2001-09-19 10:37:31 +00:00
|
|
|
# Begin of automatic generation
|
|
|
|
|
|
|
|
|
|
# Maximal error of functions:
|
Update.
2003-11-28 Ulrich Drepper <drepper@redhat.com>
* sysdeps/x86_64/fpu/libm-test-ulps: Add some more minor changes
to compensate other setup.
2003-11-27 Andreas Jaeger <aj@suse.de>
* sysdeps/x86_64/fpu/libm-test-ulps: Add ulps for new atan2 test.
* math/libm-test.inc (atan2_test): Add test that run infinitly.
Reported by "Willus" <etc231etc231@willus.com>.
2003-11-27 Michael Matz <matz@suse.de>
* sysdeps/ieee754/dbl-64/mpsqrt.c (fastiroot): Fix 64-bit problem
with wrong types.
2003-11-28 Jakub Jelinek <jakub@redhat.com>
* posix/regexec.c (acquire_init_state_context): Make inline.
Add always_inline attribute.
(check_matching): Add BE macro. Move if (cur_state->has_backref)
into if (dfa->nbackref).
(sift_states_backward): Fix comment.
(transit_state): Add BE macro. Move if (next_state->has_backref)
into if (dfa->nbackref && next_state). Don't check for next_state
!= NULL twice.
* posix/regcomp.c (peek_token): Use opr.ctx_type instead of opr.idx
for ANCHOR.
(parse_expression): Only call init_word_char if word context will be
needed.
* posix/bug-regex11.c (tests): Add new tests.
* posix/tst-regex.c: Include getopt.h.
(timing): New variable.
(main): Set timing to 1 if --timing argument is present.
Add 2 new tests.
(run_test, run_test_backwards): Handle timing.
2003-11-27 Jakub Jelinek <jakub@redhat.com>
* posix/regex_internal.h (re_string_t): Remove mbs_case field.
Add offsets, valid_raw_len, raw_len, raw_stop, mbs_allocated and
offsets_needed fields. Change icase, is_utf8 and map_notascii
type from int bitfield to unsigned char.
(MBS_ALLOCATED, MBS_CASE_ALLOCATED): Remove.
(build_wcs_upper_buffer): Change prototype to return int.
(re_string_peek_byte_case, re_string_fetch_byte_case): Remove
defines, add prototypes.
* posix/regex_internal.c (re_string_allocate): Don't initialize
stop here. Don't initialize mbs_case. Set valid_raw_len.
Use mbs_allocated instead of MBS_* macros.
(re_string_construct): Don't initialize stop and valid_len here.
Don't initialize mbs_case. Use mbs_allocated instead of MBS_*
macros. Reallocate buffers if build_wcs_upper_buffer converted
too few bytes. Set valid_len to bufs_len only for single byte
no translation and set in that case valid_raw_len as well.
(re_string_realloc_buffers): Reallocate offsets if not NULL.
Use mbs_allocated instead of MBS_ALLOCATED. Don't reallocate
mbs_case.
(re_string_construct_common): Initialize raw_len, mbs_allocated,
stop and raw_stop.
(build_wcs_buffer): Apply pstr->trans before mbrtowc instead of
after it. Set valid_raw_len. Don't set mbs_case.
(build_wcs_upper_buffer): Return REG_NOERROR or REG_ESPACE.
Only use the fast path if !pstr->offsets_needed. Apply pstr->trans
before mbrtowc instead of after it. If upper case character
uses different number of bytes than lower case, goto to the
slow path. Don't call towupper unnecessarily twice. Set
valid_raw_len as well. Handle in the slow path the case if
lower and upper case use different number of characters.
Don't set mbs_case.
(re_string_skip_chars): Use valid_raw_len instead of valid_len.
(build_upper_buffer): Don't set mbs_case. Add BE macro. Set
valid_raw_len.
(re_string_translate_buffer): Set mbs instead of mbs_case. Set
valid_raw_len.
(re_string_reconstruct): Use raw_len/raw_stop to initialize
len/stop. Clear valid_raw_len and offsets_needed when clearing
valid_len. Use mbs_allocated instead of MBS_* macros.
Check original offset against valid_raw_len instead of valid_len.
Remove mbs_case handling. Adjust valid_raw_len together with
valid_len. If is_utf8 and looking for tip context, apply
pstr->trans first. If buffers start with partial multi-byte
character, initialize mbs array as well if mbs_allocated.
Check return value of build_wcs_upper_buffer.
(re_string_peek_byte_case): New function.
(re_string_fetch_byte_case): New function.
(re_string_destruct): Use mbs_allocated instead of MBS_ALLOCATED.
Don't free mbs_case. Free offsets.
* posix/regcomp.c (init_dfa): Only check if charset name is UTF-8
if mb_cur_max == 6.
* posix/regexec.c (re_search_internal): Initialize input.raw_stop
as well. Use valid_raw_len instead of valid_len when looking
through fastmap. Adjust registers through input.offsets.
(extend_buffers): Allow build_wcs_upper_buffer to fail.
* posix/bug-regex18.c (tests): Enable #ifdefed out tests. Add new
tests.
2003-11-29 06:13:09 +00:00
|
|
|
Function: "acos":
|
2015-03-25 00:30:10 +00:00
|
|
|
float: 1
|
|
|
|
|
ifloat: 1
|
Update.
2003-11-28 Ulrich Drepper <drepper@redhat.com>
* sysdeps/x86_64/fpu/libm-test-ulps: Add some more minor changes
to compensate other setup.
2003-11-27 Andreas Jaeger <aj@suse.de>
* sysdeps/x86_64/fpu/libm-test-ulps: Add ulps for new atan2 test.
* math/libm-test.inc (atan2_test): Add test that run infinitly.
Reported by "Willus" <etc231etc231@willus.com>.
2003-11-27 Michael Matz <matz@suse.de>
* sysdeps/ieee754/dbl-64/mpsqrt.c (fastiroot): Fix 64-bit problem
with wrong types.
2003-11-28 Jakub Jelinek <jakub@redhat.com>
* posix/regexec.c (acquire_init_state_context): Make inline.
Add always_inline attribute.
(check_matching): Add BE macro. Move if (cur_state->has_backref)
into if (dfa->nbackref).
(sift_states_backward): Fix comment.
(transit_state): Add BE macro. Move if (next_state->has_backref)
into if (dfa->nbackref && next_state). Don't check for next_state
!= NULL twice.
* posix/regcomp.c (peek_token): Use opr.ctx_type instead of opr.idx
for ANCHOR.
(parse_expression): Only call init_word_char if word context will be
needed.
* posix/bug-regex11.c (tests): Add new tests.
* posix/tst-regex.c: Include getopt.h.
(timing): New variable.
(main): Set timing to 1 if --timing argument is present.
Add 2 new tests.
(run_test, run_test_backwards): Handle timing.
2003-11-27 Jakub Jelinek <jakub@redhat.com>
* posix/regex_internal.h (re_string_t): Remove mbs_case field.
Add offsets, valid_raw_len, raw_len, raw_stop, mbs_allocated and
offsets_needed fields. Change icase, is_utf8 and map_notascii
type from int bitfield to unsigned char.
(MBS_ALLOCATED, MBS_CASE_ALLOCATED): Remove.
(build_wcs_upper_buffer): Change prototype to return int.
(re_string_peek_byte_case, re_string_fetch_byte_case): Remove
defines, add prototypes.
* posix/regex_internal.c (re_string_allocate): Don't initialize
stop here. Don't initialize mbs_case. Set valid_raw_len.
Use mbs_allocated instead of MBS_* macros.
(re_string_construct): Don't initialize stop and valid_len here.
Don't initialize mbs_case. Use mbs_allocated instead of MBS_*
macros. Reallocate buffers if build_wcs_upper_buffer converted
too few bytes. Set valid_len to bufs_len only for single byte
no translation and set in that case valid_raw_len as well.
(re_string_realloc_buffers): Reallocate offsets if not NULL.
Use mbs_allocated instead of MBS_ALLOCATED. Don't reallocate
mbs_case.
(re_string_construct_common): Initialize raw_len, mbs_allocated,
stop and raw_stop.
(build_wcs_buffer): Apply pstr->trans before mbrtowc instead of
after it. Set valid_raw_len. Don't set mbs_case.
(build_wcs_upper_buffer): Return REG_NOERROR or REG_ESPACE.
Only use the fast path if !pstr->offsets_needed. Apply pstr->trans
before mbrtowc instead of after it. If upper case character
uses different number of bytes than lower case, goto to the
slow path. Don't call towupper unnecessarily twice. Set
valid_raw_len as well. Handle in the slow path the case if
lower and upper case use different number of characters.
Don't set mbs_case.
(re_string_skip_chars): Use valid_raw_len instead of valid_len.
(build_upper_buffer): Don't set mbs_case. Add BE macro. Set
valid_raw_len.
(re_string_translate_buffer): Set mbs instead of mbs_case. Set
valid_raw_len.
(re_string_reconstruct): Use raw_len/raw_stop to initialize
len/stop. Clear valid_raw_len and offsets_needed when clearing
valid_len. Use mbs_allocated instead of MBS_* macros.
Check original offset against valid_raw_len instead of valid_len.
Remove mbs_case handling. Adjust valid_raw_len together with
valid_len. If is_utf8 and looking for tip context, apply
pstr->trans first. If buffers start with partial multi-byte
character, initialize mbs array as well if mbs_allocated.
Check return value of build_wcs_upper_buffer.
(re_string_peek_byte_case): New function.
(re_string_fetch_byte_case): New function.
(re_string_destruct): Use mbs_allocated instead of MBS_ALLOCATED.
Don't free mbs_case. Free offsets.
* posix/regcomp.c (init_dfa): Only check if charset name is UTF-8
if mb_cur_max == 6.
* posix/regexec.c (re_search_internal): Initialize input.raw_stop
as well. Use valid_raw_len instead of valid_len when looking
through fastmap. Adjust registers through input.offsets.
(extend_buffers): Allow build_wcs_upper_buffer to fail.
* posix/bug-regex18.c (tests): Enable #ifdefed out tests. Add new
tests.
2003-11-29 06:13:09 +00:00
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
|
|
|
|
|
2012-04-30 09:38:06 +00:00
|
|
|
Function: "acos_downward":
|
2015-03-25 00:30:10 +00:00
|
|
|
double: 1
|
2012-04-30 09:38:06 +00:00
|
|
|
float: 1
|
2015-03-25 00:30:10 +00:00
|
|
|
idouble: 1
|
2012-04-30 09:38:06 +00:00
|
|
|
ifloat: 1
|
2015-03-25 00:30:10 +00:00
|
|
|
ildouble: 2
|
|
|
|
|
ldouble: 2
|
2012-04-30 09:38:06 +00:00
|
|
|
|
|
|
|
|
Function: "acos_towardzero":
|
2015-03-25 00:30:10 +00:00
|
|
|
double: 1
|
2012-04-30 09:38:06 +00:00
|
|
|
float: 1
|
2015-03-25 00:30:10 +00:00
|
|
|
idouble: 1
|
2012-04-30 09:38:06 +00:00
|
|
|
ifloat: 1
|
2015-03-25 00:30:10 +00:00
|
|
|
ildouble: 2
|
|
|
|
|
ldouble: 2
|
2012-04-30 09:38:06 +00:00
|
|
|
|
|
|
|
|
Function: "acos_upward":
|
2013-11-30 22:04:13 +00:00
|
|
|
double: 1
|
2015-03-25 00:30:10 +00:00
|
|
|
float: 1
|
2013-11-30 22:04:13 +00:00
|
|
|
idouble: 1
|
2015-03-25 00:30:10 +00:00
|
|
|
ifloat: 1
|
|
|
|
|
ildouble: 2
|
|
|
|
|
ldouble: 2
|
2012-04-30 09:38:06 +00:00
|
|
|
|
Add more libm-test coverage of [a-c]* real functions.
Various libm functions have inadequate test coverage in libm-test.inc
/ auto-libm-test-in - failing to cover all the usual special cases
(infinities, NaNs, zero, large and small finite values, subnormals) as
well as a reasonable range of ordinary inputs and, where appropriate,
inputs close to the thresholds for underflow and overflow.
This patch improves test coverage for real functions [a-c]* (with the
expectation of adding more coverage for other functions later).
Tested x86_64 and x86 and ulps updated accordingly (and eight glibc
bugs and one C11 DR filed for issues found in the process).
* math/auto-libm-test-in: Add more tests of acos, acosh, asin,
asinh, atan, atan2, atanh, cbrt, cos and cosh.
* math/auto-libm-test-out: Regenerated.
* math/libm-test.inc (acosh_test_data): Add more tests.
(atanh_test_data): Likewise.
(ceil_test_data): Likewise.
(copysign_test_data): Likewise.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2013-12-20 21:03:39 +00:00
|
|
|
Function: "acosh":
|
|
|
|
|
double: 1
|
2015-02-25 00:01:15 +00:00
|
|
|
float: 1
|
Add more libm-test coverage of [a-c]* real functions.
Various libm functions have inadequate test coverage in libm-test.inc
/ auto-libm-test-in - failing to cover all the usual special cases
(infinities, NaNs, zero, large and small finite values, subnormals) as
well as a reasonable range of ordinary inputs and, where appropriate,
inputs close to the thresholds for underflow and overflow.
This patch improves test coverage for real functions [a-c]* (with the
expectation of adding more coverage for other functions later).
Tested x86_64 and x86 and ulps updated accordingly (and eight glibc
bugs and one C11 DR filed for issues found in the process).
* math/auto-libm-test-in: Add more tests of acos, acosh, asin,
asinh, atan, atan2, atanh, cbrt, cos and cosh.
* math/auto-libm-test-out: Regenerated.
* math/libm-test.inc (acosh_test_data): Add more tests.
(atanh_test_data): Likewise.
(ceil_test_data): Likewise.
(copysign_test_data): Likewise.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2013-12-20 21:03:39 +00:00
|
|
|
idouble: 1
|
2015-02-25 00:01:15 +00:00
|
|
|
ifloat: 1
|
|
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
Add more libm-test coverage of [a-c]* real functions.
Various libm functions have inadequate test coverage in libm-test.inc
/ auto-libm-test-in - failing to cover all the usual special cases
(infinities, NaNs, zero, large and small finite values, subnormals) as
well as a reasonable range of ordinary inputs and, where appropriate,
inputs close to the thresholds for underflow and overflow.
This patch improves test coverage for real functions [a-c]* (with the
expectation of adding more coverage for other functions later).
Tested x86_64 and x86 and ulps updated accordingly (and eight glibc
bugs and one C11 DR filed for issues found in the process).
* math/auto-libm-test-in: Add more tests of acos, acosh, asin,
asinh, atan, atan2, atanh, cbrt, cos and cosh.
* math/auto-libm-test-out: Regenerated.
* math/libm-test.inc (acosh_test_data): Add more tests.
(atanh_test_data): Likewise.
(ceil_test_data): Likewise.
(copysign_test_data): Likewise.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2013-12-20 21:03:39 +00:00
|
|
|
|
2014-05-14 12:35:40 +00:00
|
|
|
Function: "acosh_downward":
|
2015-02-25 00:01:15 +00:00
|
|
|
double: 1
|
2014-05-14 12:35:40 +00:00
|
|
|
float: 1
|
2015-02-25 00:01:15 +00:00
|
|
|
idouble: 1
|
|
|
|
|
ifloat: 1
|
|
|
|
|
ildouble: 2
|
2014-05-14 12:35:40 +00:00
|
|
|
ldouble: 2
|
|
|
|
|
|
|
|
|
|
Function: "acosh_towardzero":
|
2015-02-25 00:01:15 +00:00
|
|
|
double: 1
|
2014-05-14 12:35:40 +00:00
|
|
|
float: 1
|
2015-02-25 00:01:15 +00:00
|
|
|
idouble: 1
|
|
|
|
|
ifloat: 1
|
|
|
|
|
ildouble: 2
|
2014-05-14 12:35:40 +00:00
|
|
|
ldouble: 2
|
|
|
|
|
|
|
|
|
|
Function: "acosh_upward":
|
|
|
|
|
double: 1
|
2015-02-25 00:01:15 +00:00
|
|
|
float: 1
|
|
|
|
|
idouble: 1
|
|
|
|
|
ifloat: 1
|
2014-05-14 12:35:40 +00:00
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
|
|
|
|
|
2015-03-25 17:53:58 +00:00
|
|
|
Function: "asin":
|
|
|
|
|
float: 1
|
|
|
|
|
ifloat: 1
|
|
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
|
|
|
|
|
2012-04-30 09:38:06 +00:00
|
|
|
Function: "asin_downward":
|
|
|
|
|
double: 1
|
|
|
|
|
float: 1
|
|
|
|
|
idouble: 1
|
|
|
|
|
ifloat: 1
|
|
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
|
|
|
|
|
|
|
|
|
Function: "asin_towardzero":
|
2015-03-25 17:53:58 +00:00
|
|
|
double: 1
|
2012-04-30 09:38:06 +00:00
|
|
|
float: 1
|
2015-03-25 17:53:58 +00:00
|
|
|
idouble: 1
|
2012-04-30 09:38:06 +00:00
|
|
|
ifloat: 1
|
2015-03-25 17:53:58 +00:00
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
2012-04-30 09:38:06 +00:00
|
|
|
|
|
|
|
|
Function: "asin_upward":
|
2013-11-30 22:04:13 +00:00
|
|
|
double: 1
|
2012-04-30 09:38:06 +00:00
|
|
|
float: 1
|
2013-11-30 22:04:13 +00:00
|
|
|
idouble: 1
|
2012-04-30 09:38:06 +00:00
|
|
|
ifloat: 1
|
|
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
|
|
|
|
|
Add more libm-test coverage of [a-c]* real functions.
Various libm functions have inadequate test coverage in libm-test.inc
/ auto-libm-test-in - failing to cover all the usual special cases
(infinities, NaNs, zero, large and small finite values, subnormals) as
well as a reasonable range of ordinary inputs and, where appropriate,
inputs close to the thresholds for underflow and overflow.
This patch improves test coverage for real functions [a-c]* (with the
expectation of adding more coverage for other functions later).
Tested x86_64 and x86 and ulps updated accordingly (and eight glibc
bugs and one C11 DR filed for issues found in the process).
* math/auto-libm-test-in: Add more tests of acos, acosh, asin,
asinh, atan, atan2, atanh, cbrt, cos and cosh.
* math/auto-libm-test-out: Regenerated.
* math/libm-test.inc (acosh_test_data): Add more tests.
(atanh_test_data): Likewise.
(ceil_test_data): Likewise.
(copysign_test_data): Likewise.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2013-12-20 21:03:39 +00:00
|
|
|
Function: "asinh":
|
|
|
|
|
double: 1
|
|
|
|
|
float: 1
|
2015-02-25 11:13:41 +00:00
|
|
|
idouble: 1
|
Add more libm-test coverage of [a-c]* real functions.
Various libm functions have inadequate test coverage in libm-test.inc
/ auto-libm-test-in - failing to cover all the usual special cases
(infinities, NaNs, zero, large and small finite values, subnormals) as
well as a reasonable range of ordinary inputs and, where appropriate,
inputs close to the thresholds for underflow and overflow.
This patch improves test coverage for real functions [a-c]* (with the
expectation of adding more coverage for other functions later).
Tested x86_64 and x86 and ulps updated accordingly (and eight glibc
bugs and one C11 DR filed for issues found in the process).
* math/auto-libm-test-in: Add more tests of acos, acosh, asin,
asinh, atan, atan2, atanh, cbrt, cos and cosh.
* math/auto-libm-test-out: Regenerated.
* math/libm-test.inc (acosh_test_data): Add more tests.
(atanh_test_data): Likewise.
