8061 POWER8 reciprocal-throughput latency master 15.5767 16.8885 new hypot 16.5371 18.4057 One way to improve might to make gcc generate xsmaxdp/xsmindp for fmax/fmin (it onl does for -ffast-math, clang does for default options). Checked on powerpc64-linux-gnu (power8) and powerpc64le-linux-gnu (power9).
Diffstat (limited to 'sysdeps/powerpc/powerpc32/power4/fpu/multiarch/Makefile')
| -rw-r--r-- | sysdeps/powerpc/powerpc32/power4/fpu/multiarch/Makefile | 5 |
1 files changed, 1 insertions, 4 deletions
diff --git a/sysdeps/powerpc/powerpc32/power4/fpu/multiarch/Makefile b/sysdeps/powerpc/powerpc32/power4/fpu/multiarch/Makefile index 60f2c95532..1de0f9b350 100644 --- a/sysdeps/powerpc/powerpc32/power4/fpu/multiarch/Makefile +++ b/sysdeps/powerpc/powerpc32/power4/fpu/multiarch/Makefile @@ -15,8 +15,7 @@ libm-sysdep_routines += s_llrintf-power6 s_llrintf-ppc32 s_llrint-power6 \ s_lrint-ppc32 s_modf-power5+ s_modf-ppc32 \ s_modff-power5+ s_modff-ppc32 s_logbl-power7 \ s_logbl-ppc32 s_logb-power7 s_logb-ppc32 \ - s_logbf-power7 s_logbf-ppc32 e_hypot-power7 \ - e_hypot-ppc32 e_hypotf-power7 e_hypotf-ppc32 + s_logbf-power7 s_logbf-ppc32 CFLAGS-s_llrintf-power6.c += -mcpu=power6 CFLAGS-s_llrintf-ppc32.c += -mcpu=power4 @@ -35,8 +34,6 @@ CFLAGS-s_modff-power5+.c = -mcpu=power5+ CFLAGS-s_logbl-power7.c = -mcpu=power7 CFLAGS-s_logb-power7.c = -mcpu=power7 CFLAGS-s_logbf-power7.c = -mcpu=power7 -CFLAGS-e_hypot-power7.c = -mcpu=power7 -CFLAGS-e_hypotf-power7.c = -mcpu=power7 # These files quiet sNaNs in a way that is optimized away without # -fsignaling-nans. |