The commit 9247f53219 triggered some regressions on loongarch and
riscv:
math/test-float-log10
math/test-float32-log10
And it is due a wrong sync with CORE-MATH for special 0.0/-0.0
inputs.
Checked on aarch64-linux-gnu and loongarch64-linux-gnu-lp64d.
On GCC 11 (x86-64), the previous code produced test failures like
this one:
Failure: Test: exp10m1_towardzero (-0x1.1p+4)
Result:
is: -1.00000000e+00 -0x1.000000p+0
should be: -9.99999940e-01 -0x1.fffffep-1
difference: 5.96046447e-08 0x1.000000p-24
ulp : 1.0000
max.ulp : 0.0000
Apply a similar fix to exp2m1f.
Co-authored-by: Paul Zimmermann <Paul.Zimmermann@inria.fr>
Reviewed-by: Adhemerval Zanella <adhemerval.zanella@linaro.org>
The CORE-MATH exp2m1f implementation showed slight worse latency
when using x86_64 baseline ABI. This patch adds a ifunc variant
with similar performance for x86_64-v3.
Reviewed-by: Noah Goldstein <goldstein.w.n@gmail.com>
Reviewed-by: DJ Delorie <dj@redhat.com>
The CORE-MATH exp10m1f implementation showed slight worse latency
when using x86_64 baseline ABI. This patch adds a ifunc variant
with similar performance for x86_64-v3.
Reviewed-by: Noah Goldstein <goldstein.w.n@gmail.com>
Reviewed-by: DJ Delorie <dj@redhat.com>
The CORE-MATH implementation is correctly rounded (for any rounding mode)
and shows better performance compared to the generic exp2m1f.
The code was adapted to glibc style and to use the definition of
math_config.h (to handle errno, overflow, and underflow). The
only change is to handle FLT_MAX_EXP for FE_DOWNWARD or FE_TOWARDZERO.
The benchmark inputs are based on exp2f ones.
Benchtest on x64_64 (Ryzen 9 5900X, gcc 14.2.1), aarch64 (Neoverse-N1,
gcc 13.3.1), and powerpc (POWER10, gcc 13.2.1):
Latency master patched improvement
x86_64 40.6042 48.7104 -19.96%
x86_64v2 40.7506 35.9032 11.90%
x86_64v3 35.2301 31.7956 9.75%
i686 102.094 94.6657 7.28%
aarch64 18.2704 15.1387 17.14%
power10 11.9444 8.2402 31.01%
reciprocal-throughput master patched improvement
x86_64 20.8683 16.1428 22.64%
x86_64v2 19.5076 10.4474 46.44%
x86_64v3 19.2106 10.4014 45.86%
i686 56.4054 59.3004 -5.13%
aarch64 12.0781 7.3953 38.77%
power10 6.5306 5.9388 9.06%
The generic implementation calls __ieee754_exp2f and x86_64 provides
an optimized ifunc version (built with -mfma -mavx2, not correctly
rounded). This explains the performance difference for x86_64.
Same for i686, where the ABI provides an optimized __ieee754_exp2f
version built with '-msse2 -mfpmath=sse'. When built wth same
flags, the new algorithm shows a better performance:
master patched improvement
latency 102.094 91.2823 10.59%
reciprocal-throughput 56.4054 52.7984 6.39%
Signed-off-by: Alexei Sibidanov <sibid@uvic.ca>
Signed-off-by: Paul Zimmermann <Paul.Zimmermann@inria.fr>
Signed-off-by: Adhemerval Zanella <adhemerval.zanella@linaro.org>
Reviewed-by: DJ Delorie <dj@redhat.com>
The CORE-MATH implementation is correctly rounded (for any rounding mode)
and shows better performance compared to the generic exp10m1f.
The code was adapted to glibc style and to use the definition of
math_config.h (to handle errno, overflow, and underflow). I mostly
fixed some small issues in corner cases (sNaN handling, -INFINITY,
a specific overflow check).