(ceil_test_data): Likewise.
(copysign_test_data): Likewise.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2013-12-20 21:03:39 +00:00
|
|
|
ifloat: 1
|
|
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
|
|
|
|
|
Test most libm functions in all rounding modes.
This patch makes libm-test.inc tests of most functions use ALL_RM_TEST
unless there was some reason to defer that change for a particular
function.
I started out planning to defer the change for pow (bug 16315), cexp /
ccos / ccosh / csin / csinh (likely fallout from exp, bug 16284) and
cpow (exact expectations for signs of exact zero results not wanted).
Testing on x86_64 and x86 showed additional failures for acosh, cacos,
catan, catanh, clog, clog10, jn, log, log10, log1p, log2, tgamma, yn,
so making the change for those functions was deferred as well, pending
investigation to show which of these represent distinct bugs (some
such bugs may already be filed) and appropriate fixing / XFAILing.
Failures include wrong signs of zero results, errors slightly above
the 9ulp bound (in such cases it may make sense for functions to set
round-to-nearest internally to reduce error accumulation), large
errors and incorrect overflow/underflow for the rounding mode (with
consequent missing errno settings in some cases). It's possible some
could be issues with test expectations, though I didn't notice any
that were obviously like that (I added NO_TEST_INLINE for cases that
were failing for ildoubl on x86 and where it seemed reasonable for
them to fail for the fast-math inlines).
There may of course be failures on other architectures for functions
that didn't fail on x86_64 or x86, in which case the usual rule
applies: file a bug (preferably identifying the underlying problem
function, in cases where function A calls function B and a problem
with function B may present in the test results for function A) if not
already in Bugzilla then fix or XFAIL.
Tested x86_64 and x86 and ulps updated accordingly.
* math/libm-test.inc (asinh_test): Use ALL_RM_TEST.
(atan_test): Likewise.
(atanh_test_data): Use NO_TEST_INLINE for two tests.
(atanh_test): Use ALL_RM_TEST.
(atan2_test_data): Likewise.
(cabs_test): Likewise.
(cacosh_test): Likewise.
(carg_test): Likewise.
(casin_test): Likewise.
(casinh_test): Likewise.
(cbrt_test): Likewise.
(csqrt_test): Likewise.
(erf_test): Likewise.
(erfc_test): Likewise.
(pow10_test): Likewise.
(exp2_test): Likewise.
(hypot_test): Likewise.
(j0_test): Likewise.
(j1_test): Likewise.
(lgamma_test): Likewise.
(gamma_test): Likewise.
(sincos_test): Likewise.
(tanh_test): Likewise.
(y0_test): Likewise.
(y1_test): Likewise.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2014-03-21 00:03:38 +00:00
|
|
|
Function: "asinh_downward":
|
2015-02-25 11:13:41 +00:00
|
|
|
double: 2
|
Test most libm functions in all rounding modes.
This patch makes libm-test.inc tests of most functions use ALL_RM_TEST
unless there was some reason to defer that change for a particular
function.
I started out planning to defer the change for pow (bug 16315), cexp /
ccos / ccosh / csin / csinh (likely fallout from exp, bug 16284) and
cpow (exact expectations for signs of exact zero results not wanted).
Testing on x86_64 and x86 showed additional failures for acosh, cacos,
catan, catanh, clog, clog10, jn, log, log10, log1p, log2, tgamma, yn,
so making the change for those functions was deferred as well, pending
investigation to show which of these represent distinct bugs (some
such bugs may already be filed) and appropriate fixing / XFAILing.
Failures include wrong signs of zero results, errors slightly above
the 9ulp bound (in such cases it may make sense for functions to set
round-to-nearest internally to reduce error accumulation), large
errors and incorrect overflow/underflow for the rounding mode (with
consequent missing errno settings in some cases). It's possible some
could be issues with test expectations, though I didn't notice any
that were obviously like that (I added NO_TEST_INLINE for cases that
were failing for ildoubl on x86 and where it seemed reasonable for
them to fail for the fast-math inlines).
There may of course be failures on other architectures for functions
that didn't fail on x86_64 or x86, in which case the usual rule
applies: file a bug (preferably identifying the underlying problem
function, in cases where function A calls function B and a problem
with function B may present in the test results for function A) if not
already in Bugzilla then fix or XFAIL.
Tested x86_64 and x86 and ulps updated accordingly.
* math/libm-test.inc (asinh_test): Use ALL_RM_TEST.
(atan_test): Likewise.
(atanh_test_data): Use NO_TEST_INLINE for two tests.
(atanh_test): Use ALL_RM_TEST.
(atan2_test_data): Likewise.
(cabs_test): Likewise.
(cacosh_test): Likewise.
(carg_test): Likewise.
(casin_test): Likewise.
(casinh_test): Likewise.
(cbrt_test): Likewise.
(csqrt_test): Likewise.
(erf_test): Likewise.
(erfc_test): Likewise.
(pow10_test): Likewise.
(exp2_test): Likewise.
(hypot_test): Likewise.
(j0_test): Likewise.
(j1_test): Likewise.
(lgamma_test): Likewise.
(gamma_test): Likewise.
(sincos_test): Likewise.
(tanh_test): Likewise.
(y0_test): Likewise.
(y1_test): Likewise.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2014-03-21 00:03:38 +00:00
|
|
|
float: 2
|
2015-02-25 11:13:41 +00:00
|
|
|
idouble: 2
|
|
|
|
|
ifloat: 2
|
|
|
|
|
ildouble: 3
|
Test most libm functions in all rounding modes.
This patch makes libm-test.inc tests of most functions use ALL_RM_TEST
unless there was some reason to defer that change for a particular
function.
I started out planning to defer the change for pow (bug 16315), cexp /
ccos / ccosh / csin / csinh (likely fallout from exp, bug 16284) and
cpow (exact expectations for signs of exact zero results not wanted).
Testing on x86_64 and x86 showed additional failures for acosh, cacos,
catan, catanh, clog, clog10, jn, log, log10, log1p, log2, tgamma, yn,
so making the change for those functions was deferred as well, pending
investigation to show which of these represent distinct bugs (some
such bugs may already be filed) and appropriate fixing / XFAILing.
Failures include wrong signs of zero results, errors slightly above
the 9ulp bound (in such cases it may make sense for functions to set
round-to-nearest internally to reduce error accumulation), large
errors and incorrect overflow/underflow for the rounding mode (with
consequent missing errno settings in some cases). It's possible some
could be issues with test expectations, though I didn't notice any
that were obviously like that (I added NO_TEST_INLINE for cases that
were failing for ildoubl on x86 and where it seemed reasonable for
them to fail for the fast-math inlines).
There may of course be failures on other architectures for functions
that didn't fail on x86_64 or x86, in which case the usual rule
applies: file a bug (preferably identifying the underlying problem
function, in cases where function A calls function B and a problem
with function B may present in the test results for function A) if not
already in Bugzilla then fix or XFAIL.
Tested x86_64 and x86 and ulps updated accordingly.
* math/libm-test.inc (asinh_test): Use ALL_RM_TEST.
(atan_test): Likewise.
(atanh_test_data): Use NO_TEST_INLINE for two tests.
(atanh_test): Use ALL_RM_TEST.
(atan2_test_data): Likewise.
(cabs_test): Likewise.
(cacosh_test): Likewise.
(carg_test): Likewise.
(casin_test): Likewise.
(casinh_test): Likewise.
(cbrt_test): Likewise.
(csqrt_test): Likewise.
(erf_test): Likewise.
(erfc_test): Likewise.
(pow10_test): Likewise.
(exp2_test): Likewise.
(hypot_test): Likewise.
(j0_test): Likewise.
(j1_test): Likewise.
(lgamma_test): Likewise.
(gamma_test): Likewise.
(sincos_test): Likewise.
(tanh_test): Likewise.
(y0_test): Likewise.
(y1_test): Likewise.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2014-03-21 00:03:38 +00:00
|
|
|
ldouble: 3
|
|
|
|
|
|
|
|
|
|
Function: "asinh_towardzero":
|
2015-02-25 11:13:41 +00:00
|
|
|
double: 2
|
|
|
|
|
float: 2
|
|
|
|
|
idouble: 2
|
|
|
|
|
ifloat: 2
|
|
|
|
|
ildouble: 3
|
|
|
|
|
ldouble: 3
|
Test most libm functions in all rounding modes.
This patch makes libm-test.inc tests of most functions use ALL_RM_TEST
unless there was some reason to defer that change for a particular
function.
I started out planning to defer the change for pow (bug 16315), cexp /
ccos / ccosh / csin / csinh (likely fallout from exp, bug 16284) and
cpow (exact expectations for signs of exact zero results not wanted).
Testing on x86_64 and x86 showed additional failures for acosh, cacos,
catan, catanh, clog, clog10, jn, log, log10, log1p, log2, tgamma, yn,
so making the change for those functions was deferred as well, pending
investigation to show which of these represent distinct bugs (some
such bugs may already be filed) and appropriate fixing / XFAILing.
Failures include wrong signs of zero results, errors slightly above
the 9ulp bound (in such cases it may make sense for functions to set
round-to-nearest internally to reduce error accumulation), large
errors and incorrect overflow/underflow for the rounding mode (with
consequent missing errno settings in some cases). It's possible some
could be issues with test expectations, though I didn't notice any
that were obviously like that (I added NO_TEST_INLINE for cases that
were failing for ildoubl on x86 and where it seemed reasonable for
them to fail for the fast-math inlines).
There may of course be failures on other architectures for functions
that didn't fail on x86_64 or x86, in which case the usual rule
applies: file a bug (preferably identifying the underlying problem
function, in cases where function A calls function B and a problem
with function B may present in the test results for function A) if not
already in Bugzilla then fix or XFAIL.
Tested x86_64 and x86 and ulps updated accordingly.
* math/libm-test.inc (asinh_test): Use ALL_RM_TEST.
(atan_test): Likewise.
(atanh_test_data): Use NO_TEST_INLINE for two tests.
(atanh_test): Use ALL_RM_TEST.
(atan2_test_data): Likewise.
(cabs_test): Likewise.
(cacosh_test): Likewise.
(carg_test): Likewise.
(casin_test): Likewise.
(casinh_test): Likewise.
(cbrt_test): Likewise.
(csqrt_test): Likewise.
(erf_test): Likewise.
(erfc_test): Likewise.
(pow10_test): Likewise.
(exp2_test): Likewise.
(hypot_test): Likewise.
(j0_test): Likewise.
(j1_test): Likewise.
(lgamma_test): Likewise.
(gamma_test): Likewise.
(sincos_test): Likewise.
(tanh_test): Likewise.
(y0_test): Likewise.
(y1_test): Likewise.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2014-03-21 00:03:38 +00:00
|
|
|
|
|
|
|
|
Function: "asinh_upward":
|
|
|
|
|
double: 2
|
|
|
|
|
float: 1
|
2015-02-25 11:13:41 +00:00
|
|
|
idouble: 2
|
Test most libm functions in all rounding modes.
This patch makes libm-test.inc tests of most functions use ALL_RM_TEST
unless there was some reason to defer that change for a particular
function.
I started out planning to defer the change for pow (bug 16315), cexp /
ccos / ccosh / csin / csinh (likely fallout from exp, bug 16284) and
cpow (exact expectations for signs of exact zero results not wanted).
Testing on x86_64 and x86 showed additional failures for acosh, cacos,
catan, catanh, clog, clog10, jn, log, log10, log1p, log2, tgamma, yn,
so making the change for those functions was deferred as well, pending
investigation to show which of these represent distinct bugs (some
such bugs may already be filed) and appropriate fixing / XFAILing.
Failures include wrong signs of zero results, errors slightly above
the 9ulp bound (in such cases it may make sense for functions to set
round-to-nearest internally to reduce error accumulation), large
errors and incorrect overflow/underflow for the rounding mode (with
consequent missing errno settings in some cases). It's possible some
could be issues with test expectations, though I didn't notice any
that were obviously like that (I added NO_TEST_INLINE for cases that
were failing for ildoubl on x86 and where it seemed reasonable for
them to fail for the fast-math inlines).
There may of course be failures on other architectures for functions
that didn't fail on x86_64 or x86, in which case the usual rule
applies: file a bug (preferably identifying the underlying problem
function, in cases where function A calls function B and a problem
with function B may present in the test results for function A) if not
already in Bugzilla then fix or XFAIL.
Tested x86_64 and x86 and ulps updated accordingly.
* math/libm-test.inc (asinh_test): Use ALL_RM_TEST.
(atan_test): Likewise.
(atanh_test_data): Use NO_TEST_INLINE for two tests.
(atanh_test): Use ALL_RM_TEST.
(atan2_test_data): Likewise.
(cabs_test): Likewise.
(cacosh_test): Likewise.
(carg_test): Likewise.
(casin_test): Likewise.
(casinh_test): Likewise.
(cbrt_test): Likewise.
(csqrt_test): Likewise.
(erf_test): Likewise.
(erfc_test): Likewise.
(pow10_test): Likewise.
(exp2_test): Likewise.
(hypot_test): Likewise.
(j0_test): Likewise.
(j1_test): Likewise.
(lgamma_test): Likewise.
(gamma_test): Likewise.
(sincos_test): Likewise.
(tanh_test): Likewise.
(y0_test): Likewise.
(y1_test): Likewise.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2014-03-21 00:03:38 +00:00
|
|
|
ifloat: 1
|
2015-02-25 11:13:41 +00:00
|
|
|
ildouble: 3
|
|
|
|
|
ldouble: 3
|
Test most libm functions in all rounding modes.
This patch makes libm-test.inc tests of most functions use ALL_RM_TEST
unless there was some reason to defer that change for a particular
function.
I started out planning to defer the change for pow (bug 16315), cexp /
ccos / ccosh / csin / csinh (likely fallout from exp, bug 16284) and
cpow (exact expectations for signs of exact zero results not wanted).
Testing on x86_64 and x86 showed additional failures for acosh, cacos,
catan, catanh, clog, clog10, jn, log, log10, log1p, log2, tgamma, yn,
so making the change for those functions was deferred as well, pending
investigation to show which of these represent distinct bugs (some
such bugs may already be filed) and appropriate fixing / XFAILing.
Failures include wrong signs of zero results, errors slightly above
the 9ulp bound (in such cases it may make sense for functions to set
round-to-nearest internally to reduce error accumulation), large
errors and incorrect overflow/underflow for the rounding mode (with
consequent missing errno settings in some cases). It's possible some
could be issues with test expectations, though I didn't notice any
that were obviously like that (I added NO_TEST_INLINE for cases that
were failing for ildoubl on x86 and where it seemed reasonable for
them to fail for the fast-math inlines).
There may of course be failures on other architectures for functions
that didn't fail on x86_64 or x86, in which case the usual rule
applies: file a bug (preferably identifying the underlying problem
function, in cases where function A calls function B and a problem
with function B may present in the test results for function A) if not
already in Bugzilla then fix or XFAIL.
Tested x86_64 and x86 and ulps updated accordingly.
* math/libm-test.inc (asinh_test): Use ALL_RM_TEST.
(atan_test): Likewise.
(atanh_test_data): Use NO_TEST_INLINE for two tests.
(atanh_test): Use ALL_RM_TEST.
(atan2_test_data): Likewise.
(cabs_test): Likewise.
(cacosh_test): Likewise.
(carg_test): Likewise.
(casin_test): Likewise.
(casinh_test): Likewise.
(cbrt_test): Likewise.
(csqrt_test): Likewise.
(erf_test): Likewise.
(erfc_test): Likewise.
(pow10_test): Likewise.
(exp2_test): Likewise.
(hypot_test): Likewise.
(j0_test): Likewise.
(j1_test): Likewise.
(lgamma_test): Likewise.
(gamma_test): Likewise.
(sincos_test): Likewise.
(tanh_test): Likewise.
(y0_test): Likewise.
(y1_test): Likewise.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2014-03-21 00:03:38 +00:00
|
|
|
|
2001-09-19 10:37:31 +00:00
|
|
|
Function: "atan2":
|
2005-07-20 18:20:48 +00:00
|
|
|
float: 1
|
|
|
|
|
ifloat: 1
|
2013-12-16 21:18:07 +00:00
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
2001-09-19 10:37:31 +00:00
|
|
|
|
Test most libm functions in all rounding modes.
This patch makes libm-test.inc tests of most functions use ALL_RM_TEST
unless there was some reason to defer that change for a particular
function.
I started out planning to defer the change for pow (bug 16315), cexp /
ccos / ccosh / csin / csinh (likely fallout from exp, bug 16284) and
cpow (exact expectations for signs of exact zero results not wanted).
Testing on x86_64 and x86 showed additional failures for acosh, cacos,
catan, catanh, clog, clog10, jn, log, log10, log1p, log2, tgamma, yn,
so making the change for those functions was deferred as well, pending
investigation to show which of these represent distinct bugs (some
such bugs may already be filed) and appropriate fixing / XFAILing.
Failures include wrong signs of zero results, errors slightly above
the 9ulp bound (in such cases it may make sense for functions to set
round-to-nearest internally to reduce error accumulation), large
errors and incorrect overflow/underflow for the rounding mode (with
consequent missing errno settings in some cases). It's possible some
could be issues with test expectations, though I didn't notice any
that were obviously like that (I added NO_TEST_INLINE for cases that
were failing for ildoubl on x86 and where it seemed reasonable for
them to fail for the fast-math inlines).
There may of course be failures on other architectures for functions
that didn't fail on x86_64 or x86, in which case the usual rule
applies: file a bug (preferably identifying the underlying problem
function, in cases where function A calls function B and a problem
with function B may present in the test results for function A) if not
already in Bugzilla then fix or XFAIL.
Tested x86_64 and x86 and ulps updated accordingly.
* math/libm-test.inc (asinh_test): Use ALL_RM_TEST.
(atan_test): Likewise.
(atanh_test_data): Use NO_TEST_INLINE for two tests.
(atanh_test): Use ALL_RM_TEST.
(atan2_test_data): Likewise.
(cabs_test): Likewise.
(cacosh_test): Likewise.
(carg_test): Likewise.
(casin_test): Likewise.
(casinh_test): Likewise.
(cbrt_test): Likewise.
(csqrt_test): Likewise.
(erf_test): Likewise.
(erfc_test): Likewise.
(pow10_test): Likewise.
(exp2_test): Likewise.
(hypot_test): Likewise.
(j0_test): Likewise.
(j1_test): Likewise.
(lgamma_test): Likewise.
(gamma_test): Likewise.
(sincos_test): Likewise.
(tanh_test): Likewise.
(y0_test): Likewise.
(y1_test): Likewise.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2014-03-21 00:03:38 +00:00
|
|
|
Function: "atan2_downward":
|
|
|
|
|
double: 1
|
|
|
|
|
float: 2
|
|
|
|
|
idouble: 1
|
|
|
|
|
ifloat: 2
|
|
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
|
|
|
|
|
|
|
|
|
Function: "atan2_towardzero":
|
|
|
|
|
double: 1
|
|
|
|
|
float: 2
|
|
|
|
|
idouble: 1
|
|
|
|
|
ifloat: 2
|
|
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
|
|
|
|
|
|
|
|
|
Function: "atan2_upward":
|
|
|
|
|
double: 1
|
|
|
|
|
float: 2
|
|
|
|
|
idouble: 1
|
|
|
|
|
ifloat: 2
|
|
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
|
|
|
|
|
|
|
|
|
Function: "atan_downward":
|
|
|
|
|
double: 1
|
|
|
|
|
float: 1
|
|
|
|
|
idouble: 1
|
|
|
|
|
ifloat: 1
|
|
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
|
|
|
|
|
|
|
|
|
Function: "atan_towardzero":
|
|
|
|
|
double: 1
|
|
|
|
|
float: 1
|
|
|
|
|
idouble: 1
|
|
|
|
|
ifloat: 1
|
|
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
|
|
|
|
|
|
|
|
|
Function: "atan_upward":
|
|
|
|
|
double: 1
|
|
|
|
|
float: 1
|
|
|
|
|
idouble: 1
|
|
|
|
|
ifloat: 1
|
|
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
|
|
|
|
|
2001-09-19 10:37:31 +00:00
|
|
|
Function: "atanh":
|
2015-02-27 17:48:37 +00:00
|
|
|
double: 1
|
2002-08-31 08:22:35 +00:00
|
|
|
float: 1
|
2015-02-27 17:48:37 +00:00
|
|
|
idouble: 1
|
2001-09-19 10:37:31 +00:00
|
|
|
ifloat: 1
|
|
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
|
|
|
|
|
Test most libm functions in all rounding modes.
