For arguments with X^2 + Y^2 close to 1, clog and clog10 avoid large
errors from log(hypot) by computing X^2 + Y^2 - 1 in a way that avoids
cancellation error and then using log1p.
However, the thresholds for using that approach still result in log
being used on argument as large as sqrt(13/16) > 0.9, leading to
significant errors, in some cases above the 9ulp maximum allowed in
glibc libm. This patch arranges for the approach using log1p to be
used in any cases where |X|, |Y| < 1 and X^2 + Y^2 >= 0.5 (with the
existing allowance for cases where one of X and Y is very small),
adjusting the __x2y2m1 functions to work with the wider range of
inputs. This way, log only gets used on arguments below sqrt(1/2) (or
substantially above 1), where the error involved is much less.
Tested for x86_64, x86, mips64 and powerpc. For the ulps regeneration
I removed the existing clog and clog10 ulps before regenerating to
allow any reduced ulps to appear. Tests added include those found by
random test generation to produce large ulps either before or after
the patch, and some found by trying inputs close to the (0.75, 0.5)
threshold where the potential errors from using log are largest.
[BZ #19016]
* sysdeps/generic/math_private.h (__x2y2m1f): Update comment to
allow more cases with X^2 + Y^2 >= 0.5.
* sysdeps/ieee754/dbl-64/x2y2m1.c (__x2y2m1): Likewise. Add -1 as
normal element in sum instead of special-casing based on values of
arguments.
* sysdeps/ieee754/dbl-64/x2y2m1f.c (__x2y2m1f): Update comment.
* sysdeps/ieee754/ldbl-128/x2y2m1l.c (__x2y2m1l): Likewise. Add
-1 as normal element in sum instead of special-casing based on
values of arguments.
* sysdeps/ieee754/ldbl-128ibm/x2y2m1l.c (__x2y2m1l): Likewise.
* sysdeps/ieee754/ldbl-96/x2y2m1.c [FLT_EVAL_METHOD != 0]
(__x2y2m1): Update comment.
* sysdeps/ieee754/ldbl-96/x2y2m1l.c (__x2y2m1l): Likewise. Add -1
as normal element in sum instead of special-casing based on values
of arguments.
* math/s_clog.c (__clog): Handle more cases using log1p without
hypot.
* math/s_clog10.c (__clog10): Likewise.
* math/s_clog10f.c (__clog10f): Likewise.
* math/s_clog10l.c (__clog10l): Likewise.
* math/s_clogf.c (__clogf): Likewise.
* math/s_clogl.c (__clogl): Likewise.
* math/auto-libm-test-in: Add more tests of clog and clog10.
* math/auto-libm-test-out: Regenerated.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
Various i386 libm functions return values with excess range and
precision; Wilco Dijkstra's patches to make isfinite etc. expand
inline cause this pre-existing issue to result in test failures (when
e.g. a result that overflows float but not long double gets counted as
overflowing for some purposes but not others).
This patch addresses those cases arising from functions defined in C,
adding a math_narrow_eval macro that forces values to memory to
eliminate excess precision if FLT_EVAL_METHOD indicates this is
needed, and is a no-op otherwise. I'll convert existing uses of
volatile and asm for this purpose to use the new macro later, once
i386 has clean test results again (which requires fixes for .S files
as well).
Tested for x86_64 and x86. Committed.
[BZ #18980]
* sysdeps/generic/math_private.h: Include <float.h>.
(math_narrow_eval): New macro.
[FLT_EVAL_METHOD != 0] (excess_precision): Likewise.
* sysdeps/ieee754/dbl-64/e_cosh.c (__ieee754_cosh): Use
math_narrow_eval on overflowing return value.
* sysdeps/ieee754/dbl-64/e_lgamma_r.c (__ieee754_lgamma_r):
Likewise.
* sysdeps/ieee754/dbl-64/e_sinh.c (__ieee754_sinh): Likewise.
* sysdeps/ieee754/flt-32/e_coshf.c (__ieee754_coshf): Likewise.
* sysdeps/ieee754/flt-32/e_lgammaf_r.c (__ieee754_lgammaf_r):
Likewise.
* sysdeps/ieee754/flt-32/e_sinhf.c (__ieee754_sinhf): Likewise.