Benchtest on x64_64 (Ryzen 9 5900X, gcc 14.2.1), aarch64 (Neoverse-N1,
gcc 13.3.1), and powerpc (POWER10, gcc 13.2.1):
Latency master patched improvement
x86_64 45.4690 49.5845 -9.05%
x86_64v2 46.1604 36.2665 21.43%
x86_64v3 37.8442 31.0359 17.99%
i686 121.367 93.0079 23.37%
aarch64 21.1126 15.0165 28.87%
power10 12.7426 8.4929 33.35%
reciprocal-throughput master patched improvement
x86_64 19.6005 17.4005 11.22%
x86_64v2 19.6008 11.1977 42.87%
x86_64v3 17.5427 10.2898 41.34%
i686 59.4215 60.9675 -2.60%
aarch64 13.9814 7.9173 43.37%
power10 6.7814 6.4258 5.24%
The generic implementation calls __ieee754_exp10f which has an
optimized version, although it is not correctly rounded, which is
the main culprit of the the latency difference for x86_64 and
throughp for i686.
Signed-off-by: Alexei Sibidanov <sibid@uvic.ca>
Signed-off-by: Paul Zimmermann <Paul.Zimmermann@inria.fr>
Signed-off-by: Adhemerval Zanella <adhemerval.zanella@linaro.org>
Reviewed-by: DJ Delorie <dj@redhat.com>
Also remove the use of builtins in favor of standard names, compiler
already inline them (if supported) with current compiler options.
It also fixes and issue where __builtin_roundeven is not support on
gcc older than version 10.
Checked on x86_64-linux-gnu and i686-linux_gnu.
Signed-off-by: Adhemerval Zanella <adhemerval.zanella@linaro.org>
Reviewed-by: DJ Delorie <dj@redhat.com>
The CORE-MATH implementation is correctly rounded (for any rounding mode).
This can be checked by exhaustive tests in a few minutes since there are
less than 2^32 values to check against for example GNU MPFR.
This patch also adds some bench values for tgammaf.
Tested on x86_64 and x86 (cfarm26).
With the initial GNU libc code it gave on an Intel(R) Core(TM) i7-8700:
"tgammaf": {
"": {
"duration": 3.50188e+09,
"iterations": 2e+07,
"max": 602.891,
"min": 65.1415,
"mean": 175.094
}
}
With the new code:
"tgammaf": {
"": {
"duration": 3.30825e+09,
"iterations": 5e+07,
"max": 211.592,
"min": 32.0325,
"mean": 66.1649
}
}
With the initial GNU libc code it gave on cfarm26 (i686):
"tgammaf": {
"": {
"duration": 3.70505e+09,
"iterations": 6e+06,
"max": 2420.23,
"min": 243.154,
"mean": 617.509
}
}
With the new code:
"tgammaf": {
"": {
"duration": 3.24497e+09,
"iterations": 1.8e+07,
"max": 1238.15,
"min": 101.155,
"mean": 180.276
}
}
Signed-off-by: Alexei Sibidanov <sibid@uvic.ca>
Signed-off-by: Paul Zimmermann <Paul.Zimmermann@inria.fr>
Changes in v2:
- include <math.h> (fix the linknamespace failures)
- restored original benchtests/strcoll-inputs/filelist#en_US.UTF-8 file
- restored original wrapper code (math/w_tgammaf_compat.c),
except for the dealing with the sign
- removed the tgammaf/float entries in all libm-test-ulps files
- address other comments from Joseph Myers
(https://sourceware.org/pipermail/libc-alpha/2024-July/158736.html)
Changes in v3:
- pass NULL argument for signgam from w_tgammaf_compat.c
- use of math_narrow_eval
- added more comments
Changes in v4:
- initialize local_signgam to 0 in math/w_tgamma_template.c
- replace sysdeps/ieee754/dbl-64/gamma_productf.c by dummy file
Changes in v5:
- do not mention local_signgam any more in math/w_tgammaf_compat.c
- initialize local_signgam to 1 instead of 0 in w_tgamma_template.c
and added comment
Changes in v6:
- pass NULL as 2nd argument of __ieee754_gammaf_r in
w_tgammaf_compat.c, and check for NULL in e_gammaf_r.c
Changes in v7:
- added Signed-off-by line for Alexei Sibidanov (author of the code)
Changes in v8:
- added Signed-off-by line for Paul Zimmermann (submitted of the patch)
Changes in v9:
- address comments from review by Adhemerval Zanella
Reviewed-by: Adhemerval Zanella <adhemerval.zanella@linaro.org>
GCC aligns global data to 16 bytes if their size is >= 16 bytes. This patch
changes the exp2f_data struct slightly so that the fields are better aligned.