This patch makes libm-test.inc tests of most functions use ALL_RM_TEST
unless there was some reason to defer that change for a particular
function.
I started out planning to defer the change for pow (bug 16315), cexp /
ccos / ccosh / csin / csinh (likely fallout from exp, bug 16284) and
cpow (exact expectations for signs of exact zero results not wanted).
Testing on x86_64 and x86 showed additional failures for acosh, cacos,
catan, catanh, clog, clog10, jn, log, log10, log1p, log2, tgamma, yn,
so making the change for those functions was deferred as well, pending
investigation to show which of these represent distinct bugs (some
such bugs may already be filed) and appropriate fixing / XFAILing.
Failures include wrong signs of zero results, errors slightly above
the 9ulp bound (in such cases it may make sense for functions to set
round-to-nearest internally to reduce error accumulation), large
errors and incorrect overflow/underflow for the rounding mode (with
consequent missing errno settings in some cases). It's possible some
could be issues with test expectations, though I didn't notice any
that were obviously like that (I added NO_TEST_INLINE for cases that
were failing for ildoubl on x86 and where it seemed reasonable for
them to fail for the fast-math inlines).
There may of course be failures on other architectures for functions
that didn't fail on x86_64 or x86, in which case the usual rule
applies: file a bug (preferably identifying the underlying problem
function, in cases where function A calls function B and a problem
with function B may present in the test results for function A) if not
already in Bugzilla then fix or XFAIL.
Tested x86_64 and x86 and ulps updated accordingly.
* math/libm-test.inc (asinh_test): Use ALL_RM_TEST.
(atan_test): Likewise.
(atanh_test_data): Use NO_TEST_INLINE for two tests.
(atanh_test): Use ALL_RM_TEST.
(atan2_test_data): Likewise.
(cabs_test): Likewise.
(cacosh_test): Likewise.
(carg_test): Likewise.
(casin_test): Likewise.
(casinh_test): Likewise.
(cbrt_test): Likewise.
(csqrt_test): Likewise.
(erf_test): Likewise.
(erfc_test): Likewise.
(pow10_test): Likewise.
(exp2_test): Likewise.
(hypot_test): Likewise.
(j0_test): Likewise.
(j1_test): Likewise.
(lgamma_test): Likewise.
(gamma_test): Likewise.
(sincos_test): Likewise.
(tanh_test): Likewise.
(y0_test): Likewise.
(y1_test): Likewise.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2014-03-21 00:03:38 +00:00
|
|
|
Function: "atanh_downward":
|
|
|
|
|
double: 1
|
|
|
|
|
float: 1
|
|
|
|
|
idouble: 1
|
|
|
|
|
ifloat: 1
|
|
|
|
|
ildouble: 2
|
|
|
|
|
ldouble: 2
|
|
|
|
|
|
|
|
|
|
Function: "atanh_towardzero":
|
2015-02-27 17:48:37 +00:00
|
|
|
double: 1
|
Test most libm functions in all rounding modes.
This patch makes libm-test.inc tests of most functions use ALL_RM_TEST
unless there was some reason to defer that change for a particular
function.
I started out planning to defer the change for pow (bug 16315), cexp /
ccos / ccosh / csin / csinh (likely fallout from exp, bug 16284) and
cpow (exact expectations for signs of exact zero results not wanted).
Testing on x86_64 and x86 showed additional failures for acosh, cacos,
catan, catanh, clog, clog10, jn, log, log10, log1p, log2, tgamma, yn,
so making the change for those functions was deferred as well, pending
investigation to show which of these represent distinct bugs (some
such bugs may already be filed) and appropriate fixing / XFAILing.
Failures include wrong signs of zero results, errors slightly above
the 9ulp bound (in such cases it may make sense for functions to set
round-to-nearest internally to reduce error accumulation), large
errors and incorrect overflow/underflow for the rounding mode (with
consequent missing errno settings in some cases). It's possible some
could be issues with test expectations, though I didn't notice any
that were obviously like that (I added NO_TEST_INLINE for cases that
were failing for ildoubl on x86 and where it seemed reasonable for
them to fail for the fast-math inlines).
There may of course be failures on other architectures for functions
that didn't fail on x86_64 or x86, in which case the usual rule
applies: file a bug (preferably identifying the underlying problem
function, in cases where function A calls function B and a problem
with function B may present in the test results for function A) if not
already in Bugzilla then fix or XFAIL.
Tested x86_64 and x86 and ulps updated accordingly.
* math/libm-test.inc (asinh_test): Use ALL_RM_TEST.
(atan_test): Likewise.
(atanh_test_data): Use NO_TEST_INLINE for two tests.
(atanh_test): Use ALL_RM_TEST.
(atan2_test_data): Likewise.
(cabs_test): Likewise.
(cacosh_test): Likewise.
(carg_test): Likewise.
(casin_test): Likewise.
(casinh_test): Likewise.
(cbrt_test): Likewise.
(csqrt_test): Likewise.
(erf_test): Likewise.
(erfc_test): Likewise.
(pow10_test): Likewise.
(exp2_test): Likewise.
(hypot_test): Likewise.
(j0_test): Likewise.
(j1_test): Likewise.
(lgamma_test): Likewise.
(gamma_test): Likewise.
(sincos_test): Likewise.
(tanh_test): Likewise.
(y0_test): Likewise.
(y1_test): Likewise.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2014-03-21 00:03:38 +00:00
|
|
|
float: 1
|
2015-02-27 17:48:37 +00:00
|
|
|
idouble: 1
|
Test most libm functions in all rounding modes.
This patch makes libm-test.inc tests of most functions use ALL_RM_TEST
unless there was some reason to defer that change for a particular
function.
I started out planning to defer the change for pow (bug 16315), cexp /
ccos / ccosh / csin / csinh (likely fallout from exp, bug 16284) and
cpow (exact expectations for signs of exact zero results not wanted).
Testing on x86_64 and x86 showed additional failures for acosh, cacos,
catan, catanh, clog, clog10, jn, log, log10, log1p, log2, tgamma, yn,
so making the change for those functions was deferred as well, pending
investigation to show which of these represent distinct bugs (some
such bugs may already be filed) and appropriate fixing / XFAILing.
Failures include wrong signs of zero results, errors slightly above
the 9ulp bound (in such cases it may make sense for functions to set
round-to-nearest internally to reduce error accumulation), large
errors and incorrect overflow/underflow for the rounding mode (with
consequent missing errno settings in some cases). It's possible some
could be issues with test expectations, though I didn't notice any
that were obviously like that (I added NO_TEST_INLINE for cases that
were failing for ildoubl on x86 and where it seemed reasonable for
them to fail for the fast-math inlines).
There may of course be failures on other architectures for functions
that didn't fail on x86_64 or x86, in which case the usual rule
applies: file a bug (preferably identifying the underlying problem
function, in cases where function A calls function B and a problem
with function B may present in the test results for function A) if not
already in Bugzilla then fix or XFAIL.
Tested x86_64 and x86 and ulps updated accordingly.
* math/libm-test.inc (asinh_test): Use ALL_RM_TEST.
(atan_test): Likewise.
(atanh_test_data): Use NO_TEST_INLINE for two tests.
(atanh_test): Use ALL_RM_TEST.
(atan2_test_data): Likewise.
(cabs_test): Likewise.
(cacosh_test): Likewise.
(carg_test): Likewise.
(casin_test): Likewise.
(casinh_test): Likewise.
(cbrt_test): Likewise.
(csqrt_test): Likewise.
(erf_test): Likewise.
(erfc_test): Likewise.
(pow10_test): Likewise.
(exp2_test): Likewise.
(hypot_test): Likewise.
(j0_test): Likewise.
(j1_test): Likewise.
(lgamma_test): Likewise.
(gamma_test): Likewise.
(sincos_test): Likewise.
(tanh_test): Likewise.
(y0_test): Likewise.
(y1_test): Likewise.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2014-03-21 00:03:38 +00:00
|
|
|
ifloat: 1
|
|
|
|
|
ildouble: 2
|
|
|
|
|
ldouble: 2
|
|
|
|
|
|
|
|
|
|
Function: "atanh_upward":
|
|
|
|
|
double: 1
|
|
|
|
|
float: 1
|
|
|
|
|
idouble: 1
|
|
|
|
|
ifloat: 1
|
|
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
|
|
|
|
|
2012-03-07 15:15:19 +00:00
|
|
|
Function: Real part of "cacos":
|
|
|
|
|
double: 1
|
2013-03-21 10:27:10 +00:00
|
|
|
float: 2
|
2012-03-07 15:15:19 +00:00
|
|
|
idouble: 1
|
2013-03-21 10:27:10 +00:00
|
|
|
ifloat: 2
|
2012-03-07 15:15:19 +00:00
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
|
|
|
|
|
2001-09-19 10:37:31 +00:00
|
|
|
Function: Imaginary part of "cacos":
|
2013-03-30 13:31:53 +00:00
|
|
|
double: 1
|
|
|
|
|
float: 2
|
|
|
|
|
idouble: 1
|
|
|
|
|
ifloat: 2
|
2002-08-31 08:22:35 +00:00
|
|
|
ildouble: 2
|
|
|
|
|
ldouble: 2
|
2001-09-19 10:37:31 +00:00
|
|
|
|
2014-05-14 12:37:24 +00:00
|
|
|
Function: Real part of "cacos_downward":
|
|
|
|
|
double: 1
|
|
|
|
|
float: 2
|
|
|
|
|
idouble: 1
|
|
|
|
|
ifloat: 2
|
|
|
|
|
ildouble: 2
|
|
|
|
|
ldouble: 2
|
|
|
|
|
|
|
|
|
|
Function: Imaginary part of "cacos_downward":
|
|
|
|
|
double: 5
|
|
|
|
|
float: 3
|
|
|
|
|
idouble: 5
|
|
|
|
|
ifloat: 3
|
|
|
|
|
ildouble: 5
|
|
|
|
|
ldouble: 5
|
|
|
|
|
|
|
|
|
|
Function: Real part of "cacos_towardzero":
|
|
|
|
|
double: 1
|
|
|
|
|
float: 2
|
|
|
|
|
idouble: 1
|
|
|
|
|
ifloat: 2
|
|
|
|
|
ildouble: 2
|
|
|
|
|
ldouble: 2
|
|
|
|
|
|
|
|
|
|
Function: Imaginary part of "cacos_towardzero":
|
|
|
|
|
double: 5
|
|
|
|
|
float: 3
|
|
|
|
|
idouble: 5
|
|
|
|
|
ifloat: 3
|
|
|
|
|
ildouble: 5
|
|
|
|
|
ldouble: 5
|
|
|
|
|
|
|
|
|
|
Function: Real part of "cacos_upward":
|
|
|
|
|
double: 2
|
|
|
|
|
float: 2
|
|
|
|
|
idouble: 2
|
|
|
|
|
ifloat: 2
|
|
|
|
|
ildouble: 2
|
|
|
|
|
ldouble: 2
|
|
|
|
|
|
|
|
|
|
Function: Imaginary part of "cacos_upward":
|
|
|
|
|
double: 4
|
|
|
|
|
float: 4
|
|
|
|
|
idouble: 4
|
|
|
|
|
ifloat: 4
|
|
|
|
|
ildouble: 5
|
|
|
|
|
ldouble: 5
|
|
|
|
|
|
2001-09-19 10:37:31 +00:00
|
|
|
Function: Real part of "cacosh":
|
|
|
|
|
double: 1
|
2013-04-02 22:54:00 +00:00
|
|
|
float: 2
|
2001-09-19 10:37:31 +00:00
|
|
|
idouble: 1
|
2013-04-02 22:54:00 +00:00
|
|
|
ifloat: 2
|
|
|
|
|
ildouble: 2
|
|
|
|
|
ldouble: 2
|
2001-09-19 10:37:31 +00:00
|
|
|
|
|
|
|
|
Function: Imaginary part of "cacosh":
|
|
|
|
|
double: 1
|
2013-04-02 22:54:00 +00:00
|
|
|
float: 2
|
2001-09-19 10:37:31 +00:00
|
|
|
idouble: 1
|
2013-04-02 22:54:00 +00:00
|
|
|
ifloat: 2
|
2001-09-19 10:37:31 +00:00
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
|
|
|
|
|
Test most libm functions in all rounding modes.
This patch makes libm-test.inc tests of most functions use ALL_RM_TEST
unless there was some reason to defer that change for a particular
function.
I started out planning to defer the change for pow (bug 16315), cexp /
ccos / ccosh / csin / csinh (likely fallout from exp, bug 16284) and
cpow (exact expectations for signs of exact zero results not wanted).
Testing on x86_64 and x86 showed additional failures for acosh, cacos,
catan, catanh, clog, clog10, jn, log, log10, log1p, log2, tgamma, yn,
so making the change for those functions was deferred as well, pending
investigation to show which of these represent distinct bugs (some
such bugs may already be filed) and appropriate fixing / XFAILing.
Failures include wrong signs of zero results, errors slightly above
the 9ulp bound (in such cases it may make sense for functions to set
round-to-nearest internally to reduce error accumulation), large
errors and incorrect overflow/underflow for the rounding mode (with
consequent missing errno settings in some cases). It's possible some
could be issues with test expectations, though I didn't notice any
that were obviously like that (I added NO_TEST_INLINE for cases that
were failing for ildoubl on x86 and where it seemed reasonable for
them to fail for the fast-math inlines).
There may of course be failures on other architectures for functions
that didn't fail on x86_64 or x86, in which case the usual rule
applies: file a bug (preferably identifying the underlying problem
function, in cases where function A calls function B and a problem
with function B may present in the test results for function A) if not
already in Bugzilla then fix or XFAIL.
Tested x86_64 and x86 and ulps updated accordingly.
* math/libm-test.inc (asinh_test): Use ALL_RM_TEST.
(atan_test): Likewise.
(atanh_test_data): Use NO_TEST_INLINE for two tests.
(atanh_test): Use ALL_RM_TEST.
(atan2_test_data): Likewise.
(cabs_test): Likewise.
(cacosh_test): Likewise.
(carg_test): Likewise.
(casin_test): Likewise.
(casinh_test): Likewise.
(cbrt_test): Likewise.
(csqrt_test): Likewise.
(erf_test): Likewise.
(erfc_test): Likewise.
(pow10_test): Likewise.
(exp2_test): Likewise.
(hypot_test): Likewise.
(j0_test): Likewise.
(j1_test): Likewise.
(lgamma_test): Likewise.
(gamma_test): Likewise.
(sincos_test): Likewise.
(tanh_test): Likewise.
(y0_test): Likewise.
(y1_test): Likewise.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2014-03-21 00:03:38 +00:00
|
|
|
Function: Real part of "cacosh_downward":
|
|
|
|
|
double: 5
|
|
|
|
|
float: 3
|
|
|
|
|
idouble: 5
|
|
|
|
|
ifloat: 3
|
|
|
|
|
ildouble: 5
|
|
|
|
|
ldouble: 5
|
|
|
|
|
|
|
|
|
|
Function: Imaginary part of "cacosh_downward":
|
|
|
|
|
double: 1
|
|
|
|
|
float: 2
|
|
|
|
|
idouble: 1
|
|
|
|
|
ifloat: 2
|
|
|
|
|
ildouble: 2
|
|
|
|
|
ldouble: 2
|
|
|
|
|
|
|
|
|
|
Function: Real part of "cacosh_towardzero":
|
|
|
|
|
double: 5
|
|
|
|
|
float: 3
|
|
|
|
|
idouble: 5
|
|
|
|
|
ifloat: 3
|
|
|
|
|
ildouble: 5
|
|
|
|
|
ldouble: 5
|
|
|
|
|
|
|
|
|
|
Function: Imaginary part of "cacosh_towardzero":
|
|
|
|
|
double: 1
|
|
|
|
|
float: 2
|
|
|
|
|
idouble: 1
|
|
|
|
|
ifloat: 2
|
|
|
|
|
ildouble: 2
|
|
|
|
|
ldouble: 2
|
|
|
|
|
|
|
|
|
|
Function: Real part of "cacosh_upward":
|
|
|
|
|
double: 4
|
|
|
|
|
float: 4
|
|
|
|
|
idouble: 4
|
|
|
|
|
ifloat: 4
|
|
|
|
|
ildouble: 5
|
|
|
|
|
ldouble: 5
|
|
|
|
|
|
|
|
|
|
Function: Imaginary part of "cacosh_upward":
|
|
|
|
|
double: 2
|
|
|
|
|
float: 2
|
|
|
|
|
idouble: 2
|
|
|
|
|
ifloat: 2
|
|
|
|
|
ildouble: 2
|
|
|
|
|
ldouble: 2
|
|
|
|
|
|
|
|
|
|
Function: "carg_downward":
|
|
|
|
|
double: 1
|
|
|
|
|
float: 2
|
|
|
|
|
idouble: 1
|
|
|
|
|
ifloat: 2
|
|
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
|
|
|
|
|
|
|
|
|
Function: "carg_towardzero":
|
|
|
|
|
float: 1
|
|
|
|
|
ifloat: 1
|
|
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
|
|
|
|
|
|
|
|
|
Function: "carg_upward":
|
|
|
|
|
double: 1
|
|
|
|
|
float: 2
|
|
|
|
|
idouble: 1
|
|
|
|
|
ifloat: 2
|
|
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
|
|
|
|
|
2001-09-19 10:37:31 +00:00
|
|
|
Function: Real part of "casin":
|
2002-12-03 15:39:37 +00:00
|
|
|
double: 1
|
|
|
|
|
float: 1
|
|
|
|
|
idouble: 1
|
2002-08-31 08:22:35 +00:00
|
|
|
ifloat: 1
|
2013-03-30 13:31:53 +00:00
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
2001-09-19 10:37:31 +00:00
|
|
|
|
|
|
|
|
Function: Imaginary part of "casin":
|
2013-03-30 13:31:53 +00:00
|
|
|
double: 1
|
|
|
|
|
float: 2
|
|
|
|
|
idouble: 1
|
|
|
|
|
ifloat: 2
|
2002-08-31 08:22:35 +00:00
|
|
|
ildouble: 2
|
|
|
|
|
ldouble: 2
|
2001-09-19 10:37:31 +00:00
|
|
|
|
Test most libm functions in all rounding modes.
This patch makes libm-test.inc tests of most functions use ALL_RM_TEST
unless there was some reason to defer that change for a particular
function.
I started out planning to defer the change for pow (bug 16315), cexp /
ccos / ccosh / csin / csinh (likely fallout from exp, bug 16284) and
cpow (exact expectations for signs of exact zero results not wanted).
Testing on x86_64 and x86 showed additional failures for acosh, cacos,
catan, catanh, clog, clog10, jn, log, log10, log1p, log2, tgamma, yn,
so making the change for those functions was deferred as well, pending
investigation to show which of these represent distinct bugs (some
such bugs may already be filed) and appropriate fixing / XFAILing.
