glibc/sysdeps
Adhemerval Zanella 1192696069 powerpc: Remove optimized finite
The powerpc finite optimization do not show much gain:

  - GCC will call libm iff -fsignaling-nans is used. This usage pattern
    is usually not performance oriented and for such calls PLT overhead
    should dominate execution time.

  - The power7 uses ftdiv to optimize for some input patterns, but at
    cost of others.  Comparing against generic C implementation built
    for powerpc64-linux-gnu-power7 (--with-cpu=power7):

    - Generic sysdeps/ieee754 implementation:
       "isfinite": {
        "": {
         "duration": 5.0082e+09,
         "iterations": 2.45299e+09,
         "max": 43.824,
         "min": 2.008,
         "mean": 2.04167
        },
        "INF": {
         "duration": 4.66554e+09,
         "iterations": 2.28288e+09,
         "max": 35.73,
         "min": 2.008,
         "mean": 2.04371
        },
        "NAN": {
         "duration": 4.66274e+09,
         "iterations": 2.28716e+09,
         "max": 34.161,
         "min": 2.009,
         "mean": 2.03866
        }
       }

    - power7 optimized one:
       "isfinite": {
        "": {
         "duration": 4.99111e+09,
         "iterations": 2.65566e+09,
         "max": 25.015,
         "min": 1.716,
         "mean": 1.87942
        },
        "INF": {
         "duration": 4.6783e+09,
         "iterations": 2.0999e+09,
         "max": 35.264,
         "min": 1.868,
         "mean": 2.22787
        },
        "NAN": {
         "duration": 4.67915e+09,
         "iterations": 2.08678e+09,
         "max": 38.099,
         "min": 1.869,
         "mean": 2.24228
        }
       }

     So it basically optimizes marginally for normal numbers while
     increasing the latency for other kind of FP.

  - The power8 implementation is just the generic implementation using
    ISA 2.07 mfvsrd instruction (which GCC uses for generic implementation).
    So generic implementation is the best option for powerpc64le.

Checked on powerpc-linux-gnu (built without --with-cpu, with
--with-cpu=power4 and with --with-cpu=power5+ and --disable-multi-arch),
powerpc64-linux-gnu (built without --with-cp and with --with-cpu=power5+
and --disable-multi-arch).

	* sysdeps/powerpc/powerpc32/power4/fpu/multiarch/Makefile
	(sysdeps_routines, libm-sysdep_routines): Remove s_finite*
	objects.
	* sysdeps/powerpc/powerpc32/power4/fpu/multiarch/s_finite-power7.S:
	Remove file.
	* sysdeps/powerpc/powerpc32/power4/fpu/multiarch/s_finite-ppc32.c:
	Likewise.
	* sysdeps/powerpc/powerpc32/power4/fpu/multiarch/s_finite.c: Likewise.
	* sysdeps/powerpc/powerpc32/power4/fpu/multiarch/s_finitef-ppc32.c:
	Likewise.
	* sysdeps/powerpc/powerpc32/power4/fpu/multiarch/s_finitef.c: Likewise.
	* sysdeps/powerpc/powerpc32/power7/fpu/s_finite.S: Likewise.
	* sysdeps/powerpc/powerpc32/power7/fpu/s_finitef.S: Likewise.
	* sysdeps/powerpc/powerpc64/fpu/multiarch/Makefile (sysdep_call):
	Remove s_finite* objects.
	* sysdeps/powerpc/powerpc64/fpu/multiarch/s_finite-power7.S: Remove file.
	* sysdeps/powerpc/powerpc64/fpu/multiarch/s_finite-power8.S: Likewise.
	* sysdeps/powerpc/powerpc64/fpu/multiarch/s_finite-ppc64.c: Likewise.
	* sysdeps/powerpc/powerpc64/fpu/multiarch/s_finite.c: Likewise.
	* sysdeps/powerpc/powerpc64/fpu/multiarch/s_finitef-ppc64.c: Likewise.
	* sysdeps/powerpc/powerpc64/fpu/multiarch/s_finitef.c: Likewise.
	* sysdeps/powerpc/powerpc64/power7/fpu/s_finite.S: Likewise.
	* sysdeps/powerpc/powerpc64/power7/fpu/s_finitef.S: Likewise.
	* sysdeps/powerpc/powerpc64/power8/fpu/s_finite.S: Likewise.
	* sysdeps/powerpc/powerpc64/power8/fpu/s_finitef.S: Likewise.

Reviewed-by: Gabriel F. T. Gomes <gabrielftg@linux.ibm.com>
2019-06-12 14:32:39 -03:00
..
aarch64 aarch64: thunderx2 memmove performance improvements 2019-05-03 11:01:34 -07:00
alpha alpha: Improve sysdeps/alpha/divqu.S and sysdeps/alpha/remqu.S 2019-04-01 16:00:37 +07:00
arm Break further lines before not after operators. 2019-02-26 15:01:50 +00:00
csky C-SKY: mark lr as undefined to stop unwinding 2019-03-11 09:51:14 +08:00
generic Add single-thread.h header 2019-05-14 08:41:15 -03:00
gnu Add UDP_GRO from Linux 5.0 to netinet/udp.h. 2019-03-25 13:16:46 +00:00
hppa Add some spaces before '('. 2019-02-27 13:55:45 +00:00
htl hurd: advertise *_setpshared as not supported 2019-01-02 22:21:34 +01:00
hurd Update copyright dates with scripts/update-copyrights. 2019-01-01 00:11:28 +00:00
i386 Add and move fall-through comments in system-specific code. 2019-02-26 02:09:18 +00:00
ia64 Refactor hp-timing rtld usage 2019-03-22 17:30:44 -03:00
ieee754 math: Use wordsize-64 version for finite 2019-06-12 14:32:39 -03:00
init_array Update copyright dates with scripts/update-copyrights. 2019-01-01 00:11:28 +00:00
m68k wcsmbs: optimize wcpcpy 2019-02-27 10:00:34 -03:00
mach misc: Add twalk_r function 2019-05-02 11:42:51 +02:00
microblaze Break more lines before not after operators. 2019-02-25 13:19:19 +00:00
mips Add and move fall-through comments in system-specific code. 2019-02-26 02:09:18 +00:00
nios2 Update copyright dates with scripts/update-copyrights. 2019-01-01 00:11:28 +00:00
nptl x86: Remove arch-specific low level lock implementation 2019-05-14 08:48:02 -03:00
posix nss: Turn __nss_database_lookup into a compatibility symbol 2019-05-15 15:07:49 +02:00
powerpc powerpc: Remove optimized finite 2019-06-12 14:32:39 -03:00
pthread <semaphore.h>: Add nonnull attributes 2019-04-29 10:11:35 +02:00
riscv riscv: remove DL_RO_DYN_SECTION 2019-04-25 10:53:08 -07:00
s390 Fix iconv buffer handling with IGNORE error handler (bug #18830) 2019-06-04 14:03:04 +02:00
sh Update copyright dates with scripts/update-copyrights. 2019-01-01 00:11:28 +00:00
sparc Refactor hp-timing rtld usage 2019-03-22 17:30:44 -03:00
unix Linux: Deprecate <sys/sysctl.h> and sysctl 2019-06-12 14:32:08 +02:00
wordsize-32 Update copyright dates with scripts/update-copyrights. 2019-01-01 00:11:28 +00:00
wordsize-64 Update copyright dates with scripts/update-copyrights. 2019-01-01 00:11:28 +00:00
x86 Refactor hp-timing rtld usage 2019-03-22 17:30:44 -03:00
x86_64 wcsmbs: optimize wcscat 2019-02-27 10:00:37 -03:00