The existing implementations of lgamma functions (except for the ia64
versions) use the reflection formula for negative arguments. This
suffers large inaccuracy from cancellation near zeros of lgamma (near
where the gamma function is +/- 1).
This patch fixes this inaccuracy. For arguments above -2, there are
no zeros and no large cancellation, while for sufficiently large
negative arguments the zeros are so close to integers that even for
integers +/- 1ulp the log(gamma(1-x)) term dominates and cancellation
is not significant. Thus, it is only necessary to take special care
about cancellation for arguments around a limited number of zeros.
Accordingly, this patch uses precomputed tables of relevant zeros,
expressed as the sum of two floating-point values. The log of the
ratio of two sines can be computed accurately using log1p in cases
where log would lose accuracy. The log of the ratio of two gamma(1-x)
values can be computed using Stirling's approximation (the difference
between two values of that approximation to lgamma being computable
without computing the two values and then subtracting), with
appropriate adjustments (which don't reduce accuracy too much) in
cases where 1-x is too small to use Stirling's approximation directly.
In the interval from -3 to -2, using the ratios of sines and of
gamma(1-x) can still produce too much cancellation between those two
parts of the computation (and that interval is also the worst interval
for computing the ratio between gamma(1-x) values, which computation
becomes more accurate, while being less critical for the final result,
for larger 1-x). Because this can result in errors slightly above
those accepted in glibc, this interval is instead dealt with by
polynomial approximations. Separate polynomial approximations to
(|gamma(x)|-1)(x-n)/(x-x0) are used for each interval of length 1/8
from -3 to -2, where n (-3 or -2) is the nearest integer to the
1/8-interval and x0 is the zero of lgamma in the relevant half-integer
interval (-3 to -2.5 or -2.5 to -2).
Together, the two approaches are intended to give sufficient accuracy
for all negative arguments in the problem range. Outside that range,
the previous implementation continues to be used.
Tested for x86_64, x86, mips64 and powerpc. The mips64 and powerpc
testing shows up pre-existing problems for ldbl-128 and ldbl-128ibm
with large negative arguments giving spurious "invalid" exceptions
(exposed by newly added tests for cases this patch doesn't affect the
logic for); I'll address those problems separately.
[BZ #2542]
[BZ #2543]
[BZ #2558]
* sysdeps/ieee754/dbl-64/e_lgamma_r.c (__ieee754_lgamma_r): Call
__lgamma_neg for arguments from -28.0 to -2.0.
* sysdeps/ieee754/flt-32/e_lgammaf_r.c (__ieee754_lgammaf_r): Call
__lgamma_negf for arguments from -15.0 to -2.0.
* sysdeps/ieee754/ldbl-128/e_lgammal_r.c (__ieee754_lgammal_r):
Call __lgamma_negl for arguments from -48.0 or -50.0 to -2.0.
* sysdeps/ieee754/ldbl-96/e_lgammal_r.c (__ieee754_lgammal_r):
Call __lgamma_negl for arguments from -33.0 to -2.0.
* sysdeps/ieee754/dbl-64/lgamma_neg.c: New file.
* sysdeps/ieee754/dbl-64/lgamma_product.c: Likewise.
* sysdeps/ieee754/flt-32/lgamma_negf.c: Likewise.
* sysdeps/ieee754/flt-32/lgamma_productf.c: Likewise.
* sysdeps/ieee754/ldbl-128/lgamma_negl.c: Likewise.
* sysdeps/ieee754/ldbl-128/lgamma_productl.c: Likewise.
* sysdeps/ieee754/ldbl-128ibm/lgamma_negl.c: Likewise.
* sysdeps/ieee754/ldbl-128ibm/lgamma_productl.c: Likewise.
* sysdeps/ieee754/ldbl-96/lgamma_negl.c: Likewise.
* sysdeps/ieee754/ldbl-96/lgamma_product.c: Likewise.
* sysdeps/ieee754/ldbl-96/lgamma_productl.c: Likewise.
* sysdeps/generic/math_private.h (__lgamma_negf): New prototype.
(__lgamma_neg): Likewise.
(__lgamma_negl): Likewise.