Failures include wrong signs of zero results, errors slightly above
the 9ulp bound (in such cases it may make sense for functions to set
round-to-nearest internally to reduce error accumulation), large
errors and incorrect overflow/underflow for the rounding mode (with
consequent missing errno settings in some cases). It's possible some
could be issues with test expectations, though I didn't notice any
that were obviously like that (I added NO_TEST_INLINE for cases that
were failing for ildoubl on x86 and where it seemed reasonable for
them to fail for the fast-math inlines).
There may of course be failures on other architectures for functions
that didn't fail on x86_64 or x86, in which case the usual rule
applies: file a bug (preferably identifying the underlying problem
function, in cases where function A calls function B and a problem
with function B may present in the test results for function A) if not
already in Bugzilla then fix or XFAIL.
Tested x86_64 and x86 and ulps updated accordingly.
* math/libm-test.inc (asinh_test): Use ALL_RM_TEST.
(atan_test): Likewise.
(atanh_test_data): Use NO_TEST_INLINE for two tests.
(atanh_test): Use ALL_RM_TEST.
(atan2_test_data): Likewise.
(cabs_test): Likewise.
(cacosh_test): Likewise.
(carg_test): Likewise.
(casin_test): Likewise.
(casinh_test): Likewise.
(cbrt_test): Likewise.
(csqrt_test): Likewise.
(erf_test): Likewise.
(erfc_test): Likewise.
(pow10_test): Likewise.
(exp2_test): Likewise.
(hypot_test): Likewise.
(j0_test): Likewise.
(j1_test): Likewise.
(lgamma_test): Likewise.
(gamma_test): Likewise.
(sincos_test): Likewise.
(tanh_test): Likewise.
(y0_test): Likewise.
(y1_test): Likewise.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2014-03-21 00:03:38 +00:00
|
|
|
Function: Real part of "casin_downward":
|
|
|
|
|
double: 3
|
|
|
|
|
float: 1
|
|
|
|
|
idouble: 3
|
|
|
|
|
ifloat: 1
|
|
|
|
|
ildouble: 2
|
|
|
|
|
ldouble: 2
|
|
|
|
|
|
|
|
|
|
Function: Imaginary part of "casin_downward":
|
|
|
|
|
double: 5
|
|
|
|
|
float: 3
|
|
|
|
|
idouble: 5
|
|
|
|
|
ifloat: 3
|
|
|
|
|
ildouble: 5
|
|
|
|
|
ldouble: 5
|
|
|
|
|
|
|
|
|
|
Function: Real part of "casin_towardzero":
|
|
|
|
|
double: 3
|
|
|
|
|
float: 1
|
|
|
|
|
idouble: 3
|
|
|
|
|
ifloat: 1
|
|
|
|
|
ildouble: 2
|
|
|
|
|
ldouble: 2
|
|
|
|
|
|
|
|
|
|
Function: Imaginary part of "casin_towardzero":
|
|
|
|
|
double: 5
|
|
|
|
|
float: 3
|
|
|
|
|
idouble: 5
|
|
|
|
|
ifloat: 3
|
|
|
|
|
ildouble: 5
|
|
|
|
|
ldouble: 5
|
|
|
|
|
|
|
|
|
|
Function: Real part of "casin_upward":
|
|
|
|
|
double: 1
|
|
|
|
|
float: 1
|
|
|
|
|
idouble: 1
|
|
|
|
|
ifloat: 1
|
|
|
|
|
ildouble: 2
|
|
|
|
|
ldouble: 2
|
|
|
|
|
|
|
|
|
|
Function: Imaginary part of "casin_upward":
|
|
|
|
|
double: 4
|
|
|
|
|
float: 4
|
|
|
|
|
idouble: 4
|
|
|
|
|
ifloat: 4
|
|
|
|
|
ildouble: 5
|
|
|
|
|
ldouble: 5
|
|
|
|
|
|
2001-09-19 10:37:31 +00:00
|
|
|
Function: Real part of "casinh":
|
2013-03-30 13:31:53 +00:00
|
|
|
double: 1
|
|
|
|
|
float: 2
|
|
|
|
|
idouble: 1
|
|
|
|
|
ifloat: 2
|
|
|
|
|
ildouble: 2
|
|
|
|
|
ldouble: 2
|
2001-09-19 10:37:31 +00:00
|
|
|
|
|
|
|
|
Function: Imaginary part of "casinh":
|
2013-03-30 13:31:53 +00:00
|
|
|
double: 1
|
|
|
|
|
float: 1
|
|
|
|
|
idouble: 1
|
|
|
|
|
ifloat: 1
|
|
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
2001-09-19 10:37:31 +00:00
|
|
|
|
Test most libm functions in all rounding modes.
This patch makes libm-test.inc tests of most functions use ALL_RM_TEST
unless there was some reason to defer that change for a particular
function.
I started out planning to defer the change for pow (bug 16315), cexp /
ccos / ccosh / csin / csinh (likely fallout from exp, bug 16284) and
cpow (exact expectations for signs of exact zero results not wanted).
Testing on x86_64 and x86 showed additional failures for acosh, cacos,
catan, catanh, clog, clog10, jn, log, log10, log1p, log2, tgamma, yn,
so making the change for those functions was deferred as well, pending
investigation to show which of these represent distinct bugs (some
such bugs may already be filed) and appropriate fixing / XFAILing.
Failures include wrong signs of zero results, errors slightly above
the 9ulp bound (in such cases it may make sense for functions to set
round-to-nearest internally to reduce error accumulation), large
errors and incorrect overflow/underflow for the rounding mode (with
consequent missing errno settings in some cases). It's possible some
could be issues with test expectations, though I didn't notice any
that were obviously like that (I added NO_TEST_INLINE for cases that
were failing for ildoubl on x86 and where it seemed reasonable for
them to fail for the fast-math inlines).
There may of course be failures on other architectures for functions
that didn't fail on x86_64 or x86, in which case the usual rule
applies: file a bug (preferably identifying the underlying problem
function, in cases where function A calls function B and a problem
with function B may present in the test results for function A) if not
already in Bugzilla then fix or XFAIL.
Tested x86_64 and x86 and ulps updated accordingly.
* math/libm-test.inc (asinh_test): Use ALL_RM_TEST.
(atan_test): Likewise.
(atanh_test_data): Use NO_TEST_INLINE for two tests.
(atanh_test): Use ALL_RM_TEST.
(atan2_test_data): Likewise.
(cabs_test): Likewise.
(cacosh_test): Likewise.
(carg_test): Likewise.
(casin_test): Likewise.
(casinh_test): Likewise.
(cbrt_test): Likewise.
(csqrt_test): Likewise.
(erf_test): Likewise.
(erfc_test): Likewise.
(pow10_test): Likewise.
(exp2_test): Likewise.
(hypot_test): Likewise.
(j0_test): Likewise.
(j1_test): Likewise.
(lgamma_test): Likewise.
(gamma_test): Likewise.
(sincos_test): Likewise.
(tanh_test): Likewise.
(y0_test): Likewise.
(y1_test): Likewise.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2014-03-21 00:03:38 +00:00
|
|
|
Function: Real part of "casinh_downward":
|
|
|
|
|
double: 5
|
|
|
|
|
float: 3
|
|
|
|
|
idouble: 5
|
|
|
|
|
ifloat: 3
|
|
|
|
|
ildouble: 5
|
|
|
|
|
ldouble: 5
|
|
|
|
|
|
|
|
|
|
Function: Imaginary part of "casinh_downward":
|
|
|
|
|
double: 3
|
|
|
|
|
float: 1
|
|
|
|
|
idouble: 3
|
|
|
|
|
ifloat: 1
|
|
|
|
|
ildouble: 2
|
|
|
|
|
ldouble: 2
|
|
|
|
|
|
|
|
|
|
Function: Real part of "casinh_towardzero":
|
|
|
|
|
double: 5
|
|
|
|
|
float: 3
|
|
|
|
|
idouble: 5
|
|
|
|
|
ifloat: 3
|
|
|
|
|
ildouble: 5
|
|
|
|
|
ldouble: 5
|
|
|
|
|
|
|
|
|
|
Function: Imaginary part of "casinh_towardzero":
|
|
|
|
|
double: 3
|
|
|
|
|
float: 1
|
|
|
|
|
idouble: 3
|
|
|
|
|
ifloat: 1
|
|
|
|
|
ildouble: 2
|
|
|
|
|
ldouble: 2
|
|
|
|
|
|
|
|
|
|
Function: Real part of "casinh_upward":
|
|
|
|
|
double: 4
|
|
|
|
|
float: 4
|
|
|
|
|
idouble: 4
|
|
|
|
|
ifloat: 4
|
|
|
|
|
ildouble: 5
|
|
|
|
|
ldouble: 5
|
|
|
|
|
|
|
|
|
|
Function: Imaginary part of "casinh_upward":
|
|
|
|
|
double: 1
|
|
|
|
|
float: 2
|
|
|
|
|
idouble: 1
|
|
|
|
|
ifloat: 2
|
|
|
|
|
ildouble: 2
|
|
|
|
|
ldouble: 2
|
|
|
|
|
|
2001-09-19 10:37:31 +00:00
|
|
|
Function: Real part of "catan":
|
2013-07-02 20:00:48 +00:00
|
|
|
float: 1
|
|
|
|
|
ifloat: 1
|
2001-09-19 10:37:31 +00:00
|
|
|
|
|
|
|
|
Function: Imaginary part of "catan":
|
|
|
|
|
double: 1
|
|
|
|
|
float: 1
|
|
|
|
|
idouble: 1
|
|
|
|
|
ifloat: 1
|
2013-04-27 14:56:34 +00:00
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
2001-09-19 10:37:31 +00:00
|
|
|
|
2014-04-02 17:41:02 +00:00
|
|
|
Function: Real part of "catan_downward":
|
|
|
|
|
double: 1
|
|
|
|
|
float: 1
|
|
|
|
|
idouble: 1
|
|
|
|
|
ifloat: 1
|
|
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
|
|
|
|
|
|
|
|
|
Function: Imaginary part of "catan_downward":
|
|
|
|
|
double: 2
|
|
|
|
|
float: 2
|
|
|
|
|
idouble: 2
|
|
|
|
|
ifloat: 2
|
|
|
|
|
ildouble: 4
|
|
|
|
|
ldouble: 4
|
|
|
|
|
|
|
|
|
|
Function: Real part of "catan_towardzero":
|
|
|
|
|
double: 1
|
|
|
|
|
float: 1
|
|
|
|
|
idouble: 1
|
|
|
|
|
ifloat: 1
|
|
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
|
|
|
|
|
|
|
|
|
Function: Imaginary part of "catan_towardzero":
|
|
|
|
|
double: 2
|
|
|
|
|
float: 1
|
|
|
|
|
idouble: 2
|
|
|
|
|
ifloat: 1
|
|
|
|
|
ildouble: 4
|
|
|
|
|
ldouble: 4
|
|
|
|
|
|
|
|
|
|
Function: Real part of "catan_upward":
|
|
|
|
|
double: 1
|
|
|
|
|
float: 1
|
|
|
|
|
idouble: 1
|
|
|
|
|
ifloat: 1
|
|
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
|
|
|
|
|
|
|
|
|
Function: Imaginary part of "catan_upward":
|
|
|
|
|
double: 3
|
|
|
|
|
float: 3
|
|
|
|
|
idouble: 3
|
|
|
|
|
ifloat: 3
|
|
|
|
|
ildouble: 3
|
|
|
|
|
ldouble: 3
|
|
|
|
|
|
2001-09-19 10:37:31 +00:00
|
|
|
Function: Real part of "catanh":
|
2013-07-02 20:00:48 +00:00
|
|
|
double: 1
|
2013-04-30 11:27:35 +00:00
|
|
|
float: 1
|
2013-07-02 20:00:48 +00:00
|
|
|
idouble: 1
|
2013-04-30 11:27:35 +00:00
|
|
|
ifloat: 1
|
2001-09-19 10:37:31 +00:00
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
|
|
|
|
|
|
|
|
|
Function: Imaginary part of "catanh":
|
2013-04-30 11:27:35 +00:00
|
|
|
float: 1
|
|
|
|
|
ifloat: 1
|
2001-09-19 10:37:31 +00:00
|
|
|
|
2014-04-02 17:41:02 +00:00
|
|
|
Function: Real part of "catanh_downward":
|
|
|
|
|
double: 2
|
|
|
|
|
float: 2
|
|
|
|
|
idouble: 2
|
|
|
|
|
ifloat: 2
|
|
|
|
|
ildouble: 4
|
|
|
|
|
ldouble: 4
|
|
|
|
|
|
|
|
|
|
Function: Imaginary part of "catanh_downward":
|
|
|
|
|
double: 1
|
|
|
|
|
float: 2
|
|
|
|
|
idouble: 1
|
|
|
|
|
ifloat: 2
|
|
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
|
|
|
|
|
|
|
|
|
Function: Real part of "catanh_towardzero":
|
|
|
|
|
double: 2
|
|
|
|
|
float: 1
|
|
|
|
|
idouble: 2
|
|
|
|
|
ifloat: 1
|
|
|
|
|
ildouble: 4
|
|
|
|
|
ldouble: 4
|
|
|
|
|
|
|
|
|
|
Function: Imaginary part of "catanh_towardzero":
|
|
|
|
|
double: 1
|
|
|
|
|
float: 2
|
|
|
|
|
idouble: 1
|
|
|
|
|
ifloat: 2
|
|
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
|
|
|
|
|
|
|
|
|
Function: Real part of "catanh_upward":
|
|
|
|
|
double: 4
|
|
|
|
|
float: 3
|
|
|
|
|
idouble: 4
|
|
|
|
|
ifloat: 3
|
|
|
|
|
ildouble: 4
|
|
|
|
|
ldouble: 4
|
|
|
|
|
|
|
|
|
|
Function: Imaginary part of "catanh_upward":
|
|
|
|
|
double: 1
|
|
|
|
|
float: 1
|
|
|
|
|
idouble: 1
|
|
|
|
|
ifloat: 1
|
|
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
|
|
|
|
|
2001-09-19 10:37:31 +00:00
|
|
|
Function: "cbrt":
|
|
|
|
|
double: 1
|
2013-11-30 22:04:13 +00:00
|
|
|
float: 1
|
2001-09-19 10:37:31 +00:00
|
|
|
idouble: 1
|
2013-11-30 22:04:13 +00:00
|
|
|
ifloat: 1
|
2002-08-31 08:22:35 +00:00
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
2001-09-19 10:37:31 +00:00
|
|
|
|
Test most libm functions in all rounding modes.
This patch makes libm-test.inc tests of most functions use ALL_RM_TEST
unless there was some reason to defer that change for a particular
function.
I started out planning to defer the change for pow (bug 16315), cexp /
ccos / ccosh / csin / csinh (likely fallout from exp, bug 16284) and
cpow (exact expectations for signs of exact zero results not wanted).
Testing on x86_64 and x86 showed additional failures for acosh, cacos,
catan, catanh, clog, clog10, jn, log, log10, log1p, log2, tgamma, yn,
so making the change for those functions was deferred as well, pending
investigation to show which of these represent distinct bugs (some
such bugs may already be filed) and appropriate fixing / XFAILing.
Failures include wrong signs of zero results, errors slightly above
the 9ulp bound (in such cases it may make sense for functions to set
round-to-nearest internally to reduce error accumulation), large
errors and incorrect overflow/underflow for the rounding mode (with
consequent missing errno settings in some cases). It's possible some
could be issues with test expectations, though I didn't notice any
that were obviously like that (I added NO_TEST_INLINE for cases that
were failing for ildoubl on x86 and where it seemed reasonable for
them to fail for the fast-math inlines).
There may of course be failures on other architectures for functions
that didn't fail on x86_64 or x86, in which case the usual rule
applies: file a bug (preferably identifying the underlying problem
function, in cases where function A calls function B and a problem
with function B may present in the test results for function A) if not
already in Bugzilla then fix or XFAIL.
Tested x86_64 and x86 and ulps updated accordingly.
* math/libm-test.inc (asinh_test): Use ALL_RM_TEST.
(atan_test): Likewise.
(atanh_test_data): Use NO_TEST_INLINE for two tests.
(atanh_test): Use ALL_RM_TEST.
(atan2_test_data): Likewise.
(cabs_test): Likewise.
(cacosh_test): Likewise.
(carg_test): Likewise.
(casin_test): Likewise.
(casinh_test): Likewise.
(cbrt_test): Likewise.
(csqrt_test): Likewise.
(erf_test): Likewise.
(erfc_test): Likewise.
(pow10_test): Likewise.
(exp2_test): Likewise.
(hypot_test): Likewise.
(j0_test): Likewise.
(j1_test): Likewise.
(lgamma_test): Likewise.
(gamma_test): Likewise.
(sincos_test): Likewise.
(tanh_test): Likewise.
(y0_test): Likewise.