As a result on targets that support them, load-pair instructions accessing
poly_scaled and invln2_scaled are now 16-byte aligned.
Reviewed-by: Adhemerval Zanella <adhemerval.zanella@linaro.org>
As discussed at the patch review meeting
Signed-off-by: Andreas K. Hüttel <dilfridge@gentoo.org>
Reviewed-by: Simon Chopin <simon.chopin@canonical.com>
C23 adds various <math.h> function families originally defined in TS
18661-4. Add the exp2m1 and exp10m1 functions (exp2(x)-1 and
exp10(x)-1, like expm1).
As with other such functions, these use type-generic templates that
could be replaced with faster and more accurate type-specific
implementations in future. Test inputs are copied from those for
expm1, plus some additions close to the overflow threshold (copied
from exp2 and exp10) and also some near the underflow threshold.
exp2m1 has the unusual property of having an input (M_MAX_EXP) where
whether the function overflows (under IEEE semantics) depends on the
rounding mode. Although these could reasonably be XFAILed in the
testsuite (as we do in some cases for arguments very close to a
function's overflow threshold when an error of a few ulps in the
implementation can result in the implementation not agreeing with an
ideal one on whether overflow takes place - the testsuite isn't smart
enough to handle this automatically), since these functions aren't
required to be correctly rounding, I made the implementation check for
and handle this case specially.
The Makefile ordering expected by lint-makefiles for the new functions
is a bit peculiar, but I implemented it in this patch so that the test
passes; I don't know why log2 also needed moving in one Makefile
variable setting when it didn't in my previous patches, but the
failure showed a different place was expected for that function as
well.
The powerpc64le IFUNC setup seems not to be as self-contained as one
might hope; it shouldn't be necessary to add IFUNCs for new functions
such as these simply to get them building, but without setting up
IFUNCs for the new functions, there were undefined references to
__GI___expm1f128 (that IFUNC machinery results in no such function
being defined, but doesn't stop include/math.h from doing the
redirection resulting in the exp2m1f128 and exp10m1f128
implementations expecting to call it).
Tested for x86_64 and x86, and with build-many-glibcs.py.
C23 adds various <math.h> function families originally defined in TS
18661-4. Add the log10p1 functions (log10(1+x): like log1p, but for
base-10 logarithms).
This is directly analogous to the log2p1 implementation (except that
whereas log2p1 has a smaller underflow range than log1p, log10p1 has a
larger underflow range). The test inputs are copied from those for
log1p and log2p1, plus a few more inputs in that wider underflow
range.
Tested for x86_64 and x86, and with build-many-glibcs.py.
C23 adds various <math.h> function families originally defined in TS
18661-4. Add the logp1 functions (aliases for log1p functions - the
name is intended to be more consistent with the new log2p1 and
log10p1, where clearly it would have been very confusing to name those
functions log21p and log101p). As aliases rather than new functions,
the content of this patch is somewhat different from those actually
adding new functions.