(__lgamma_product): Likewise.
(__lgamma_productl): Likewise.
* math/Makefile (libm-calls): Add lgamma_neg and lgamma_product.
* math/auto-libm-test-in: Add more tests of lgamma.
* math/auto-libm-test-out: Regenerated.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
Concluding the fixes for C90 libm functions calling C99 fe* functions,
this patch fixes the case of feupdateenv by making it a weak alias for
__feupdateenv and making the affected code call __feupdateenv.
Tested for x86_64 (testsuite, and that installed stripped shared
libraries are unchanged by the patch). Also tested for ARM
(soft-float) that the math.h linknamespace tests now pass.
[BZ #17748]
* include/fenv.h (__feupdateenv): Use libm_hidden_proto.
* math/feupdateenv.c (__feupdateenv): Use libm_hidden_def.
* sysdeps/aarch64/fpu/feupdateenv.c (feupdateenv): Rename to
__feupdateenv and define as weak alias of __feupdateenv. Use
libm_hidden_weak.
* sysdeps/alpha/fpu/feupdateenv.c (__feupdateenv): Use
libm_hidden_def.
* sysdeps/arm/feupdateenv.c (feupdateenv): Rename to __feupdateenv
and define as weak alias of __feupdateenv. Use libm_hidden_weak.
* sysdeps/hppa/fpu/feupdateenv.c (feupdateenv): Likewise.
* sysdeps/i386/fpu/feupdateenv.c (__feupdateenv): Use
libm_hidden_def.
* sysdeps/ia64/fpu/feupdateenv.c (feupdateenv): Rename to
__feupdateenv and define as weak alias of __feupdateenv. Use
libm_hidden_weak.
* sysdeps/m68k/fpu/feupdateenv.c (__feupdateenv): Use
libm_hidden_def.
* sysdeps/mips/fpu/feupdateenv.c (feupdateenv): Rename to
__feupdateenv and define as weak alias of __feupdateenv. Use
libm_hidden_weak.
* sysdeps/powerpc/fpu/feupdateenv.c (__feupdateenv): Use
libm_hidden_def.
* sysdeps/powerpc/nofpu/feupdateenv.c (__feupdateenv): Likewise.
* sysdeps/powerpc/powerpc32/e500/nofpu/feupdateenv.c
(__feupdateenv): Likewise.
* sysdeps/s390/fpu/feupdateenv.c (feupdateenv): Rename to
__feupdateenv and define as weak alias of __feupdateenv. Use
libm_hidden_weak.
* sysdeps/sh/sh4/fpu/feupdateenv.c (feupdateenv): Likewise.
* sysdeps/sparc/fpu/feupdateenv.c (__feupdateenv): Use
libm_hidden_def.
* sysdeps/tile/math_private.h (__feupdateenv): New inline
function.
* sysdeps/x86_64/fpu/feupdateenv.c (__feupdateenv): Use
libm_hidden_def.
* sysdeps/generic/math_private.h (default_libc_feupdateenv): Call
__feupdateenv instead of feupdateenv.
(default_libc_feupdateenv_test): Likewise.
(libc_feresetround_ctx): Likewise.
Continuing the fixes for C90 libm functions calling C99 fe* functions,
this patch fixes the case of fesetround by making it a weak alias of
__fesetround and making the affected code call __fesetround. An
existing __fesetround function in fenv_libc.h for powerpc is renamed
to __fesetround_inline.
Tested for x86_64 (testsuite, and that disassembly of installed shared
libraries is unchanged by the patch). Also tested for ARM
(soft-float) that fesetround failures disappear from the linknamespace
test results (feupdateenv remains to be addressed to complete fixing
bug 17748).
[BZ #17748]
* include/fenv.h (__fesetround): Declare. Use libm_hidden_proto.
* math/fesetround.c (fesetround): Rename to __fesetround and
define as weak alias of __fesetround. Use libm_hidden_weak.
* sysdeps/aarch64/fpu/fesetround.c (fesetround): Likewise.
* sysdeps/alpha/fpu/fesetround.c (fesetround): Likewise.
* sysdeps/arm/fesetround.c (fesetround): Likewise.
* sysdeps/hppa/fpu/fesetround.c (fesetround): Likewise.