(y1_test): Likewise.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2014-03-21 00:03:38 +00:00
|
|
|
Function: "cbrt_downward":
|
|
|
|
|
double: 2
|
|
|
|
|
float: 1
|
|
|
|
|
idouble: 2
|
|
|
|
|
ifloat: 1
|
|
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
|
|
|
|
|
|
|
|
|
Function: "cbrt_towardzero":
|
|
|
|
|
double: 2
|
|
|
|
|
idouble: 2
|
|
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
|
|
|
|
|
|
|
|
|
Function: "cbrt_upward":
|
|
|
|
|
double: 3
|
|
|
|
|
float: 1
|
|
|
|
|
idouble: 3
|
|
|
|
|
ifloat: 1
|
|
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
|
|
|
|
|
2001-09-19 10:37:31 +00:00
|
|
|
Function: Real part of "ccos":
|
|
|
|
|
double: 1
|
2002-08-31 08:22:35 +00:00
|
|
|
float: 1
|
2001-09-19 10:37:31 +00:00
|
|
|
idouble: 1
|
2002-08-31 08:22:35 +00:00
|
|
|
ifloat: 1
|
|
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
2001-09-19 10:37:31 +00:00
|
|
|
|
|
|
|
|
Function: Imaginary part of "ccos":
|
2012-05-19 15:35:29 +00:00
|
|
|
double: 1
|
2001-09-19 10:37:31 +00:00
|
|
|
float: 1
|
2012-05-19 15:35:29 +00:00
|
|
|
idouble: 1
|
2001-09-19 10:37:31 +00:00
|
|
|
ifloat: 1
|
|
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
|
|
|
|
|
2014-03-24 12:18:45 +00:00
|
|
|
Function: Real part of "ccos_downward":
|
|
|
|
|
double: 1
|
|
|
|
|
float: 1
|
|
|
|
|
idouble: 1
|
|
|
|
|
ifloat: 1
|
|
|
|
|
ildouble: 3
|
|
|
|
|
ldouble: 3
|
|
|
|
|
|
|
|
|
|
Function: Imaginary part of "ccos_downward":
|
|
|
|
|
double: 2
|
|
|
|
|
float: 3
|
|
|
|
|
idouble: 2
|
|
|
|
|
ifloat: 3
|
|
|
|
|
ildouble: 3
|
|
|
|
|
ldouble: 3
|
|
|
|
|
|
|
|
|
|
Function: Real part of "ccos_towardzero":
|
|
|
|
|
double: 1
|
|
|
|
|
float: 2
|
|
|
|
|
idouble: 1
|
|
|
|
|
ifloat: 2
|
|
|
|
|
ildouble: 3
|
|
|
|
|
ldouble: 3
|
|
|
|
|
|
|
|
|
|
Function: Imaginary part of "ccos_towardzero":
|
|
|
|
|
double: 2
|
|
|
|
|
float: 3
|
|
|
|
|
idouble: 2
|
|
|
|
|
ifloat: 3
|
|
|
|
|
ildouble: 3
|
|
|
|
|
ldouble: 3
|
|
|
|
|
|
|
|
|
|
Function: Real part of "ccos_upward":
|
|
|
|
|
double: 1
|
|
|
|
|
float: 2
|
|
|
|
|
idouble: 1
|
|
|
|
|
ifloat: 2
|
|
|
|
|
ildouble: 2
|
|
|
|
|
ldouble: 2
|
|
|
|
|
|
|
|
|
|
Function: Imaginary part of "ccos_upward":
|
|
|
|
|
double: 2
|
|
|
|
|
float: 2
|
|
|
|
|
idouble: 2
|
|
|
|
|
ifloat: 2
|
|
|
|
|
ildouble: 2
|
|
|
|
|
ldouble: 2
|
|
|
|
|
|
2001-09-19 10:37:31 +00:00
|
|
|
Function: Real part of "ccosh":
|
|
|
|
|
double: 1
|
|
|
|
|
float: 1
|
|
|
|
|
idouble: 1
|
|
|
|
|
ifloat: 1
|
|
|
|
|
|
|
|
|
|
Function: Imaginary part of "ccosh":
|
|
|
|
|
double: 1
|
|
|
|
|
float: 1
|
|
|
|
|
idouble: 1
|
|
|
|
|
ifloat: 1
|
|
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
|
|
|
|
|
2014-03-24 12:18:45 +00:00
|
|
|
Function: Real part of "ccosh_downward":
|
|
|
|
|
double: 1
|
|
|
|
|
float: 2
|
|
|
|
|
idouble: 1
|
|
|
|
|
ifloat: 2
|
|
|
|
|
ildouble: 3
|
|
|
|
|
ldouble: 3
|
|
|
|
|
|
|
|
|
|
Function: Imaginary part of "ccosh_downward":
|
|
|
|
|
double: 2
|
|
|
|
|
float: 3
|
|
|
|
|
idouble: 2
|
|
|
|
|
ifloat: 3
|
|
|
|
|
ildouble: 3
|
|
|
|
|
ldouble: 3
|
|
|
|
|
|
|
|
|
|
Function: Real part of "ccosh_towardzero":
|
|
|
|
|
double: 1
|
|
|
|
|
float: 3
|
|
|
|
|
idouble: 1
|
|
|
|
|
ifloat: 3
|
|
|
|
|
ildouble: 3
|
|
|
|
|
ldouble: 3
|
|
|
|
|
|
|
|
|
|
Function: Imaginary part of "ccosh_towardzero":
|
|
|
|
|
double: 2
|
|
|
|
|
float: 3
|
|
|
|
|
idouble: 2
|
|
|
|
|
ifloat: 3
|
|
|
|
|
ildouble: 3
|
|
|
|
|
ldouble: 3
|
|
|
|
|
|
|
|
|
|
Function: Real part of "ccosh_upward":
|
|
|
|
|
double: 1
|
|
|
|
|
float: 2
|
|
|
|
|
idouble: 1
|
|
|
|
|
ifloat: 2
|
|
|
|
|
ildouble: 2
|
|
|
|
|
ldouble: 2
|
|
|
|
|
|
|
|
|
|
Function: Imaginary part of "ccosh_upward":
|
|
|
|
|
double: 2
|
|
|
|
|
float: 2
|
|
|
|
|
idouble: 2
|
|
|
|
|
ifloat: 2
|
|
|
|
|
ildouble: 2
|
|
|
|
|
ldouble: 2
|
|
|
|
|
|
2001-09-19 10:37:31 +00:00
|
|
|
Function: Real part of "cexp":
|
2012-03-21 15:28:05 +00:00
|
|
|
double: 2
|
2001-09-19 10:37:31 +00:00
|
|
|
float: 1
|
2012-03-21 15:28:05 +00:00
|
|
|
idouble: 2
|
2001-09-19 10:37:31 +00:00
|
|
|
ifloat: 1
|
2012-03-21 15:28:05 +00:00
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
2001-09-19 10:37:31 +00:00
|
|
|
|
|
|
|
|
Function: Imaginary part of "cexp":
|
2012-03-21 15:28:05 +00:00
|
|
|
double: 1
|
2012-03-22 19:38:09 +00:00
|
|
|
float: 2
|
2012-03-21 15:28:05 +00:00
|
|
|
idouble: 1
|
2012-03-22 19:38:09 +00:00
|
|
|
ifloat: 2
|
2002-08-31 08:22:35 +00:00
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
|
|
|
|
|
|
|
|
|
Function: Real part of "clog":
|
2012-03-19 20:13:10 +00:00
|
|
|
double: 1
|
2001-09-19 10:37:31 +00:00
|
|
|
float: 1
|
2012-03-19 20:13:10 +00:00
|
|
|
idouble: 1
|
2001-09-19 10:37:31 +00:00
|
|
|
ifloat: 1
|
|
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
|
|
|
|
|
|
|
|
|
Function: Imaginary part of "clog":
|
2013-07-02 20:00:48 +00:00
|
|
|
float: 1
|
|
|
|
|
ifloat: 1
|
2013-07-02 22:01:13 +00:00
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
2001-09-19 10:37:31 +00:00
|
|
|
|
|
|
|
|
Function: Real part of "clog10":
|
2012-07-31 14:21:19 +00:00
|
|
|
double: 2
|
2012-09-25 19:43:49 +00:00
|
|
|
float: 2
|
2012-07-31 14:21:19 +00:00
|
|
|
idouble: 2
|
2012-09-25 19:43:49 +00:00
|
|
|
ifloat: 2
|
2014-02-19 14:26:29 +00:00
|
|
|
ildouble: 2
|
|
|
|
|
ldouble: 2
|
2001-09-19 10:37:31 +00:00
|
|
|
|
|
|
|
|
Function: Imaginary part of "clog10":
|
|
|
|
|
double: 1
|
2012-09-25 19:43:49 +00:00
|
|
|
float: 1
|
2001-09-19 10:37:31 +00:00
|
|
|
idouble: 1
|
2012-09-25 19:43:49 +00:00
|
|
|
ifloat: 1
|
2014-02-19 14:26:29 +00:00
|
|
|
ildouble: 2
|
|
|
|
|
ldouble: 2
|
2002-08-31 08:22:35 +00:00
|
|
|
|
2014-04-02 13:10:19 +00:00
|
|
|
Function: Real part of "clog10_downward":
|
|
|
|
|
double: 3
|
|
|
|
|
float: 3
|
|
|
|
|
idouble: 3
|
|
|
|
|
ifloat: 3
|
|
|
|
|
ildouble: 4
|
|
|
|
|
ldouble: 4
|
|
|
|
|
|
|
|
|
|
Function: Imaginary part of "clog10_downward":
|
|
|
|
|
double: 3
|
|
|
|
|
float: 2
|
|
|
|
|
idouble: 3
|
|
|
|
|
ifloat: 2
|
|
|
|
|
ildouble: 2
|
|
|
|
|
ldouble: 2
|
|
|
|
|
|
|
|
|
|
Function: Real part of "clog10_towardzero":
|
|
|
|
|
double: 3
|
|
|
|
|
float: 2
|
|
|
|
|
idouble: 3
|
|
|
|
|
ifloat: 2
|
|
|
|
|
ildouble: 4
|
|
|
|
|
ldouble: 4
|
|
|
|
|
|
|
|
|
|
Function: Imaginary part of "clog10_towardzero":
|
|
|
|
|
double: 3
|
|
|
|
|
float: 2
|
|
|
|
|
idouble: 3
|
|
|
|
|
ifloat: 2
|
|
|
|
|
ildouble: 2
|
|
|
|
|
ldouble: 2
|
|
|
|
|
|
|
|
|
|
Function: Real part of "clog10_upward":
|
|
|
|
|
double: 4
|
|
|
|
|
float: 3
|
|
|
|
|
idouble: 4
|
|
|
|
|
ifloat: 3
|
|
|
|
|
ildouble: 4
|
|
|
|
|
ldouble: 4
|
|
|
|
|
|
|
|
|
|
Function: Imaginary part of "clog10_upward":
|
|
|
|
|
double: 2
|
|
|
|
|
float: 2
|
|
|
|
|
idouble: 2
|
|
|
|
|
ifloat: 2
|
|
|
|
|
ildouble: 2
|
|
|
|
|
ldouble: 2
|
|
|
|
|
|
|
|
|
|
Function: Real part of "clog_downward":
|
|
|
|
|
double: 2
|
|
|
|
|
float: 2
|
|
|
|
|
idouble: 2
|
|
|
|
|
ifloat: 2
|
|
|
|
|
ildouble: 2
|
|
|
|
|
ldouble: 2
|
|
|
|
|
|
|
|
|
|
Function: Imaginary part of "clog_downward":
|
|
|
|
|
double: 1
|
|
|
|
|
float: 2
|
|
|
|
|
idouble: 1
|
|
|
|
|
ifloat: 2
|
|
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
|
|
|
|
|
|
|
|
|
Function: Real part of "clog_towardzero":
|
|
|
|
|
double: 2
|
|
|
|
|
float: 2
|
|
|
|
|
idouble: 2
|
|
|
|
|
ifloat: 2
|
|
|
|
|
ildouble: 2
|
|
|
|
|
ldouble: 2
|
|
|
|
|
|
|
|
|
|
Function: Imaginary part of "clog_towardzero":
|
|
|
|
|
double: 1
|
|
|
|
|
float: 2
|
|
|
|
|
idouble: 1
|
|
|
|
|
ifloat: 2
|
|
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
|
|
|
|
|
|
|
|
|
Function: Real part of "clog_upward":
|
|
|
|
|
double: 2
|
|
|
|
|
float: 1
|
|
|
|
|
idouble: 2
|
|
|
|
|
ifloat: 1
|
|
|
|
|
ildouble: 2
|
|
|
|
|
ldouble: 2
|
|
|
|
|
|
|
|
|
|
Function: Imaginary part of "clog_upward":
|
|
|
|
|
double: 2
|
|
|
|
|
float: 2
|
|
|
|
|
idouble: 2
|
|
|
|
|
ifloat: 2
|
|
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
|
|
|
|
|
2001-09-19 10:37:31 +00:00
|
|
|
Function: "cos":
|
2002-12-03 15:39:37 +00:00
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
2001-09-19 10:37:31 +00:00
|
|
|
|
2012-03-02 20:51:39 +00:00
|
|
|
Function: "cos_downward":
|
2013-11-30 22:04:13 +00:00
|
|
|
double: 1
|
|
|
|
|
idouble: 1
|
|
|
|
|
ildouble: 2
|
|
|
|
|
ldouble: 2
|
|
|
|
|
|
2012-03-02 20:51:39 +00:00
|
|
|
Function: "cos_towardzero":
|
2013-11-30 22:04:13 +00:00
|
|
|
double: 1
|
|
|
|
|
idouble: 1
|
2012-03-02 20:51:39 +00:00
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
|
|
|
|
|
|
|
|
|
Function: "cos_upward":
|
2013-11-30 22:04:13 +00:00
|
|
|
double: 1
|
|
|
|
|
idouble: 1
|
|
|
|
|
ildouble: 2
|
|
|
|
|
ldouble: 2
|
|
|
|
|
|
|
|
|
|
Function: "cosh":
|
|
|
|
|
double: 1
|
Add more libm-test coverage of [a-c]* real functions.
Various libm functions have inadequate test coverage in libm-test.inc
/ auto-libm-test-in - failing to cover all the usual special cases
(infinities, NaNs, zero, large and small finite values, subnormals) as
well as a reasonable range of ordinary inputs and, where appropriate,
inputs close to the thresholds for underflow and overflow.
This patch improves test coverage for real functions [a-c]* (with the
expectation of adding more coverage for other functions later).
Tested x86_64 and x86 and ulps updated accordingly (and eight glibc
bugs and one C11 DR filed for issues found in the process).
* math/auto-libm-test-in: Add more tests of acos, acosh, asin,
asinh, atan, atan2, atanh, cbrt, cos and cosh.
* math/auto-libm-test-out: Regenerated.
* math/libm-test.inc (acosh_test_data): Add more tests.
(atanh_test_data): Likewise.
(ceil_test_data): Likewise.
(copysign_test_data): Likewise.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2013-12-20 21:03:39 +00:00
|
|
|
float: 1
|
2013-11-30 22:04:13 +00:00
|
|
|
idouble: 1
|
Add more libm-test coverage of [a-c]* real functions.
Various libm functions have inadequate test coverage in libm-test.inc
/ auto-libm-test-in - failing to cover all the usual special cases
(infinities, NaNs, zero, large and small finite values, subnormals) as
well as a reasonable range of ordinary inputs and, where appropriate,
inputs close to the thresholds for underflow and overflow.
This patch improves test coverage for real functions [a-c]* (with the
expectation of adding more coverage for other functions later).
Tested x86_64 and x86 and ulps updated accordingly (and eight glibc
bugs and one C11 DR filed for issues found in the process).
* math/auto-libm-test-in: Add more tests of acos, acosh, asin,
asinh, atan, atan2, atanh, cbrt, cos and cosh.
* math/auto-libm-test-out: Regenerated.
* math/libm-test.inc (acosh_test_data): Add more tests.
(atanh_test_data): Likewise.
(ceil_test_data): Likewise.
(copysign_test_data): Likewise.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2013-12-20 21:03:39 +00:00
|
|
|
ifloat: 1
|
2012-03-02 20:51:39 +00:00
|
|
|
ildouble: 1
|
Add more libm-test coverage of [a-c]* real functions.
Various libm functions have inadequate test coverage in libm-test.inc
/ auto-libm-test-in - failing to cover all the usual special cases
(infinities, NaNs, zero, large and small finite values, subnormals) as
well as a reasonable range of ordinary inputs and, where appropriate,
inputs close to the thresholds for underflow and overflow.
This patch improves test coverage for real functions [a-c]* (with the
expectation of adding more coverage for other functions later).
Tested x86_64 and x86 and ulps updated accordingly (and eight glibc
bugs and one C11 DR filed for issues found in the process).
* math/auto-libm-test-in: Add more tests of acos, acosh, asin,
asinh, atan, atan2, atanh, cbrt, cos and cosh.
* math/auto-libm-test-out: Regenerated.
* math/libm-test.inc (acosh_test_data): Add more tests.
(atanh_test_data): Likewise.
(ceil_test_data): Likewise.
(copysign_test_data): Likewise.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2013-12-20 21:03:39 +00:00
|
|
|
ldouble: 2
|
2012-03-02 20:51:39 +00:00
|
|
|
|
2012-03-05 12:20:24 +00:00
|
|
|
Function: "cosh_downward":
|
2013-11-30 22:04:13 +00:00
|
|
|
double: 1
|
2012-03-05 12:20:24 +00:00
|
|
|
float: 1
|
2013-11-30 22:04:13 +00:00
|
|
|
idouble: 1
|
2012-03-05 12:20:24 +00:00
|
|
|
ifloat: 1
|
|
|
|
|
ildouble: 2
|
Add more libm-test coverage of [a-c]* real functions.
Various libm functions have inadequate test coverage in libm-test.inc
/ auto-libm-test-in - failing to cover all the usual special cases
(infinities, NaNs, zero, large and small finite values, subnormals) as
well as a reasonable range of ordinary inputs and, where appropriate,
inputs close to the thresholds for underflow and overflow.
This patch improves test coverage for real functions [a-c]* (with the
expectation of adding more coverage for other functions later).
Tested x86_64 and x86 and ulps updated accordingly (and eight glibc
bugs and one C11 DR filed for issues found in the process).
* math/auto-libm-test-in: Add more tests of acos, acosh, asin,
asinh, atan, atan2, atanh, cbrt, cos and cosh.
* math/auto-libm-test-out: Regenerated.
* math/libm-test.inc (acosh_test_data): Add more tests.
(atanh_test_data): Likewise.
(ceil_test_data): Likewise.
(copysign_test_data): Likewise.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2013-12-20 21:03:39 +00:00
|
|
|
ldouble: 3
|
2012-03-05 12:20:24 +00:00
|
|
|
|
|
|
|
|
Function: "cosh_towardzero":
|
2013-11-30 22:04:13 +00:00
|
|
|
double: 1
|
2012-03-05 12:20:24 +00:00
|
|
|
float: 1
|
2013-11-30 22:04:13 +00:00
|
|
|
idouble: 1
|
2012-03-05 12:20:24 +00:00
|
|
|
ifloat: 1
|
2014-01-01 14:34:38 +00:00
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 2
|
2012-03-05 12:20:24 +00:00
|
|
|
|
|
|
|
|
Function: "cosh_upward":
|
2013-11-30 22:04:13 +00:00
|
|
|
double: 1
|
Add more libm-test coverage of [a-c]* real functions.
Various libm functions have inadequate test coverage in libm-test.inc
/ auto-libm-test-in - failing to cover all the usual special cases
(infinities, NaNs, zero, large and small finite values, subnormals) as
well as a reasonable range of ordinary inputs and, where appropriate,
inputs close to the thresholds for underflow and overflow.
This patch improves test coverage for real functions [a-c]* (with the
expectation of adding more coverage for other functions later).
Tested x86_64 and x86 and ulps updated accordingly (and eight glibc
bugs and one C11 DR filed for issues found in the process).
* math/auto-libm-test-in: Add more tests of acos, acosh, asin,
asinh, atan, atan2, atanh, cbrt, cos and cosh.
* math/auto-libm-test-out: Regenerated.
* math/libm-test.inc (acosh_test_data): Add more tests.
(atanh_test_data): Likewise.
(ceil_test_data): Likewise.
(copysign_test_data): Likewise.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2013-12-20 21:03:39 +00:00
|
|
|
float: 2
|
2013-11-30 22:04:13 +00:00
|
|
|
idouble: 1
|
Add more libm-test coverage of [a-c]* real functions.
Various libm functions have inadequate test coverage in libm-test.inc
/ auto-libm-test-in - failing to cover all the usual special cases
(infinities, NaNs, zero, large and small finite values, subnormals) as
well as a reasonable range of ordinary inputs and, where appropriate,
inputs close to the thresholds for underflow and overflow.
This patch improves test coverage for real functions [a-c]* (with the
expectation of adding more coverage for other functions later).
Tested x86_64 and x86 and ulps updated accordingly (and eight glibc
bugs and one C11 DR filed for issues found in the process).
* math/auto-libm-test-in: Add more tests of acos, acosh, asin,
asinh, atan, atan2, atanh, cbrt, cos and cosh.
* math/auto-libm-test-out: Regenerated.
* math/libm-test.inc (acosh_test_data): Add more tests.
(atanh_test_data): Likewise.
(ceil_test_data): Likewise.
(copysign_test_data): Likewise.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2013-12-20 21:03:39 +00:00
|
|
|
ifloat: 2
|
2015-03-24 23:48:04 +00:00
|
|
|
ildouble: 2
|
Add more libm-test coverage of [a-c]* real functions.
Various libm functions have inadequate test coverage in libm-test.inc
/ auto-libm-test-in - failing to cover all the usual special cases
(infinities, NaNs, zero, large and small finite values, subnormals) as
well as a reasonable range of ordinary inputs and, where appropriate,
inputs close to the thresholds for underflow and overflow.
This patch improves test coverage for real functions [a-c]* (with the
expectation of adding more coverage for other functions later).
Tested x86_64 and x86 and ulps updated accordingly (and eight glibc
bugs and one C11 DR filed for issues found in the process).
* math/auto-libm-test-in: Add more tests of acos, acosh, asin,
asinh, atan, atan2, atanh, cbrt, cos and cosh.
* math/auto-libm-test-out: Regenerated.
* math/libm-test.inc (acosh_test_data): Add more tests.
