i386: Replace assembly versions of e_powf with generic e_powf.c
This patch replaces i386 assembly versions of e_powf with generic
e_powf.c. For workload-spec2017.wrf, on Nehalem, it improves
performance by:
Before After Improvement
reciprocal-throughput 230.855 78.3358 194%
latency 231.685 94.1259 146%
On Skylake, it improves performance by:
Before After Improvement
reciprocal-throughput 239.858 47.4713 405%
latency 247.57 93.8798 163%
On IvyBridge with --disable-multi-arch, it improves performance by:
Before After Improvement
reciprocal-throughput 269.078 63.3758 324%
latency 271.473 102.091 165%
* sysdeps/i386/fpu/e_powf.S: Removed.
* sysdeps/i386/fpu/e_powf_log2_data.c: Likewise.
* sysdeps/i386/fpu/w_powf.c: Likewise.
* sysdeps/i386/fpu/libm-test-ulps: Updated for generic e_powf.c.
* sysdeps/i386/i686/fpu/multiarch/libm-test-ulps: Likewise.
* sysdeps/i386/i686/fpu/multiarch/Makefile (libm-sysdep_routines):
Add e_powf-sse2.
(CFLAGS-e_powf-sse2.c): New.
* sysdeps/i386/i686/fpu/multiarch/e_powf-sse2.c: New file.
* sysdeps/i386/i686/fpu/multiarch/e_powf.c: Likewise.