* sysdeps/i386/fpu/fesetround.c (fesetround): Likewise.
* sysdeps/ia64/fpu/fesetround.c (fesetround): Likewise.
* sysdeps/m68k/fpu/fesetround.c (fesetround): Likewise.
* sysdeps/mips/fpu/fesetround.c (fesetround): Likewise.
* sysdeps/powerpc/fpu/fenv_libc.h (__fesetround): Rename to
__fesetround_inline.
* sysdeps/powerpc/fpu/fenv_private.h (libc_fesetround_ppc): Call
__fesetround_inline instead of __fesetround.
* sysdeps/powerpc/fpu/fesetround.c (fesetround): Rename to
__fesetround and define as weak alias of __fesetround. Use
libm_hidden_weak. Call __fesetround_inline instead of
__fesetround.
* sysdeps/powerpc/nofpu/fesetround.c (fesetround): Rename to
__fesetround and define as weak alias of __fesetround. Use
libm_hidden_weak.
* sysdeps/powerpc/powerpc32/e500/nofpu/fesetround.c (fesetround):
Likewise.
* sysdeps/s390/fpu/fesetround.c (fesetround): Likewise.
* sysdeps/sh/sh4/fpu/fesetround.c (fesetround): Likewise.
* sysdeps/sparc/fpu/fesetround.c (fesetround): Likewise.
* sysdeps/tile/math_private.h (__fesetround): New inline function.
* sysdeps/x86_64/fpu/fesetround.c (fesetround): Rename to
__fesetround and define as weak alias of __fesetround. Use
libm_hidden_weak.
* sysdeps/generic/math_private.h (default_libc_fesetround): Call
__fesetround instead of fesetround.
(default_libc_feholdexcept_setround): Likewise.
(libc_feholdsetround_ctx): Likewise.
(libc_feholdsetround_noex_ctx): Likewise.
Continuing the fixes for C90 libm functions calling C99 fe* functions,
this patch fixes the case of fesetenv by making it a weak alias of
__fesetenv and making the affected code (including various copies of
feupdateenv which also gets called from C90 functions) call
__fesetenv.
Tested for x86_64 (testsuite, and that disassembly of installed shared
libraries is unchanged by the patch). Also tested for ARM
(soft-float) that fesetenv failures disappear from the linknamespace
test results (fsetround and feupdateenv remain to be addressed to
complete fixing bug 17748).
[BZ #17748]
* include/fenv.h (__fesetenv): Use libm_hidden_proto.
* math/fesetenv.c (__fesetenv): Use libm_hidden_def.
* sysdeps/aarch64/fpu/fesetenv.c (fesetenv): Rename to __fesetenv
and define as weak alias of __fesetenv. Use libm_hidden_weak.
* sysdeps/alpha/fpu/fesetenv.c (__fesetenv): Use libm_hidden_def.
* sysdeps/arm/fesetenv.c (fesetenv): Rename to __fesetenv and
define as weak alias of __fesetenv. Use libm_hidden_weak.
* sysdeps/hppa/fpu/fesetenv.c (fesetenv): Likewise.
* sysdeps/i386/fpu/fesetenv.c (__fesetenv): Use libm_hidden_def.
* sysdeps/ia64/fpu/fesetenv.c (fesetenv): Rename to __fesetenv and
define as weak alias of __fesetenv. Use libm_hidden_weak.
* sysdeps/m68k/fpu/fesetenv.c (__fesetenv): Use libm_hidden_def.
* sysdeps/mips/fpu/fesetenv.c (fesetenv): Rename to __fesetenv and
define as weak alias of __fesetenv. Use libm_hidden_weak.
* sysdeps/powerpc/fpu/fesetenv.c (__fesetenv): Use
libm_hidden_def.
* sysdeps/powerpc/nofpu/fesetenv.c (__fesetenv): Likewise.
* sysdeps/powerpc/powerpc32/e500/nofpu/fesetenv.c (__fesetenv):
Likewise.
* sysdeps/s390/fpu/fesetenv.c (fesetenv): Rename to __fesetenv and
define as weak alias of __fesetenv. Use libm_hidden_weak.