(atanh_test_data): Likewise.
(ceil_test_data): Likewise.
(copysign_test_data): Likewise.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2013-12-20 21:03:39 +00:00
|
|
|
ldouble: 3
|
2012-03-05 12:20:24 +00:00
|
|
|
|
2001-09-19 10:37:31 +00:00
|
|
|
Function: Real part of "cpow":
|
2002-08-31 10:32:59 +00:00
|
|
|
double: 2
|
2002-12-03 15:39:37 +00:00
|
|
|
float: 5
|
2002-08-31 10:32:59 +00:00
|
|
|
idouble: 2
|
2002-12-03 15:39:37 +00:00
|
|
|
ifloat: 5
|
2013-07-02 20:00:48 +00:00
|
|
|
ildouble: 3
|
|
|
|
|
ldouble: 3
|
2001-09-19 10:37:31 +00:00
|
|
|
|
|
|
|
|
Function: Imaginary part of "cpow":
|
|
|
|
|
float: 2
|
|
|
|
|
ifloat: 2
|
2011-10-23 00:06:23 +00:00
|
|
|
ildouble: 4
|
|
|
|
|
ldouble: 4
|
2001-09-19 10:37:31 +00:00
|
|
|
|
2014-06-23 20:15:14 +00:00
|
|
|
Function: Real part of "cpow_downward":
|
|
|
|
|
double: 4
|
|
|
|
|
float: 8
|
|
|
|
|
idouble: 4
|
|
|
|
|
ifloat: 8
|
|
|
|
|
ildouble: 7
|
|
|
|
|
ldouble: 7
|
|
|
|
|
|
|
|
|
|
Function: Imaginary part of "cpow_downward":
|
|
|
|
|
double: 2
|
|
|
|
|
float: 2
|
|
|
|
|
idouble: 2
|
|
|
|
|
ifloat: 2
|
|
|
|
|
ildouble: 2
|
|
|
|
|
ldouble: 2
|
|
|
|
|
|
|
|
|
|
Function: Real part of "cpow_towardzero":
|
|
|
|
|
double: 4
|
|
|
|
|
float: 8
|
|
|
|
|
idouble: 4
|
|
|
|
|
ifloat: 8
|
|
|
|
|
ildouble: 7
|
|
|
|
|
ldouble: 7
|
|
|
|
|
|
|
|
|
|
Function: Imaginary part of "cpow_towardzero":
|
|
|
|
|
double: 2
|
|
|
|
|
float: 2
|
|
|
|
|
idouble: 2
|
|
|
|
|
ifloat: 2
|
|
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
|
|
|
|
|
|
|
|
|
Function: Real part of "cpow_upward":
|
|
|
|
|
double: 4
|
|
|
|
|
float: 1
|
|
|
|
|
idouble: 4
|
|
|
|
|
ifloat: 1
|
|
|
|
|
ildouble: 2
|
|
|
|
|
ldouble: 2
|
|
|
|
|
|
|
|
|
|
Function: Imaginary part of "cpow_upward":
|
|
|
|
|
double: 1
|
|
|
|
|
float: 2
|
|
|
|
|
idouble: 1
|
|
|
|
|
ifloat: 2
|
|
|
|
|
ildouble: 2
|
|
|
|
|
ldouble: 2
|
|
|
|
|
|
2002-08-31 08:22:35 +00:00
|
|
|
Function: Real part of "csin":
|
2012-05-19 15:35:29 +00:00
|
|
|
double: 1
|
|
|
|
|
float: 1
|
|
|
|
|
idouble: 1
|
|
|
|
|
ifloat: 1
|
2002-08-31 08:22:35 +00:00
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
|
|
|
|
|
2014-03-24 12:18:45 +00:00
|
|
|
Function: Real part of "csin_downward":
|
|
|
|
|
double: 2
|
|
|
|
|
float: 3
|
|
|
|
|
idouble: 2
|
|
|
|
|
ifloat: 3
|
|
|
|
|
ildouble: 3
|
|
|
|
|
ldouble: 3
|
|
|
|
|
|
|
|
|
|
Function: Imaginary part of "csin_downward":
|
|
|
|
|
double: 1
|
|
|
|
|
float: 2
|
|
|
|
|
idouble: 1
|
|
|
|
|
ifloat: 2
|
|
|
|
|
ildouble: 3
|
|
|
|
|
ldouble: 3
|
|
|
|
|
|
|
|
|
|
Function: Real part of "csin_towardzero":
|
|
|
|
|
double: 2
|
|
|
|
|
float: 3
|
|
|
|
|
idouble: 2
|
|
|
|
|
ifloat: 3
|
|
|
|
|
ildouble: 3
|
|
|
|
|
ldouble: 3
|
|
|
|
|
|
|
|
|
|
Function: Imaginary part of "csin_towardzero":
|
|
|
|
|
double: 2
|
|
|
|
|
float: 2
|
|
|
|
|
idouble: 2
|
|
|
|
|
ifloat: 2
|
|
|
|
|
ildouble: 3
|
|
|
|
|
ldouble: 3
|
|
|
|
|
|
|
|
|
|
Function: Real part of "csin_upward":
|
|
|
|
|
double: 1
|
|
|
|
|
float: 3
|
|
|
|
|
idouble: 1
|
|
|
|
|
ifloat: 3
|
|
|
|
|
ildouble: 3
|
|
|
|
|
ldouble: 3
|
|
|
|
|
|
|
|
|
|
Function: Imaginary part of "csin_upward":
|
|
|
|
|
double: 1
|
|
|
|
|
float: 3
|
|
|
|
|
idouble: 1
|
|
|
|
|
ifloat: 3
|
|
|
|
|
ildouble: 3
|
|
|
|
|
ldouble: 3
|
|
|
|
|
|
2001-09-19 10:37:31 +00:00
|
|
|
Function: Real part of "csinh":
|
|
|
|
|
float: 1
|
|
|
|
|
ifloat: 1
|
2002-08-31 08:22:35 +00:00
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
2001-09-19 10:37:31 +00:00
|
|
|
|
|
|
|
|
Function: Imaginary part of "csinh":
|
|
|
|
|
double: 1
|
|
|
|
|
float: 1
|
|
|
|
|
idouble: 1
|
|
|
|
|
ifloat: 1
|
|
|
|
|
|
2014-03-24 12:18:45 +00:00
|
|
|
Function: Real part of "csinh_downward":
|
|
|
|
|
double: 1
|
|
|
|
|
float: 2
|
|
|
|
|
idouble: 1
|
|
|
|
|
ifloat: 2
|
|
|
|
|
ildouble: 3
|
|
|
|
|
ldouble: 3
|
|
|
|
|
|
|
|
|
|
Function: Imaginary part of "csinh_downward":
|
|
|
|
|
double: 2
|
|
|
|
|
float: 3
|
|
|
|
|
idouble: 2
|
|
|
|
|
ifloat: 3
|
|
|
|
|
ildouble: 3
|
|
|
|
|
ldouble: 3
|
|
|
|
|
|
|
|
|
|
Function: Real part of "csinh_towardzero":
|
|
|
|
|
double: 2
|
|
|
|
|
float: 2
|
|
|
|
|
idouble: 2
|
|
|
|
|
ifloat: 2
|
|
|
|
|
ildouble: 3
|
|
|
|
|
ldouble: 3
|
|
|
|
|
|
|
|
|
|
Function: Imaginary part of "csinh_towardzero":
|
|
|
|
|
double: 2
|
|
|
|
|
float: 3
|
|
|
|
|
idouble: 2
|
|
|
|
|
ifloat: 3
|
|
|
|
|
ildouble: 3
|
|
|
|
|
ldouble: 3
|
|
|
|
|
|
|
|
|
|
Function: Real part of "csinh_upward":
|
|
|
|
|
double: 1
|
|
|
|
|
float: 3
|
|
|
|
|
idouble: 1
|
|
|
|
|
ifloat: 3
|
|
|
|
|
ildouble: 3
|
|
|
|
|
ldouble: 3
|
|
|
|
|
|
|
|
|
|
Function: Imaginary part of "csinh_upward":
|
|
|
|
|
double: 2
|
|
|
|
|
float: 3
|
|
|
|
|
idouble: 2
|
|
|
|
|
ifloat: 3
|
|
|
|
|
ildouble: 3
|
|
|
|
|
ldouble: 3
|
|
|
|
|
|
2001-09-19 10:37:31 +00:00
|
|
|
Function: Real part of "csqrt":
|
2012-03-14 11:53:32 +00:00
|
|
|
double: 1
|
|
|
|
|
float: 1
|
|
|
|
|
idouble: 1
|
|
|
|
|
ifloat: 1
|
2012-07-05 11:02:13 +00:00
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
2012-03-14 11:53:32 +00:00
|
|
|
|
|
|
|
|
Function: Imaginary part of "csqrt":
|
|
|
|
|
double: 1
|
2001-09-19 10:37:31 +00:00
|
|
|
float: 1
|
2012-03-14 11:53:32 +00:00
|
|
|
idouble: 1
|
2001-09-19 10:37:31 +00:00
|
|
|
ifloat: 1
|
2012-03-14 11:53:32 +00:00
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
2001-09-19 10:37:31 +00:00
|
|
|
|
Test most libm functions in all rounding modes.
This patch makes libm-test.inc tests of most functions use ALL_RM_TEST
unless there was some reason to defer that change for a particular
function.
I started out planning to defer the change for pow (bug 16315), cexp /
ccos / ccosh / csin / csinh (likely fallout from exp, bug 16284) and
cpow (exact expectations for signs of exact zero results not wanted).
Testing on x86_64 and x86 showed additional failures for acosh, cacos,
catan, catanh, clog, clog10, jn, log, log10, log1p, log2, tgamma, yn,
so making the change for those functions was deferred as well, pending
investigation to show which of these represent distinct bugs (some
such bugs may already be filed) and appropriate fixing / XFAILing.
Failures include wrong signs of zero results, errors slightly above
the 9ulp bound (in such cases it may make sense for functions to set
round-to-nearest internally to reduce error accumulation), large
errors and incorrect overflow/underflow for the rounding mode (with
consequent missing errno settings in some cases). It's possible some
could be issues with test expectations, though I didn't notice any
that were obviously like that (I added NO_TEST_INLINE for cases that
were failing for ildoubl on x86 and where it seemed reasonable for
them to fail for the fast-math inlines).
There may of course be failures on other architectures for functions
that didn't fail on x86_64 or x86, in which case the usual rule
applies: file a bug (preferably identifying the underlying problem
function, in cases where function A calls function B and a problem
with function B may present in the test results for function A) if not
already in Bugzilla then fix or XFAIL.
Tested x86_64 and x86 and ulps updated accordingly.
* math/libm-test.inc (asinh_test): Use ALL_RM_TEST.
(atan_test): Likewise.
(atanh_test_data): Use NO_TEST_INLINE for two tests.
(atanh_test): Use ALL_RM_TEST.
(atan2_test_data): Likewise.
(cabs_test): Likewise.
(cacosh_test): Likewise.
(carg_test): Likewise.
(casin_test): Likewise.
(casinh_test): Likewise.
(cbrt_test): Likewise.
(csqrt_test): Likewise.
(erf_test): Likewise.
(erfc_test): Likewise.
(pow10_test): Likewise.
(exp2_test): Likewise.
(hypot_test): Likewise.
(j0_test): Likewise.
(j1_test): Likewise.
(lgamma_test): Likewise.
(gamma_test): Likewise.
(sincos_test): Likewise.
(tanh_test): Likewise.
(y0_test): Likewise.
(y1_test): Likewise.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2014-03-21 00:03:38 +00:00
|
|
|
Function: Real part of "csqrt_downward":
|
|
|
|
|
double: 3
|
|
|
|
|
float: 3
|
|
|
|
|
idouble: 3
|
|
|
|
|
ifloat: 3
|
|
|
|
|
ildouble: 3
|
|
|
|
|
ldouble: 3
|
|
|
|
|
|
|
|
|
|
Function: Imaginary part of "csqrt_downward":
|
|
|
|
|
double: 2
|
|
|
|
|
float: 2
|
|
|
|
|
idouble: 2
|
|
|
|
|
ifloat: 2
|
|
|
|
|
ildouble: 2
|
|
|
|
|
ldouble: 2
|
|
|
|
|
|
|
|
|
|
Function: Real part of "csqrt_towardzero":
|
|
|
|
|
double: 2
|
|
|
|
|
float: 2
|
|
|
|
|
idouble: 2
|
|
|
|
|
ifloat: 2
|
|
|
|
|
ildouble: 2
|
|
|
|
|
ldouble: 2
|
|
|
|
|
|
|
|
|
|
Function: Imaginary part of "csqrt_towardzero":
|
|
|
|
|
double: 2
|
|
|
|
|
float: 2
|
|
|
|
|
idouble: 2
|
|
|
|
|
ifloat: 2
|
|
|
|
|
ildouble: 2
|
|
|
|
|
ldouble: 2
|
|
|
|
|
|
|
|
|
|
Function: Real part of "csqrt_upward":
|
|
|
|
|
double: 3
|
|
|
|
|
float: 2
|
|
|
|
|
idouble: 3
|
|
|
|
|
ifloat: 2
|
|
|
|
|
ildouble: 3
|
|
|
|
|
ldouble: 3
|
|
|
|
|
|
|
|
|
|
Function: Imaginary part of "csqrt_upward":
|
|
|
|
|
double: 2
|
|
|
|
|
float: 2
|
|
|
|
|
idouble: 2
|
|
|
|
|
ifloat: 2
|
|
|
|
|
ildouble: 2
|
|
|
|
|
ldouble: 2
|
|
|
|
|
|
2001-09-19 10:37:31 +00:00
|
|
|
Function: Real part of "ctan":
|
|
|
|
|
double: 1
|
2013-07-02 20:00:48 +00:00
|
|
|
float: 1
|
2001-09-19 10:37:31 +00:00
|
|
|
idouble: 1
|
2013-07-02 20:00:48 +00:00
|
|
|
ifloat: 1
|
2012-04-09 22:31:35 +00:00
|
|
|
ildouble: 2
|
|
|
|
|
ldouble: 2
|
2001-09-19 10:37:31 +00:00
|
|
|
|
|
|
|
|
Function: Imaginary part of "ctan":
|
Move various TEST_c_c tests from libm-test.inc to auto-libm-test-inc.
This patch moves tests of ccos, ccosh, cexp, clog, csqrt, ctan and
ctanh to auto-libm-test-in, adding the required support to
gen-auto-libm-tests. Other TEST_c_c functions aren't moved for now
(although the relevant table entries are put in gen-auto-libm-tests
for it to know how to handle them): clog10 because of a known MPC bug
causing it to hang for at least some pure imaginary inputs (fixed in
SVN, but I'd rather not rely on unreleased versions of MPFR or MPC
even if relying on very recent releases); the inverse trig and
hyperbolic functions because of known slowness in special cases; and
csin / csinh because of observed slowness that I need to investigate
and report to the MPC maintainers. Slowness can be bypassed by moving
to incremental generation (only for new / changed tests) rather than
regenerating the whole of auto-libm-test-out every time, but that
needs implementing. (This patch takes the time for running
gen-auto-libm-tests from about one second to seven, on my system,
which I think is reasonable. The slow functions would make it take
several minutes at least, which seems unreasonable.)
Tested x86_64 and x86 and ulps updated accordingly.
* math/auto-libm-test-in: Add tests of ccos, ccosh, cexp, clog,
csqrt, ctan and ctanh.
* math/auto-libm-test-out: Regenerated.
* math/libm-test.inc (TEST_COND_x86_64): New macro.
(TEST_COND_x86): Likewise.
(ccos_test_data): Use AUTO_TESTS_c_c.
(ccosh_test_data): Likewise.
(cexp_test_data): Likewise.
(clog_test_data): Likewise.
(csqrt_test_data): Likewise.
(ctan_test_data): Likewise.
(ctan_tonearest_test_data): Likewise.
(ctan_towardzero_test_data): Likewise.
(ctan_downward_test_data): Likewise.
(ctan_upward_test_data): Likewise.
(ctanh_test_data): Likewise.
(ctanh_tonearest_test_data): Likewise.
(ctanh_towardzero_test_data): Likewise.
(ctanh_downward_test_data): Likewise.
(ctanh_upward_test_data): Likewise.
* math/gen-auto-libm-tests.c (func_calc_method): Add value
mpc_c_c.
(func_calc_desc): Add mpc_c_c union field.
(FUNC_mpc_c_c): New macro.
(test_functions): Add cacos, cacosh, casin, casinh, catan, catanh,
ccos, ccosh, cexp, clog, clog10, csin, csinh, csqrt, ctan and
ctanh.
(special_fill_min_subnorm_p120): New function.
(special_real_inputs): Add min_subnorm_p120.
(calc_generic_results): Handle mpc_c_c.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2013-12-20 12:32:44 +00:00
|
|
|
double: 2
|
2012-09-25 18:44:46 +00:00
|
|
|
float: 2
|
Move various TEST_c_c tests from libm-test.inc to auto-libm-test-inc.
This patch moves tests of ccos, ccosh, cexp, clog, csqrt, ctan and
ctanh to auto-libm-test-in, adding the required support to
gen-auto-libm-tests. Other TEST_c_c functions aren't moved for now
(although the relevant table entries are put in gen-auto-libm-tests
for it to know how to handle them): clog10 because of a known MPC bug
causing it to hang for at least some pure imaginary inputs (fixed in
SVN, but I'd rather not rely on unreleased versions of MPFR or MPC
even if relying on very recent releases); the inverse trig and
hyperbolic functions because of known slowness in special cases; and
csin / csinh because of observed slowness that I need to investigate
and report to the MPC maintainers. Slowness can be bypassed by moving
to incremental generation (only for new / changed tests) rather than
regenerating the whole of auto-libm-test-out every time, but that
needs implementing. (This patch takes the time for running
gen-auto-libm-tests from about one second to seven, on my system,
which I think is reasonable. The slow functions would make it take
several minutes at least, which seems unreasonable.)
Tested x86_64 and x86 and ulps updated accordingly.
* math/auto-libm-test-in: Add tests of ccos, ccosh, cexp, clog,
csqrt, ctan and ctanh.
* math/auto-libm-test-out: Regenerated.
* math/libm-test.inc (TEST_COND_x86_64): New macro.
(TEST_COND_x86): Likewise.
(ccos_test_data): Use AUTO_TESTS_c_c.
(ccosh_test_data): Likewise.
(cexp_test_data): Likewise.
(clog_test_data): Likewise.
(csqrt_test_data): Likewise.
(ctan_test_data): Likewise.
(ctan_tonearest_test_data): Likewise.
(ctan_towardzero_test_data): Likewise.
(ctan_downward_test_data): Likewise.
(ctan_upward_test_data): Likewise.
(ctanh_test_data): Likewise.
(ctanh_tonearest_test_data): Likewise.
(ctanh_towardzero_test_data): Likewise.
(ctanh_downward_test_data): Likewise.
(ctanh_upward_test_data): Likewise.
* math/gen-auto-libm-tests.c (func_calc_method): Add value
mpc_c_c.
(func_calc_desc): Add mpc_c_c union field.
(FUNC_mpc_c_c): New macro.
(test_functions): Add cacos, cacosh, casin, casinh, catan, catanh,
ccos, ccosh, cexp, clog, clog10, csin, csinh, csqrt, ctan and
ctanh.
(special_fill_min_subnorm_p120): New function.
(special_real_inputs): Add min_subnorm_p120.