Tests are shared with log1p, so this patch *does* mechanically update
all affected libm-test-ulps files to expect the same errors for both
functions.
The vector versions of log1p on aarch64 and x86_64 are *not* updated
to have logp1 aliases (and thus there are no corresponding header,
tests, abilist or ulps changes for vector functions either). It would
be reasonable for such vector aliases and corresponding changes to
other files to be made separately. For now, the log1p tests instead
avoid testing logp1 in the vector case (a Makefile change is needed to
avoid problems with grep, used in generating the .c files for vector
function tests, matching more than one ALL_RM_TEST line in a file
testing multiple functions with the same inputs, when it assumes that
the .inc file only has a single such line).
Tested for x86_64 and x86, and with build-many-glibcs.py.
Left shift of ki is undefined when ki<0, copy the logic from exp,
which uses unsigned arithmetics, to fix it.
Reviewed-by: Adhemerval Zanella <adhemerval.zanella@linaro.org>
The libc.a for alpha, s390, and sparcv9 does not provide
copysignf64x, copysignf128, frexpf64x, frexpf128, modff64x, and
modff128.
Checked with a static build for the affected ABIs.
Reviewed-by: H.J. Lu <hjl.tools@gmail.com>
Don't provide __nexttowardf128_do_not_use, nexttowardf128_do_not_use,
finitef128_do_not_use, isinff128_do_not_use and isnanf128_do_not_use.
This fixes BZ #31757.
Signed-off-by: H.J. Lu <hjl.tools@gmail.com>
Reviewed-by: Adhemerval Zanella <adhemerval.zanella@linaro.org>
Some static implementation of float128 routines might call __isnanf128,
which is not provided by the static object.
Checked on x86_64-linux-gnu.
Reviewed-by: H.J. Lu <hjl.tools@gmail.com>
The commit 16439f419b removed the static fmod/fmodf on i386 and m68k
with and empty w_fmod.c (required for the ABIs that uses the newly
implementation). This patch fixes by adding the required symbols on
the arch-specific w_fmod{f}_compat.c implementation.
To statically build fmod fails on some ABI (alpha, s390, sparc) because
it does not export the ldexpf128, this is also fixed by this patch.
Checked on i686-linux-gnu and with a build for m68k-linux-gnu.
Reviewed-by: Aurelien Jarno <aurelien@aurel32.net>
Tested-by: Aurelien Jarno <aurelien@aurel32.net>
C23 adds various <math.h> function families originally defined in TS
18661-4. Add the log2p1 functions (log2(1+x): like log1p, but for
base-2 logarithms).
This illustrates the intended structure of implementations of all
these function families: define them initially with a type-generic
template implementation. If someone wishes to add type-specific
implementations, it is likely such implementations can be both faster
and more accurate than the type-generic one and can then override it
for types for which they are implemented (adding benchmarks would be
desirable in such cases to demonstrate that a new implementation is
indeed faster).
The test inputs are copied from those for log1p. Note that these
changes make gen-auto-libm-tests depend on MPFR 4.2 (or later).
The bulk of the changes are fairly generic for any such new function.
(sysdeps/powerpc/nofpu/Makefile only needs changing for those
type-generic templates that use fabs.)
Tested for x86_64 and x86, and with build-many-glibcs.py.
Fix BZ #31759 by not defining nearbyint aliases when used in IFUNC.
Signed-off-by: H.J. Lu <hjl.tools@gmail.com>
Reviewed-by: Noah Goldstein <goldstein.w.n@gmail.com>
Complete the internal renaming from "C2X" and related names in GCC by
renaming *-c2x and *-gnu2x tests to *-c23 and *-gnu23.
Tested for x86_64, and with build-many-glibcs.py for powerpc64le.
WG14 decided to use the name C23 as the informal name of the next
revision of the C standard (notwithstanding the publication date in
2024). Update references to C2X in glibc to use the C23 name.