* sysdeps/sh/sh4/fpu/fesetenv.c (fesetenv): Likewise.
* sysdeps/sparc/fpu/fesetenv.c (__fesetenv): Use libm_hidden_def.
* sysdeps/tile/math_private.h (__fesetenv): New inline function.
* sysdeps/x86_64/fpu/fesetenv.c (fesetenv): Rename to __fesetenv
and define as weak alias of __fesetenv. Use libm_hidden_weak.
* sysdeps/generic/math_private.h (default_libc_fesetenv): Use
__fesetenv instead of fesetenv.
(libc_feresetround_noex_ctx): Likewise.
* sysdeps/alpha/fpu/feupdateenv.c (__feupdateenv): Likewise.
* sysdeps/hppa/fpu/feupdateenv.c (feupdateenv): Likewise.
* sysdeps/i386/fpu/feupdateenv.c (__feupdateenv): Likewise.
* sysdeps/ia64/fpu/feupdateenv.c (feupdateenv): Likewise.
* sysdeps/m68k/fpu/feupdateenv.c (__feupdateenv): Likewise.
* sysdeps/mips/fpu/feupdateenv.c (feupdateenv): Likewise.
* sysdeps/powerpc/nofpu/feupdateenv.c (__feupdateenv): Likewise.
* sysdeps/powerpc/powerpc32/e500/nofpu/feupdateenv.c
(__feupdateenv): Likewise.
* sysdeps/s390/fpu/feupdateenv.c (feupdateenv): Likewise.
* sysdeps/sh/sh4/fpu/feupdateenv.c (feupdateenv): Likewise.
* sysdeps/sparc/fpu/feupdateenv.c (__feupdateenv): Likewise.
* sysdeps/x86_64/fpu/feupdateenv.c (__feupdateenv): Likewise.
Continuing the fixes for C90 libm functions calling C99 fe* functions,
this patch fixes the case of feholdexcept by making it a weak alias of
__feholdexcept and making the affected code call __feholdexcept.
Tested for x86_64 (testsuite, and that disassembly of installed shared
libraries is unchanged by the patch). Also tested for ARM
(soft-float) that feholdexcept failures disappear from the
linknamespace test failures (fesetenv, fsetround and feupdateenv
remain to be addressed to complete fixing bug 17748).
[BZ #17748]
* include/fenv.h (__feholdexcept): Declare. Use
libm_hidden_proto.
* math/feholdexcpt.c (feholdexcept): Rename to __feholdexcept and
define as weak alias of __feholdexcept. Use libm_hidden_weak.
* sysdeps/aarch64/fpu/feholdexcpt.c (feholdexcept): Likewise.
* sysdeps/alpha/fpu/feholdexcpt.c (feholdexcept): Likewise.
* sysdeps/arm/feholdexcpt.c (feholdexcept): Likewise.
* sysdeps/hppa/fpu/feholdexcpt.c (feholdexcept): Likewise.
* sysdeps/i386/fpu/feholdexcpt.c (feholdexcept): Likewise.
* sysdeps/ia64/fpu/feholdexcpt.c (feholdexcept): Likewise.
* sysdeps/m68k/fpu/feholdexcpt.c (feholdexcept): Likewise.
* sysdeps/mips/fpu/feholdexcpt.c (feholdexcept): Likewise.
* sysdeps/powerpc/fpu/feholdexcpt.c (feholdexcept): Likewise.
* sysdeps/powerpc/nofpu/feholdexcpt.c (feholdexcept): Likewise.
* sysdeps/powerpc/powerpc32/e500/nofpu/feholdexcpt.c
(feholdexcept): Likewise.
* sysdeps/s390/fpu/feholdexcpt.c (feholdexcept): Likewise.
* sysdeps/sh/sh4/fpu/feholdexcpt.c (feholdexcept): Likewise.
* sysdeps/sparc/fpu/feholdexcpt.c (feholdexcept): Likewise.
* sysdeps/x86_64/fpu/feholdexcpt.c (feholdexcept): Likewise.
* sysdeps/generic/math_private.h (default_libc_feholdexcept): Use
__feholdexcept instead of feholdexcept.
(default_libc_feholdexcept_setround): Likewise.