(calc_generic_results): Handle mpc_c_c.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2013-12-20 12:32:44 +00:00
|
|
|
idouble: 2
|
2012-09-25 18:44:46 +00:00
|
|
|
ifloat: 2
|
2012-04-09 22:31:35 +00:00
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
2001-09-19 10:37:31 +00:00
|
|
|
|
2012-07-04 09:55:26 +00:00
|
|
|
Function: Real part of "ctan_downward":
|
Move various TEST_c_c tests from libm-test.inc to auto-libm-test-inc.
This patch moves tests of ccos, ccosh, cexp, clog, csqrt, ctan and
ctanh to auto-libm-test-in, adding the required support to
gen-auto-libm-tests. Other TEST_c_c functions aren't moved for now
(although the relevant table entries are put in gen-auto-libm-tests
for it to know how to handle them): clog10 because of a known MPC bug
causing it to hang for at least some pure imaginary inputs (fixed in
SVN, but I'd rather not rely on unreleased versions of MPFR or MPC
even if relying on very recent releases); the inverse trig and
hyperbolic functions because of known slowness in special cases; and
csin / csinh because of observed slowness that I need to investigate
and report to the MPC maintainers. Slowness can be bypassed by moving
to incremental generation (only for new / changed tests) rather than
regenerating the whole of auto-libm-test-out every time, but that
needs implementing. (This patch takes the time for running
gen-auto-libm-tests from about one second to seven, on my system,
which I think is reasonable. The slow functions would make it take
several minutes at least, which seems unreasonable.)
Tested x86_64 and x86 and ulps updated accordingly.
* math/auto-libm-test-in: Add tests of ccos, ccosh, cexp, clog,
csqrt, ctan and ctanh.
* math/auto-libm-test-out: Regenerated.
* math/libm-test.inc (TEST_COND_x86_64): New macro.
(TEST_COND_x86): Likewise.
(ccos_test_data): Use AUTO_TESTS_c_c.
(ccosh_test_data): Likewise.
(cexp_test_data): Likewise.
(clog_test_data): Likewise.
(csqrt_test_data): Likewise.
(ctan_test_data): Likewise.
(ctan_tonearest_test_data): Likewise.
(ctan_towardzero_test_data): Likewise.
(ctan_downward_test_data): Likewise.
(ctan_upward_test_data): Likewise.
(ctanh_test_data): Likewise.
(ctanh_tonearest_test_data): Likewise.
(ctanh_towardzero_test_data): Likewise.
(ctanh_downward_test_data): Likewise.
(ctanh_upward_test_data): Likewise.
* math/gen-auto-libm-tests.c (func_calc_method): Add value
mpc_c_c.
(func_calc_desc): Add mpc_c_c union field.
(FUNC_mpc_c_c): New macro.
(test_functions): Add cacos, cacosh, casin, casinh, catan, catanh,
ccos, ccosh, cexp, clog, clog10, csin, csinh, csqrt, ctan and
ctanh.
(special_fill_min_subnorm_p120): New function.
(special_real_inputs): Add min_subnorm_p120.
(calc_generic_results): Handle mpc_c_c.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2013-12-20 12:32:44 +00:00
|
|
|
double: 6
|
|
|
|
|
float: 5
|
|
|
|
|
idouble: 6
|
|
|
|
|
ifloat: 5
|
|
|
|
|
ildouble: 5
|
|
|
|
|
ldouble: 5
|
2012-07-04 09:55:26 +00:00
|
|
|
|
|
|
|
|
Function: Imaginary part of "ctan_downward":
|
Move various TEST_c_c tests from libm-test.inc to auto-libm-test-inc.
This patch moves tests of ccos, ccosh, cexp, clog, csqrt, ctan and
ctanh to auto-libm-test-in, adding the required support to
gen-auto-libm-tests. Other TEST_c_c functions aren't moved for now
(although the relevant table entries are put in gen-auto-libm-tests
for it to know how to handle them): clog10 because of a known MPC bug
causing it to hang for at least some pure imaginary inputs (fixed in
SVN, but I'd rather not rely on unreleased versions of MPFR or MPC
even if relying on very recent releases); the inverse trig and
hyperbolic functions because of known slowness in special cases; and
csin / csinh because of observed slowness that I need to investigate
and report to the MPC maintainers. Slowness can be bypassed by moving
to incremental generation (only for new / changed tests) rather than
regenerating the whole of auto-libm-test-out every time, but that
needs implementing. (This patch takes the time for running
gen-auto-libm-tests from about one second to seven, on my system,
which I think is reasonable. The slow functions would make it take
several minutes at least, which seems unreasonable.)
Tested x86_64 and x86 and ulps updated accordingly.
* math/auto-libm-test-in: Add tests of ccos, ccosh, cexp, clog,
csqrt, ctan and ctanh.
* math/auto-libm-test-out: Regenerated.
* math/libm-test.inc (TEST_COND_x86_64): New macro.
(TEST_COND_x86): Likewise.
(ccos_test_data): Use AUTO_TESTS_c_c.
(ccosh_test_data): Likewise.
(cexp_test_data): Likewise.
(clog_test_data): Likewise.
(csqrt_test_data): Likewise.
(ctan_test_data): Likewise.
(ctan_tonearest_test_data): Likewise.
(ctan_towardzero_test_data): Likewise.
(ctan_downward_test_data): Likewise.
(ctan_upward_test_data): Likewise.
(ctanh_test_data): Likewise.
(ctanh_tonearest_test_data): Likewise.
(ctanh_towardzero_test_data): Likewise.
(ctanh_downward_test_data): Likewise.
(ctanh_upward_test_data): Likewise.
* math/gen-auto-libm-tests.c (func_calc_method): Add value
mpc_c_c.
(func_calc_desc): Add mpc_c_c union field.
(FUNC_mpc_c_c): New macro.
(test_functions): Add cacos, cacosh, casin, casinh, catan, catanh,
ccos, ccosh, cexp, clog, clog10, csin, csinh, csqrt, ctan and
ctanh.
(special_fill_min_subnorm_p120): New function.
(special_real_inputs): Add min_subnorm_p120.
(calc_generic_results): Handle mpc_c_c.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2013-12-20 12:32:44 +00:00
|
|
|
double: 2
|
|
|
|
|
float: 2
|
|
|
|
|
idouble: 2
|
|
|
|
|
ifloat: 2
|
2012-07-04 09:55:26 +00:00
|
|
|
ildouble: 4
|
|
|
|
|
ldouble: 4
|
|
|
|
|
|
|
|
|
|
Function: Real part of "ctan_towardzero":
|
Move various TEST_c_c tests from libm-test.inc to auto-libm-test-inc.
This patch moves tests of ccos, ccosh, cexp, clog, csqrt, ctan and
ctanh to auto-libm-test-in, adding the required support to
gen-auto-libm-tests. Other TEST_c_c functions aren't moved for now
(although the relevant table entries are put in gen-auto-libm-tests
for it to know how to handle them): clog10 because of a known MPC bug
causing it to hang for at least some pure imaginary inputs (fixed in
SVN, but I'd rather not rely on unreleased versions of MPFR or MPC
even if relying on very recent releases); the inverse trig and
hyperbolic functions because of known slowness in special cases; and
csin / csinh because of observed slowness that I need to investigate
and report to the MPC maintainers. Slowness can be bypassed by moving
to incremental generation (only for new / changed tests) rather than
regenerating the whole of auto-libm-test-out every time, but that
needs implementing. (This patch takes the time for running
gen-auto-libm-tests from about one second to seven, on my system,
which I think is reasonable. The slow functions would make it take
several minutes at least, which seems unreasonable.)
Tested x86_64 and x86 and ulps updated accordingly.
* math/auto-libm-test-in: Add tests of ccos, ccosh, cexp, clog,
csqrt, ctan and ctanh.
* math/auto-libm-test-out: Regenerated.
* math/libm-test.inc (TEST_COND_x86_64): New macro.
(TEST_COND_x86): Likewise.
(ccos_test_data): Use AUTO_TESTS_c_c.
(ccosh_test_data): Likewise.
(cexp_test_data): Likewise.
(clog_test_data): Likewise.
(csqrt_test_data): Likewise.
(ctan_test_data): Likewise.
(ctan_tonearest_test_data): Likewise.
(ctan_towardzero_test_data): Likewise.
(ctan_downward_test_data): Likewise.
(ctan_upward_test_data): Likewise.
(ctanh_test_data): Likewise.
(ctanh_tonearest_test_data): Likewise.
(ctanh_towardzero_test_data): Likewise.
(ctanh_downward_test_data): Likewise.
(ctanh_upward_test_data): Likewise.
* math/gen-auto-libm-tests.c (func_calc_method): Add value
mpc_c_c.
(func_calc_desc): Add mpc_c_c union field.
(FUNC_mpc_c_c): New macro.
(test_functions): Add cacos, cacosh, casin, casinh, catan, catanh,
ccos, ccosh, cexp, clog, clog10, csin, csinh, csqrt, ctan and
ctanh.
(special_fill_min_subnorm_p120): New function.
(special_real_inputs): Add min_subnorm_p120.
(calc_generic_results): Handle mpc_c_c.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2013-12-20 12:32:44 +00:00
|
|
|
double: 5
|
|
|
|
|
float: 3
|
|
|
|
|
idouble: 5
|
|
|
|
|
ifloat: 3
|
|
|
|
|
ildouble: 5
|
|
|
|
|
ldouble: 5
|
2012-07-04 09:55:26 +00:00
|
|
|
|
|
|
|
|
Function: Imaginary part of "ctan_towardzero":
|
Move various TEST_c_c tests from libm-test.inc to auto-libm-test-inc.
This patch moves tests of ccos, ccosh, cexp, clog, csqrt, ctan and
ctanh to auto-libm-test-in, adding the required support to
gen-auto-libm-tests. Other TEST_c_c functions aren't moved for now
(although the relevant table entries are put in gen-auto-libm-tests
for it to know how to handle them): clog10 because of a known MPC bug
causing it to hang for at least some pure imaginary inputs (fixed in
SVN, but I'd rather not rely on unreleased versions of MPFR or MPC
even if relying on very recent releases); the inverse trig and
hyperbolic functions because of known slowness in special cases; and
csin / csinh because of observed slowness that I need to investigate
and report to the MPC maintainers. Slowness can be bypassed by moving
to incremental generation (only for new / changed tests) rather than
regenerating the whole of auto-libm-test-out every time, but that
needs implementing. (This patch takes the time for running
gen-auto-libm-tests from about one second to seven, on my system,
which I think is reasonable. The slow functions would make it take
several minutes at least, which seems unreasonable.)
Tested x86_64 and x86 and ulps updated accordingly.
* math/auto-libm-test-in: Add tests of ccos, ccosh, cexp, clog,
csqrt, ctan and ctanh.
* math/auto-libm-test-out: Regenerated.
* math/libm-test.inc (TEST_COND_x86_64): New macro.
(TEST_COND_x86): Likewise.
(ccos_test_data): Use AUTO_TESTS_c_c.
(ccosh_test_data): Likewise.
(cexp_test_data): Likewise.
(clog_test_data): Likewise.
(csqrt_test_data): Likewise.
(ctan_test_data): Likewise.
(ctan_tonearest_test_data): Likewise.
(ctan_towardzero_test_data): Likewise.
(ctan_downward_test_data): Likewise.
(ctan_upward_test_data): Likewise.
(ctanh_test_data): Likewise.
(ctanh_tonearest_test_data): Likewise.
(ctanh_towardzero_test_data): Likewise.
(ctanh_downward_test_data): Likewise.
(ctanh_upward_test_data): Likewise.
* math/gen-auto-libm-tests.c (func_calc_method): Add value
mpc_c_c.
(func_calc_desc): Add mpc_c_c union field.
(FUNC_mpc_c_c): New macro.
(test_functions): Add cacos, cacosh, casin, casinh, catan, catanh,
ccos, ccosh, cexp, clog, clog10, csin, csinh, csqrt, ctan and
ctanh.
(special_fill_min_subnorm_p120): New function.
(special_real_inputs): Add min_subnorm_p120.
(calc_generic_results): Handle mpc_c_c.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2013-12-20 12:32:44 +00:00
|
|
|
double: 2
|
|
|
|
|
float: 2
|
|
|
|
|
idouble: 2
|
|
|
|
|
ifloat: 2
|
2012-07-04 09:55:26 +00:00
|
|
|
ildouble: 4
|
|
|
|
|
ldouble: 4
|
|
|
|
|
|
|
|
|
|
Function: Real part of "ctan_upward":
|
|
|
|
|
double: 2
|
Move various TEST_c_c tests from libm-test.inc to auto-libm-test-inc.
This patch moves tests of ccos, ccosh, cexp, clog, csqrt, ctan and
ctanh to auto-libm-test-in, adding the required support to
gen-auto-libm-tests. Other TEST_c_c functions aren't moved for now
(although the relevant table entries are put in gen-auto-libm-tests
for it to know how to handle them): clog10 because of a known MPC bug
causing it to hang for at least some pure imaginary inputs (fixed in
SVN, but I'd rather not rely on unreleased versions of MPFR or MPC
even if relying on very recent releases); the inverse trig and
hyperbolic functions because of known slowness in special cases; and
csin / csinh because of observed slowness that I need to investigate
and report to the MPC maintainers. Slowness can be bypassed by moving
to incremental generation (only for new / changed tests) rather than
regenerating the whole of auto-libm-test-out every time, but that
needs implementing. (This patch takes the time for running
gen-auto-libm-tests from about one second to seven, on my system,
which I think is reasonable. The slow functions would make it take
several minutes at least, which seems unreasonable.)
Tested x86_64 and x86 and ulps updated accordingly.
* math/auto-libm-test-in: Add tests of ccos, ccosh, cexp, clog,
csqrt, ctan and ctanh.
* math/auto-libm-test-out: Regenerated.
* math/libm-test.inc (TEST_COND_x86_64): New macro.
(TEST_COND_x86): Likewise.
(ccos_test_data): Use AUTO_TESTS_c_c.
(ccosh_test_data): Likewise.
(cexp_test_data): Likewise.
(clog_test_data): Likewise.
(csqrt_test_data): Likewise.
(ctan_test_data): Likewise.
(ctan_tonearest_test_data): Likewise.
(ctan_towardzero_test_data): Likewise.
(ctan_downward_test_data): Likewise.
(ctan_upward_test_data): Likewise.
(ctanh_test_data): Likewise.
(ctanh_tonearest_test_data): Likewise.
(ctanh_towardzero_test_data): Likewise.
(ctanh_downward_test_data): Likewise.
(ctanh_upward_test_data): Likewise.
* math/gen-auto-libm-tests.c (func_calc_method): Add value
mpc_c_c.
(func_calc_desc): Add mpc_c_c union field.
(FUNC_mpc_c_c): New macro.
(test_functions): Add cacos, cacosh, casin, casinh, catan, catanh,
ccos, ccosh, cexp, clog, clog10, csin, csinh, csqrt, ctan and
ctanh.
(special_fill_min_subnorm_p120): New function.
(special_real_inputs): Add min_subnorm_p120.
(calc_generic_results): Handle mpc_c_c.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2013-12-20 12:32:44 +00:00
|
|
|
float: 4
|
2012-07-04 09:55:26 +00:00
|
|
|
idouble: 2
|
Move various TEST_c_c tests from libm-test.inc to auto-libm-test-inc.
This patch moves tests of ccos, ccosh, cexp, clog, csqrt, ctan and
ctanh to auto-libm-test-in, adding the required support to
gen-auto-libm-tests. Other TEST_c_c functions aren't moved for now
(although the relevant table entries are put in gen-auto-libm-tests
for it to know how to handle them): clog10 because of a known MPC bug
causing it to hang for at least some pure imaginary inputs (fixed in
SVN, but I'd rather not rely on unreleased versions of MPFR or MPC
even if relying on very recent releases); the inverse trig and
hyperbolic functions because of known slowness in special cases; and
csin / csinh because of observed slowness that I need to investigate
and report to the MPC maintainers. Slowness can be bypassed by moving
to incremental generation (only for new / changed tests) rather than
regenerating the whole of auto-libm-test-out every time, but that
needs implementing. (This patch takes the time for running
gen-auto-libm-tests from about one second to seven, on my system,
which I think is reasonable. The slow functions would make it take
several minutes at least, which seems unreasonable.)
Tested x86_64 and x86 and ulps updated accordingly.
* math/auto-libm-test-in: Add tests of ccos, ccosh, cexp, clog,
csqrt, ctan and ctanh.
* math/auto-libm-test-out: Regenerated.
* math/libm-test.inc (TEST_COND_x86_64): New macro.
(TEST_COND_x86): Likewise.
(ccos_test_data): Use AUTO_TESTS_c_c.
(ccosh_test_data): Likewise.
(cexp_test_data): Likewise.
(clog_test_data): Likewise.
(csqrt_test_data): Likewise.
(ctan_test_data): Likewise.
(ctan_tonearest_test_data): Likewise.
(ctan_towardzero_test_data): Likewise.
(ctan_downward_test_data): Likewise.
(ctan_upward_test_data): Likewise.
(ctanh_test_data): Likewise.
(ctanh_tonearest_test_data): Likewise.
(ctanh_towardzero_test_data): Likewise.
(ctanh_downward_test_data): Likewise.
(ctanh_upward_test_data): Likewise.
* math/gen-auto-libm-tests.c (func_calc_method): Add value
mpc_c_c.
(func_calc_desc): Add mpc_c_c union field.
(FUNC_mpc_c_c): New macro.
(test_functions): Add cacos, cacosh, casin, casinh, catan, catanh,
ccos, ccosh, cexp, clog, clog10, csin, csinh, csqrt, ctan and
ctanh.
(special_fill_min_subnorm_p120): New function.
(special_real_inputs): Add min_subnorm_p120.
(calc_generic_results): Handle mpc_c_c.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2013-12-20 12:32:44 +00:00
|
|
|
ifloat: 4
|
|
|
|
|
ildouble: 3
|
|
|
|
|
ldouble: 3
|
2012-07-04 09:55:26 +00:00
|
|
|
|
|
|
|
|
Function: Imaginary part of "ctan_upward":
|
Move various TEST_c_c tests from libm-test.inc to auto-libm-test-inc.
This patch moves tests of ccos, ccosh, cexp, clog, csqrt, ctan and
ctanh to auto-libm-test-in, adding the required support to
gen-auto-libm-tests. Other TEST_c_c functions aren't moved for now
(although the relevant table entries are put in gen-auto-libm-tests
for it to know how to handle them): clog10 because of a known MPC bug
causing it to hang for at least some pure imaginary inputs (fixed in
SVN, but I'd rather not rely on unreleased versions of MPFR or MPC
even if relying on very recent releases); the inverse trig and
hyperbolic functions because of known slowness in special cases; and
csin / csinh because of observed slowness that I need to investigate
and report to the MPC maintainers. Slowness can be bypassed by moving
to incremental generation (only for new / changed tests) rather than
regenerating the whole of auto-libm-test-out every time, but that
needs implementing. (This patch takes the time for running
gen-auto-libm-tests from about one second to seven, on my system,
which I think is reasonable. The slow functions would make it take
several minutes at least, which seems unreasonable.)
Tested x86_64 and x86 and ulps updated accordingly.
* math/auto-libm-test-in: Add tests of ccos, ccosh, cexp, clog,
csqrt, ctan and ctanh.
* math/auto-libm-test-out: Regenerated.
* math/libm-test.inc (TEST_COND_x86_64): New macro.
(TEST_COND_x86): Likewise.
(ccos_test_data): Use AUTO_TESTS_c_c.
(ccosh_test_data): Likewise.
(cexp_test_data): Likewise.
(clog_test_data): Likewise.
(csqrt_test_data): Likewise.
(ctan_test_data): Likewise.
(ctan_tonearest_test_data): Likewise.
(ctan_towardzero_test_data): Likewise.
(ctan_downward_test_data): Likewise.
(ctan_upward_test_data): Likewise.