This is intended to update everything *except* where it involves
renaming files (the changes involving renaming tests are intended to
be done separately). In the case of the _ISOC2X_SOURCE feature test
macro - the only user-visible interface involved - support for that
macro is kept for backwards compatibility, while adding
_ISOC23_SOURCE.
Tested for x86_64.
Remove the error handling wrapper from exp10. This is very similar to
the changes done to exp and exp2, except that we also need to handle
pow10 and pow10l.
Reviewed-by: Adhemerval Zanella <adhemerval.zanella@linaro.org>
New implementation is based on the existing exp/exp2, with different
reduction constants and polynomial. Worst-case error in round-to-
nearest is 0.513 ULP.
The exp/exp2 shared table is reused for exp10 - .rodata size of
e_exp_data increases by 64 bytes.
As for exp/exp2, targets with single-instruction rounding/conversion
intrinsics can use them by toggling TOINT_INTRINSICS=1 and adding the
necessary code to their math_private.h.
Improvements on Neoverse V1 compared to current GLIBC master:
exp10 thruput: 3.3x in [-0x1.439b746e36b52p+8 0x1.34413509f79ffp+8]
exp10 latency: 1.8x in [-0x1.439b746e36b52p+8 0x1.34413509f79ffp+8]
Tested on:
aarch64-linux-gnu (TOINT_INTRINSICS, fma contraction) and
x86_64-linux-gnu (!TOINT_INTRINSICS, no fma contraction)
Reviewed-by: Szabolcs Nagy <szabolcs.nagy@arm.com>
With GCC 14 on 32-bit x86 the compiler emits a maybe-uninitialized
warning:
../sysdeps/ieee754/dbl-64/k_rem_pio2.c: In function '__kernel_rem_pio2':
../sysdeps/ieee754/dbl-64/k_rem_pio2.c:364:20: error: 'fq' may be used uninitialized [-Werror=maybe-uninitialized]
364 | y[0] = fq[0]; y[1] = fq[1]; y[2] = fw;
| ~~^~~
This is similar to the warning that is suppressed in the other branch of
the switch. Help the compiler knowing that the variable is always
initialized, which also makes the suppression obsolete.
On Skylake, it changes log1p bench performance by:
Before After Improvement
max 63.349 58.347 8%
min 4.448 5.651 -30%
mean 12.0674 10.336 14%
The minimum code path is
if (hx < 0x3FDA827A) /* x < 0.41422 */
{
if (__glibc_unlikely (ax >= 0x3ff00000)) /* x <= -1.0 */
{
...
}
if (__glibc_unlikely (ax < 0x3e200000)) /* |x| < 2**-29 */
{
math_force_eval (two54 + x); /* raise inexact */
if (ax < 0x3c900000) /* |x| < 2**-54 */
{
...
}
else
return x - x * x * 0.5;
FMA and non-FMA code sequences look similar. Non-FMA version is slightly
faster. Since log1p is called by asinh and atanh, it improves asinh
performance by:
Before After Improvement
max 75.645 63.135 16%
min 10.074 10.071 0%
mean 15.9483 14.9089 6%
and improves atanh performance by:
Before After Improvement
max 91.768 75.081 18%
min 15.548 13.883 10%
mean 18.3713 16.8011 8%
On Skylake, it improves expm1 bench performance by:
Before After Improvement
max 70.204 68.054 3%
min 20.709 16.2 22%
mean 22.1221 16.7367 24%
NB: Add
extern long double __expm1l (long double);
extern long double __expm1f128 (long double);
for __typeof (__expm1l) and __typeof (__expm1f128) when __expm1 is
defined since __expm1 may be expanded in their declarations which
causes the build failure.
Bump autoconf requirement to 2.71 to allow regenerating configure on
more recent distributions. autoconf 2.71 has been in Fedora since F36
and is the current version in Debian stable (bookworm). It appears to
be current in Gentoo as well.