Some C90 libm functions call fegetenv via libc_feholdsetround*
functions in math_private.h. This patch makes them call __fegetenv
instead, making fegetenv into a weak alias for __fegetenv as needed.
Tested for x86_64 (testsuite, and that disassembly of installed shared
libraries is unchanged by the patch). Also tested for ARM
(soft-float) that fegetenv failures disappear from the linknamespace
test failures (however, similar fixes will also be needed for
fegetround, feholdexcept, fesetenv, fesetround and feupdateenv before
this set of namespace issues covered by bug 17748 is fully fixed and
those linknamespace tests start passing).
[BZ #17748]
* include/fenv.h (__fegetenv): Use libm_hidden_proto.
* math/fegetenv.c (__fegetenv): Use libm_hidden_def.
* sysdeps/aarch64/fpu/fegetenv.c (fegetenv): Rename to __fegetenv
and define as weak alias of __fegetenv. Use libm_hidden_weak.
* sysdeps/alpha/fpu/fegetenv.c (__fegetenv): Use libm_hidden_def.
* sysdeps/arm/fegetenv.c (fegetenv): Rename to __fegetenv and
define as weak alias of __fegetenv. Use libm_hidden_weak.
* sysdeps/hppa/fpu/fegetenv.c (fegetenv): Likewise.
* sysdeps/i386/fpu/fegetenv.c (__fegetenv): Use libm_hidden_def.
* sysdeps/ia64/fpu/fegetenv.c (fegetenv): Rename to __fegetenv and
define as weak alias of __fegetenv. Use libm_hidden_weak.
* sysdeps/m68k/fpu/fegetenv.c (__fegetenv): Use libm_hidden_def.
* sysdeps/mips/fpu/fegetenv.c (fegetenv): Rename to __fegetenv and
define as weak alias of __fegetenv. Use libm_hidden_weak.
* sysdeps/powerpc/fpu/fegetenv.c (__fegetenv): Use
libm_hidden_def.
* sysdeps/powerpc/nofpu/fegetenv.c (__fegetenv): Likewise.
* sysdeps/powerpc/powerpc32/e500/nofpu/fegetenv.c (__fegetenv):
Likewise.
* sysdeps/s390/fpu/fegetenv.c (fegetenv): Rename to __fegetenv and
define as weak alias of __fegetenv. Use libm_hidden_weak.
* sysdeps/sh/sh4/fpu/fegetenv.c (fegetenv): Likewise.
* sysdeps/sparc/fpu/fegetenv.c (__fegetenv): Use libm_hidden_def.
* sysdeps/tile/math_private.h (__fegetenv): New inline function.
* sysdeps/x86_64/fpu/fegetenv.c (fegetenv): Rename to __fegetenv
and define as weak alias of __fegetenv. Use libm_hidden_weak.
* sysdeps/generic/math_private.h (libc_feholdsetround_ctx): Use
__fegetenv instead of fegetenv.
(libc_feholdsetround_noex_ctx): Likewise.
This patch adds a generic implementation of HAVE_RM_CTX using standard
fenv calls. As a result math functions using SET_RESTORE_ROUND* macros
do not suffer from a large slowdown on targets which do not implement
optimized libc_fe*_ctx inline functions. Most of the libc_fe* inline
functions are now unused and could be removed in the future (there are
a few math functions left which use a mixture of standard fenv calls
and libc_fe* inline functions - they could be updated to use
SET_RESTORE_ROUND or improved to avoid expensive fenv manipulations
across just a few FP instructions).
libc_feholdsetround*_noex_ctx is added to enable better optimization of
SET_RESTORE_ROUND_NOEX* implementations.
Performance measurements on ARM and x86 of sin() show significant gains
over the current default, fairly close to a highly optimized fenv_private:
ARM x86
no fenv_private : 100% 100%
generic HAVE_RM_CTX : 250% 350%
fenv_private (CTX) : 250% 450%
2014-06-23 Will Newton <will.newton@linaro.org>
Wilco <wdijkstr@arm.com>
* sysdeps/generic/math_private.h: Add generic HAVE_RM_CTX
implementation. Include get-rounding-mode.h.
[!HAVE_RM_CTX]: Define HAVE_RM_CTX to zero.