All sysdeps configure and preconfigure scripts have also been
regenerated; all changes are trivial transformations that do not affect
functionality.
Signed-off-by: Siddhesh Poyarekar <siddhesh@sourceware.org>
Reviewed-by: Carlos O'Donell <carlos@redhat.com>
Return value from *scanf and *asprintf routines are now properly checked
in test-scanf-ldbl-compat-template.c and test-printf-ldbl-compat.c.
Reviewed-by: Siddhesh Poyarekar <siddhesh@sourceware.org>
This allows to include bits/syslog-decl.h in include/sys/syslog.h and
therefore be able to create the libc_hidden_builtin_proto (__syslog_chk)
prototype.
Reviewed-by: Siddhesh Poyarekar <siddhesh@sourceware.org>
If libc_hidden_builtin_{def,proto} isn't properly set for *_chk routines,
there are unwanted PLT entries in libc.so.
There is a special case with __asprintf_chk:
If ldbl_* macros are used for asprintf, ABI gets broken on s390x,
if it isn't, ppc64le isn't building due to multiple asm redirections.
This is due to the inclusion of bits/stdio-lbdl.h for ppc64le whereas it
isn't for s390x. This header creates redirections, which are not
compatible with the ones generated using libc_hidden_def.
Yet, we can't use libc_hidden_ldbl_proto on s390x since it will not
create a simple strong alias (e.g. as done on x86_64), but a versioned
alias, leading to ABI breakage.
This results in errors on s390x:
/usr/bin/ld: glibc/iconv/../libio/bits/stdio2.h:137: undefined reference
to `__asprintf_chk'
Original __asprintf_chk symbols:
00000000001395b0 T __asprintf_chk
0000000000177e90 T __nldbl___asprintf_chk
__asprintf_chk symbols with ldbl_* macros:
000000000012d590 t ___asprintf_chk
000000000012d590 t __asprintf_chk@@GLIBC_2.4
000000000012d590 t __GI___asprintf_chk
000000000012d590 t __GL____asprintf_chk___asprintf_chk
0000000000172240 T __nldbl___asprintf_chk
__asprintf_chk symbols with the patch:
000000000012d590 t ___asprintf_chk
000000000012d590 T __asprintf_chk
000000000012d590 t __GI___asprintf_chk
0000000000172240 T __nldbl___asprintf_chk
Reviewed-by: Carlos O'Donell <carlos@redhat.com>
The *_chk routines naming doesn't match the name that would be generated
using libc_hidden_ldbl_proto. Since the macro is needed for some of
these *_chk functions for _FORTIFY_SOURCE to be enabled, that needed to
be fixed.
While at it, all the *_chk function get renamed appropriately for
consistency, even if not strictly necessary.
Reviewed-by: Siddhesh Poyarekar <siddhesh@sourceware.org>
Reviewed-by: Paul E. Murphy <murphyp@linux.ibm.com>
Since the _FORTIFY_SOURCE feature uses some routines of Glibc, they need to
be excluded from the fortification.
On top of that:
- some tests explicitly verify that some level of fortification works
appropriately, we therefore shouldn't modify the level set for them.
- some objects need to be build with optimization disabled, which
prevents _FORTIFY_SOURCE to be used for them.
Assembler files that implement architecture specific versions of the
fortified routines were not excluded from _FORTIFY_SOURCE as there is no
C header included that would impact their behavior.
Reviewed-by: Siddhesh Poyarekar <siddhesh@sourceware.org>
Replace the loop-over-scalar placeholder routines with optimised
implementations from Arm Optimized Routines (AOR).
Also add some headers containing utilities for aarch64 libmvec
routines, and update libm-test-ulps.
Data tables for new routines are used via a pointer with a
barrier on it, in order to prevent overly aggressive constant
inlining in GCC. This allows a single adrp, combined with offset
loads, to be used for every constant in the table.
Special-case handlers are marked NOINLINE in order to confine the
save/restore overhead of switching from vector to normal calling
standard. This way we only incur the extra memory access in the
exceptional cases. NOINLINE definitions have been moved to
math_private.h in order to reduce duplication.
AOR exposes a config option, WANT_SIMD_EXCEPT, to enable
selective masking (and later fixing up) of invalid lanes, in
order to trigger fp exceptions correctly (AdvSIMD only). This is
tested and maintained in AOR, however it is configured off at
source level here for performance reasons. We keep the
WANT_SIMD_EXCEPT blocks in routine sources to greatly simplify
the upstreaming process from AOR to glibc.
Reviewed-by: Szabolcs Nagy <szabolcs.nagy@arm.com>
This patch redirects the error functions to the appropriate
longdouble variants which enables the compiler to optimize
for the abi ieeelongdouble.
Signed-off-by: Sachin Monga <smonga@linux.ibm.com>
Optimize the fast paths (x < y) and (x/y < 2^12). Delay handling of special
cases to reduce the number of instructions executed before the fast paths.
Performance improvements for fmod:
Skylake Zen2 Neoverse V1
subnormals 11.8% 4.2% 11.5%
normal 3.9% 0.01% -0.5%
close-exponents 6.3% 5.6% 19.4%
Reviewed-by: Adhemerval Zanella <adhemerval.zanella@linaro.org>
The error handling is moved to sysdeps/ieee754 version with no SVID
support. The compatibility symbol versions still use the wrapper
with SVID error handling around the new code. There is no new symbol
version nor compatibility code on !LIBM_SVID_COMPAT targets
(e.g. riscv).
The ia64 is unchanged, since it still uses the arch specific
__libm_error_region on its implementation. For both i686 and m68k,
which provive arch specific implementation, wrappers are added so
no new symbol are added (which would require to change the
implementations).
It shows an small improvement, the results for fmod:
Architecture | Input | master | patch
-----------------|-----------------|----------|--------
x86_64 (Ryzen 9) | subnormals | 12.5049 | 9.40992
x86_64 (Ryzen 9) | normal | 296.939 | 296.738
x86_64 (Ryzen 9) | close-exponents | 16.0244 | 13.119
aarch64 (N1) | subnormal | 6.81778 | 4.33313
aarch64 (N1) | normal | 155.620 | 152.915
aarch64 (N1) | close-exponents | 8.21306 | 5.76138
armhf (N1) | subnormal | 15.1083 | 14.5746
armhf (N1) | normal | 244.833 | 241.738
armhf (N1) | close-exponents | 21.8182 | 22.457
Checked on x86_64-linux-gnu, i686-linux-gnu, and aarch64-linux-gnu.
Reviewed-by: Wilco Dijkstra <Wilco.Dijkstra@arm.com>
This uses a new algorithm similar to already proposed earlier [1].
With x = mx * 2^ex and y = my * 2^ey (mx, my, ex, ey being integers),
the simplest implementation is:
mx * 2^ex == 2 * mx * 2^(ex - 1)
while (ex > ey)
{
mx *= 2;
--ex;
mx %= my;
}
With mx/my being mantissa of double floating pointer, on each step the
argument reduction can be improved 8 (which is sizeof of uint32_t minus
MANTISSA_WIDTH plus the signal bit):
while (ex > ey)
{
mx << 8;
ex -= 8;
mx %= my;
} */
The implementation uses builtin clz and ctz, along with shifts to
convert hx/hy back to doubles. Different than the original patch,
this path assume modulo/divide operation is slow, so use multiplication
with invert values.