[!libc_feholdsetround_noex_ctx]: Define
libc_feholdsetround_noex_ctx.
[!libc_feholdsetround_noexf_ctx]: Define
libc_feholdsetround_noexf_ctx.
[!libc_feholdsetround_noexl_ctx]: Define
libc_feholdsetround_noexl_ctx.
(libc_feholdsetround_ctx): New function.
(libc_feresetround_ctx): New function.
(libc_feholdsetround_noex_ctx): New function.
(libc_feresetround_noex_ctx): New function.
ChangeLog:
2014-03-17 Will Newton <will.newton@linaro.org>
* sysdeps/generic/math_private.h: Check whether
HAVE_RM_CTX is defined with #ifdef rather
than #if.
The most common use case of math functions is with default rounding
mode, i.e. rounding to nearest. Setting and restoring rounding mode
is an unnecessary overhead for this, so I've added support for a
context, which does the set/restore only if the FP status needs a
change. The code is written such that only x86 uses these. Other
architectures should be unaffected by it, but would definitely benefit
if the set/restore has as much overhead relative to the rest of the
code, as the x86 bits do.
Here's a summary of the performance improvement due to these
improvements; I've only mentioned functions that use the set/restore
and have benchmark inputs for x86_64:
Before:
cos(): ITERS:4.69335e+08: TOTAL:28884.6Mcy, MAX:4080.28cy, MIN:57.562cy, 16248.6 calls/Mcy
exp(): ITERS:4.47604e+08: TOTAL:28796.2Mcy, MAX:207.721cy, MIN:62.385cy, 15543.9 calls/Mcy
pow(): ITERS:1.63485e+08: TOTAL:28879.9Mcy, MAX:362.255cy, MIN:172.469cy, 5660.86 calls/Mcy
sin(): ITERS:3.89578e+08: TOTAL:28900Mcy, MAX:704.859cy, MIN:47.583cy, 13480.2 calls/Mcy
tan(): ITERS:7.0971e+07: TOTAL:28902.2Mcy, MAX:1357.79cy, MIN:388.58cy, 2455.55 calls/Mcy
After:
cos(): ITERS:6.0014e+08: TOTAL:28875.9Mcy, MAX:364.283cy, MIN:45.716cy, 20783.4 calls/Mcy
exp(): ITERS:5.48578e+08: TOTAL:28764.9Mcy, MAX:191.617cy, MIN:51.011cy, 19071.1 calls/Mcy
pow(): ITERS:1.70013e+08: TOTAL:28873.6Mcy, MAX:689.522cy, MIN:163.989cy, 5888.18 calls/Mcy
sin(): ITERS:4.64079e+08: TOTAL:28891.5Mcy, MAX:6959.3cy, MIN:36.189cy, 16062.8 calls/Mcy
tan(): ITERS:7.2354e+07: TOTAL:28898.9Mcy, MAX:1295.57cy, MIN:380.698cy, 2503.7 calls/Mcy
So the improvements are:
cos: 27.9089%
exp: 22.6919%
pow: 4.01564%
sin: 19.1585%
tan: 1.96086%
The downside of the change is that it will have an adverse performance
impact on non-default rounding modes, but I think the tradeoff is
justified.
[BZ #6794]
Following Joseph comments about bug 6794, here is a proposed fix. It turned out
to be a large fix mainly because I had to move some file along to follow libm
files/names conventions.
Basically I have added wrappers (w_ilogb.c, w_ilogbf.c, w_ilogbl.c) that now calls
the symbol '__ieee754_ilogb'. The wrappers checks for '__ieee754_ilogb' output and
set the errno and raise exceptions as expected.
The '__ieee754_ilogb' is implemented in sysdeps. I have moved the 's_ilogb[f|l]' files
to e_ilogb[f|l] and renamed the '__ilogb[f|l]' to '__ieee754_ilogb[f|l]'.
I also found out a bug in i386 and x86-64 assembly coded ilogb implementation where
it raises a FE_DIVBYZERO when argument is '0.0'. I corrected this issue as well.
Finally I added the errno and FE_INVALID tests for 0.0, NaN and +-InF argument. Tested
on i386, x86-64, ppc32 and ppc64.