I see the following performance improvements using fmod benchtests
(result only show the 'mean' result):
Architecture | Input | master | patch
-----------------|-----------------|----------|--------
x86_64 (Ryzen 9) | subnormals | 17.2549 | 12.0318
x86_64 (Ryzen 9) | normal | 85.4096 | 49.9641
x86_64 (Ryzen 9) | close-exponents | 19.1072 | 15.8224
aarch64 (N1) | subnormal | 10.2182 | 6.81778
aarch64 (N1) | normal | 60.0616 | 20.3667
aarch64 (N1) | close-exponents | 11.5256 | 8.39685
I also see similar improvements on arm-linux-gnueabihf when running on
the N1 aarch64 chips, where it a lot of soft-fp implementation (for
modulo, and multiplication):
Architecture | Input | master | patch
-----------------|-----------------|----------|--------
armhf (N1) | subnormal | 11.6662 | 10.8955
armhf (N1) | normal | 69.2759 | 34.1524
armhf (N1) | close-exponents | 13.6472 | 18.2131
Instead of using the math_private.h definitions, I used the
math_config.h instead which is used on newer math implementations.
Co-authored-by: kirill <kirill.okhotnikov@gmail.com>
[1] https://sourceware.org/pipermail/libc-alpha/2020-November/119794.html
Reviewed-by: Wilco Dijkstra <Wilco.Dijkstra@arm.com>
This uses a new algorithm similar to already proposed earlier [1].
With x = mx * 2^ex and y = my * 2^ey (mx, my, ex, ey being integers),
the simplest implementation is:
mx * 2^ex == 2 * mx * 2^(ex - 1)
while (ex > ey)
{
mx *= 2;
--ex;
mx %= my;
}
With mx/my being mantissa of double floating pointer, on each step the
argument reduction can be improved 11 (which is sizeo of uint64_t minus
MANTISSA_WIDTH plus the signal bit):
while (ex > ey)
{
mx << 11;
ex -= 11;
mx %= my;
} */
The implementation uses builtin clz and ctz, along with shifts to
convert hx/hy back to doubles. Different than the original patch,
this path assume modulo/divide operation is slow, so use multiplication
with invert values.
I see the following performance improvements using fmod benchtests
(result only show the 'mean' result):
Architecture | Input | master | patch
-----------------|-----------------|----------|--------
x86_64 (Ryzen 9) | subnormals | 19.1584 | 12.5049
x86_64 (Ryzen 9) | normal | 1016.51 | 296.939
x86_64 (Ryzen 9) | close-exponents | 18.4428 | 16.0244
aarch64 (N1) | subnormal | 11.153 | 6.81778
aarch64 (N1) | normal | 528.649 | 155.62
aarch64 (N1) | close-exponents | 11.4517 | 8.21306
I also see similar improvements on arm-linux-gnueabihf when running on
the N1 aarch64 chips, where it a lot of soft-fp implementation (for
modulo, clz, ctz, and multiplication):
Architecture | Input | master | patch
-----------------|-----------------|----------|--------
armhf (N1) | subnormal | 15.908 | 15.1083
armhf (N1) | normal | 837.525 | 244.833
armhf (N1) | close-exponents | 16.2111 | 21.8182
Instead of using the math_private.h definitions, I used the
math_config.h instead which is used on newer math implementations.
Co-authored-by: kirill <kirill.okhotnikov@gmail.com>
[1] https://sourceware.org/pipermail/libc-alpha/2020-November/119794.html
Reviewed-by: Wilco Dijkstra <Wilco.Dijkstra@arm.com>
They are both used by __libc_freeres to free all library malloc
allocated resources to help tooling like mtrace or valgrind with
memory leak tracking.
The current scheme uses assembly markers and linker script entries
to consolidate the free routine function pointers in the RELRO segment
and to be freed buffers in BSS.
This patch changes it to use specific free functions for
libc_freeres_ptrs buffers and call the function pointer array directly
with call_function_static_weak.
It allows the removal of both the internal macros and the linker
script sections.
Checked on x86_64-linux-gnu, i686-linux-gnu, and aarch64-linux-gnu.
Reviewed-by: Carlos O'Donell <carlos@redhat.com>