(ctanh_test_data): Likewise.
(ctanh_tonearest_test_data): Likewise.
(ctanh_towardzero_test_data): Likewise.
(ctanh_downward_test_data): Likewise.
(ctanh_upward_test_data): Likewise.
* math/gen-auto-libm-tests.c (func_calc_method): Add value
mpc_c_c.
(func_calc_desc): Add mpc_c_c union field.
(FUNC_mpc_c_c): New macro.
(test_functions): Add cacos, cacosh, casin, casinh, catan, catanh,
ccos, ccosh, cexp, clog, clog10, csin, csinh, csqrt, ctan and
ctanh.
(special_fill_min_subnorm_p120): New function.
(special_real_inputs): Add min_subnorm_p120.
(calc_generic_results): Handle mpc_c_c.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2013-12-20 12:32:44 +00:00
|
|
|
double: 2
|
|
|
|
|
float: 1
|
|
|
|
|
idouble: 2
|
|
|
|
|
ifloat: 1
|
|
|
|
|
ildouble: 3
|
|
|
|
|
ldouble: 3
|
2012-07-04 09:55:26 +00:00
|
|
|
|
2001-09-19 10:37:31 +00:00
|
|
|
Function: Real part of "ctanh":
|
Move various TEST_c_c tests from libm-test.inc to auto-libm-test-inc.
This patch moves tests of ccos, ccosh, cexp, clog, csqrt, ctan and
ctanh to auto-libm-test-in, adding the required support to
gen-auto-libm-tests. Other TEST_c_c functions aren't moved for now
(although the relevant table entries are put in gen-auto-libm-tests
for it to know how to handle them): clog10 because of a known MPC bug
causing it to hang for at least some pure imaginary inputs (fixed in
SVN, but I'd rather not rely on unreleased versions of MPFR or MPC
even if relying on very recent releases); the inverse trig and
hyperbolic functions because of known slowness in special cases; and
csin / csinh because of observed slowness that I need to investigate
and report to the MPC maintainers. Slowness can be bypassed by moving
to incremental generation (only for new / changed tests) rather than
regenerating the whole of auto-libm-test-out every time, but that
needs implementing. (This patch takes the time for running
gen-auto-libm-tests from about one second to seven, on my system,
which I think is reasonable. The slow functions would make it take
several minutes at least, which seems unreasonable.)
Tested x86_64 and x86 and ulps updated accordingly.
* math/auto-libm-test-in: Add tests of ccos, ccosh, cexp, clog,
csqrt, ctan and ctanh.
* math/auto-libm-test-out: Regenerated.
* math/libm-test.inc (TEST_COND_x86_64): New macro.
(TEST_COND_x86): Likewise.
(ccos_test_data): Use AUTO_TESTS_c_c.
(ccosh_test_data): Likewise.
(cexp_test_data): Likewise.
(clog_test_data): Likewise.
(csqrt_test_data): Likewise.
(ctan_test_data): Likewise.
(ctan_tonearest_test_data): Likewise.
(ctan_towardzero_test_data): Likewise.
(ctan_downward_test_data): Likewise.
(ctan_upward_test_data): Likewise.
(ctanh_test_data): Likewise.
(ctanh_tonearest_test_data): Likewise.
(ctanh_towardzero_test_data): Likewise.
(ctanh_downward_test_data): Likewise.
(ctanh_upward_test_data): Likewise.
* math/gen-auto-libm-tests.c (func_calc_method): Add value
mpc_c_c.
(func_calc_desc): Add mpc_c_c union field.
(FUNC_mpc_c_c): New macro.
(test_functions): Add cacos, cacosh, casin, casinh, catan, catanh,
ccos, ccosh, cexp, clog, clog10, csin, csinh, csqrt, ctan and
ctanh.
(special_fill_min_subnorm_p120): New function.
(special_real_inputs): Add min_subnorm_p120.
(calc_generic_results): Handle mpc_c_c.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2013-12-20 12:32:44 +00:00
|
|
|
double: 2
|
2012-09-25 18:44:46 +00:00
|
|
|
float: 2
|
Move various TEST_c_c tests from libm-test.inc to auto-libm-test-inc.
This patch moves tests of ccos, ccosh, cexp, clog, csqrt, ctan and
ctanh to auto-libm-test-in, adding the required support to
gen-auto-libm-tests. Other TEST_c_c functions aren't moved for now
(although the relevant table entries are put in gen-auto-libm-tests
for it to know how to handle them): clog10 because of a known MPC bug
causing it to hang for at least some pure imaginary inputs (fixed in
SVN, but I'd rather not rely on unreleased versions of MPFR or MPC
even if relying on very recent releases); the inverse trig and
hyperbolic functions because of known slowness in special cases; and
csin / csinh because of observed slowness that I need to investigate
and report to the MPC maintainers. Slowness can be bypassed by moving
to incremental generation (only for new / changed tests) rather than
regenerating the whole of auto-libm-test-out every time, but that
needs implementing. (This patch takes the time for running
gen-auto-libm-tests from about one second to seven, on my system,
which I think is reasonable. The slow functions would make it take
several minutes at least, which seems unreasonable.)
Tested x86_64 and x86 and ulps updated accordingly.
* math/auto-libm-test-in: Add tests of ccos, ccosh, cexp, clog,
csqrt, ctan and ctanh.
* math/auto-libm-test-out: Regenerated.
* math/libm-test.inc (TEST_COND_x86_64): New macro.
(TEST_COND_x86): Likewise.
(ccos_test_data): Use AUTO_TESTS_c_c.
(ccosh_test_data): Likewise.
(cexp_test_data): Likewise.
(clog_test_data): Likewise.
(csqrt_test_data): Likewise.
(ctan_test_data): Likewise.
(ctan_tonearest_test_data): Likewise.
(ctan_towardzero_test_data): Likewise.
(ctan_downward_test_data): Likewise.
(ctan_upward_test_data): Likewise.
(ctanh_test_data): Likewise.
(ctanh_tonearest_test_data): Likewise.
(ctanh_towardzero_test_data): Likewise.
(ctanh_downward_test_data): Likewise.
(ctanh_upward_test_data): Likewise.
* math/gen-auto-libm-tests.c (func_calc_method): Add value
mpc_c_c.
(func_calc_desc): Add mpc_c_c union field.
(FUNC_mpc_c_c): New macro.
(test_functions): Add cacos, cacosh, casin, casinh, catan, catanh,
ccos, ccosh, cexp, clog, clog10, csin, csinh, csqrt, ctan and
ctanh.
(special_fill_min_subnorm_p120): New function.
(special_real_inputs): Add min_subnorm_p120.
(calc_generic_results): Handle mpc_c_c.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2013-12-20 12:32:44 +00:00
|
|
|
idouble: 2
|
2012-09-25 18:44:46 +00:00
|
|
|
ifloat: 2
|
2012-02-29 20:40:50 +00:00
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
2001-09-19 10:37:31 +00:00
|
|
|
|
2012-04-09 22:31:35 +00:00
|
|
|
Function: Imaginary part of "ctanh":
|
Move various TEST_c_c tests from libm-test.inc to auto-libm-test-inc.
This patch moves tests of ccos, ccosh, cexp, clog, csqrt, ctan and
ctanh to auto-libm-test-in, adding the required support to
gen-auto-libm-tests. Other TEST_c_c functions aren't moved for now
(although the relevant table entries are put in gen-auto-libm-tests
for it to know how to handle them): clog10 because of a known MPC bug
causing it to hang for at least some pure imaginary inputs (fixed in
SVN, but I'd rather not rely on unreleased versions of MPFR or MPC
even if relying on very recent releases); the inverse trig and
hyperbolic functions because of known slowness in special cases; and
csin / csinh because of observed slowness that I need to investigate
and report to the MPC maintainers. Slowness can be bypassed by moving
to incremental generation (only for new / changed tests) rather than
regenerating the whole of auto-libm-test-out every time, but that
needs implementing. (This patch takes the time for running
gen-auto-libm-tests from about one second to seven, on my system,
which I think is reasonable. The slow functions would make it take
several minutes at least, which seems unreasonable.)
Tested x86_64 and x86 and ulps updated accordingly.
* math/auto-libm-test-in: Add tests of ccos, ccosh, cexp, clog,
csqrt, ctan and ctanh.
* math/auto-libm-test-out: Regenerated.
* math/libm-test.inc (TEST_COND_x86_64): New macro.
(TEST_COND_x86): Likewise.
(ccos_test_data): Use AUTO_TESTS_c_c.
(ccosh_test_data): Likewise.
(cexp_test_data): Likewise.
(clog_test_data): Likewise.
(csqrt_test_data): Likewise.
(ctan_test_data): Likewise.
(ctan_tonearest_test_data): Likewise.
(ctan_towardzero_test_data): Likewise.
(ctan_downward_test_data): Likewise.
(ctan_upward_test_data): Likewise.
(ctanh_test_data): Likewise.
(ctanh_tonearest_test_data): Likewise.
(ctanh_towardzero_test_data): Likewise.
(ctanh_downward_test_data): Likewise.
(ctanh_upward_test_data): Likewise.
* math/gen-auto-libm-tests.c (func_calc_method): Add value
mpc_c_c.
(func_calc_desc): Add mpc_c_c union field.
(FUNC_mpc_c_c): New macro.
(test_functions): Add cacos, cacosh, casin, casinh, catan, catanh,
ccos, ccosh, cexp, clog, clog10, csin, csinh, csqrt, ctan and
ctanh.
(special_fill_min_subnorm_p120): New function.
(special_real_inputs): Add min_subnorm_p120.
(calc_generic_results): Handle mpc_c_c.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2013-12-20 12:32:44 +00:00
|
|
|
double: 2
|
2012-04-09 22:31:35 +00:00
|
|
|
float: 2
|
Move various TEST_c_c tests from libm-test.inc to auto-libm-test-inc.
This patch moves tests of ccos, ccosh, cexp, clog, csqrt, ctan and
ctanh to auto-libm-test-in, adding the required support to
gen-auto-libm-tests. Other TEST_c_c functions aren't moved for now
(although the relevant table entries are put in gen-auto-libm-tests
for it to know how to handle them): clog10 because of a known MPC bug
causing it to hang for at least some pure imaginary inputs (fixed in
SVN, but I'd rather not rely on unreleased versions of MPFR or MPC
even if relying on very recent releases); the inverse trig and
hyperbolic functions because of known slowness in special cases; and
csin / csinh because of observed slowness that I need to investigate
and report to the MPC maintainers. Slowness can be bypassed by moving
to incremental generation (only for new / changed tests) rather than
regenerating the whole of auto-libm-test-out every time, but that
needs implementing. (This patch takes the time for running
gen-auto-libm-tests from about one second to seven, on my system,
which I think is reasonable. The slow functions would make it take
several minutes at least, which seems unreasonable.)
Tested x86_64 and x86 and ulps updated accordingly.
* math/auto-libm-test-in: Add tests of ccos, ccosh, cexp, clog,
csqrt, ctan and ctanh.
* math/auto-libm-test-out: Regenerated.
* math/libm-test.inc (TEST_COND_x86_64): New macro.
(TEST_COND_x86): Likewise.
(ccos_test_data): Use AUTO_TESTS_c_c.
(ccosh_test_data): Likewise.
(cexp_test_data): Likewise.
(clog_test_data): Likewise.
(csqrt_test_data): Likewise.
(ctan_test_data): Likewise.
(ctan_tonearest_test_data): Likewise.
(ctan_towardzero_test_data): Likewise.
(ctan_downward_test_data): Likewise.
(ctan_upward_test_data): Likewise.
(ctanh_test_data): Likewise.
(ctanh_tonearest_test_data): Likewise.
(ctanh_towardzero_test_data): Likewise.
(ctanh_downward_test_data): Likewise.
(ctanh_upward_test_data): Likewise.
* math/gen-auto-libm-tests.c (func_calc_method): Add value
mpc_c_c.
(func_calc_desc): Add mpc_c_c union field.
(FUNC_mpc_c_c): New macro.
(test_functions): Add cacos, cacosh, casin, casinh, catan, catanh,
ccos, ccosh, cexp, clog, clog10, csin, csinh, csqrt, ctan and
ctanh.
(special_fill_min_subnorm_p120): New function.
(special_real_inputs): Add min_subnorm_p120.
(calc_generic_results): Handle mpc_c_c.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2013-12-20 12:32:44 +00:00
|
|
|
idouble: 2
|
2012-04-09 22:31:35 +00:00
|
|
|
ifloat: 2
|
|
|
|
|
ildouble: 2
|
|
|
|
|
ldouble: 2
|
|
|
|
|
|
2012-07-04 09:55:26 +00:00
|
|
|
Function: Real part of "ctanh_downward":
|
Move various TEST_c_c tests from libm-test.inc to auto-libm-test-inc.
This patch moves tests of ccos, ccosh, cexp, clog, csqrt, ctan and
ctanh to auto-libm-test-in, adding the required support to
gen-auto-libm-tests. Other TEST_c_c functions aren't moved for now
(although the relevant table entries are put in gen-auto-libm-tests
for it to know how to handle them): clog10 because of a known MPC bug
causing it to hang for at least some pure imaginary inputs (fixed in
SVN, but I'd rather not rely on unreleased versions of MPFR or MPC
even if relying on very recent releases); the inverse trig and
hyperbolic functions because of known slowness in special cases; and
csin / csinh because of observed slowness that I need to investigate
and report to the MPC maintainers. Slowness can be bypassed by moving
to incremental generation (only for new / changed tests) rather than
regenerating the whole of auto-libm-test-out every time, but that
needs implementing. (This patch takes the time for running
gen-auto-libm-tests from about one second to seven, on my system,
which I think is reasonable. The slow functions would make it take
several minutes at least, which seems unreasonable.)
Tested x86_64 and x86 and ulps updated accordingly.
* math/auto-libm-test-in: Add tests of ccos, ccosh, cexp, clog,
csqrt, ctan and ctanh.
* math/auto-libm-test-out: Regenerated.
* math/libm-test.inc (TEST_COND_x86_64): New macro.
(TEST_COND_x86): Likewise.
(ccos_test_data): Use AUTO_TESTS_c_c.
(ccosh_test_data): Likewise.
(cexp_test_data): Likewise.
(clog_test_data): Likewise.
(csqrt_test_data): Likewise.
(ctan_test_data): Likewise.
(ctan_tonearest_test_data): Likewise.
(ctan_towardzero_test_data): Likewise.
(ctan_downward_test_data): Likewise.
(ctan_upward_test_data): Likewise.
(ctanh_test_data): Likewise.
(ctanh_tonearest_test_data): Likewise.
(ctanh_towardzero_test_data): Likewise.
(ctanh_downward_test_data): Likewise.
(ctanh_upward_test_data): Likewise.
* math/gen-auto-libm-tests.c (func_calc_method): Add value
mpc_c_c.
(func_calc_desc): Add mpc_c_c union field.
(FUNC_mpc_c_c): New macro.
(test_functions): Add cacos, cacosh, casin, casinh, catan, catanh,
ccos, ccosh, cexp, clog, clog10, csin, csinh, csqrt, ctan and
ctanh.
(special_fill_min_subnorm_p120): New function.
(special_real_inputs): Add min_subnorm_p120.
(calc_generic_results): Handle mpc_c_c.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2013-12-20 12:32:44 +00:00
|
|
|
double: 4
|
|
|
|
|
float: 2
|
|
|
|
|
idouble: 4
|
|
|
|
|
ifloat: 2
|
2012-07-04 09:55:26 +00:00
|
|
|
ildouble: 4
|
|
|
|
|
ldouble: 4
|
|
|
|
|
|
|
|
|
|
Function: Imaginary part of "ctanh_downward":
|
Move various TEST_c_c tests from libm-test.inc to auto-libm-test-inc.
This patch moves tests of ccos, ccosh, cexp, clog, csqrt, ctan and
ctanh to auto-libm-test-in, adding the required support to
gen-auto-libm-tests. Other TEST_c_c functions aren't moved for now
(although the relevant table entries are put in gen-auto-libm-tests
for it to know how to handle them): clog10 because of a known MPC bug
causing it to hang for at least some pure imaginary inputs (fixed in
SVN, but I'd rather not rely on unreleased versions of MPFR or MPC
even if relying on very recent releases); the inverse trig and
hyperbolic functions because of known slowness in special cases; and
csin / csinh because of observed slowness that I need to investigate
and report to the MPC maintainers. Slowness can be bypassed by moving
to incremental generation (only for new / changed tests) rather than
regenerating the whole of auto-libm-test-out every time, but that
needs implementing. (This patch takes the time for running
gen-auto-libm-tests from about one second to seven, on my system,
which I think is reasonable. The slow functions would make it take
several minutes at least, which seems unreasonable.)
Tested x86_64 and x86 and ulps updated accordingly.
* math/auto-libm-test-in: Add tests of ccos, ccosh, cexp, clog,
csqrt, ctan and ctanh.
* math/auto-libm-test-out: Regenerated.
* math/libm-test.inc (TEST_COND_x86_64): New macro.
(TEST_COND_x86): Likewise.
(ccos_test_data): Use AUTO_TESTS_c_c.
(ccosh_test_data): Likewise.
(cexp_test_data): Likewise.
(clog_test_data): Likewise.
(csqrt_test_data): Likewise.
(ctan_test_data): Likewise.
(ctan_tonearest_test_data): Likewise.
(ctan_towardzero_test_data): Likewise.
(ctan_downward_test_data): Likewise.
(ctan_upward_test_data): Likewise.
(ctanh_test_data): Likewise.
(ctanh_tonearest_test_data): Likewise.
(ctanh_towardzero_test_data): Likewise.
(ctanh_downward_test_data): Likewise.
(ctanh_upward_test_data): Likewise.
* math/gen-auto-libm-tests.c (func_calc_method): Add value
mpc_c_c.
(func_calc_desc): Add mpc_c_c union field.
(FUNC_mpc_c_c): New macro.
(test_functions): Add cacos, cacosh, casin, casinh, catan, catanh,
ccos, ccosh, cexp, clog, clog10, csin, csinh, csqrt, ctan and
ctanh.
(special_fill_min_subnorm_p120): New function.
(special_real_inputs): Add min_subnorm_p120.
(calc_generic_results): Handle mpc_c_c.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2013-12-20 12:32:44 +00:00
|
|
|
double: 6
|
|
|
|
|
float: 5
|
|
|
|
|
idouble: 6
|
|
|
|
|
ifloat: 5
|
|
|
|
|
ildouble: 4
|
|
|
|
|
ldouble: 4
|
2012-07-04 09:55:26 +00:00
|
|
|
|
|
|
|
|
Function: Real part of "ctanh_towardzero":
|
Move various TEST_c_c tests from libm-test.inc to auto-libm-test-inc.
This patch moves tests of ccos, ccosh, cexp, clog, csqrt, ctan and
ctanh to auto-libm-test-in, adding the required support to
gen-auto-libm-tests. Other TEST_c_c functions aren't moved for now
(although the relevant table entries are put in gen-auto-libm-tests
for it to know how to handle them): clog10 because of a known MPC bug
causing it to hang for at least some pure imaginary inputs (fixed in
SVN, but I'd rather not rely on unreleased versions of MPFR or MPC
even if relying on very recent releases); the inverse trig and
hyperbolic functions because of known slowness in special cases; and
csin / csinh because of observed slowness that I need to investigate
and report to the MPC maintainers. Slowness can be bypassed by moving
to incremental generation (only for new / changed tests) rather than
regenerating the whole of auto-libm-test-out every time, but that
needs implementing. (This patch takes the time for running
gen-auto-libm-tests from about one second to seven, on my system,
which I think is reasonable. The slow functions would make it take
several minutes at least, which seems unreasonable.)
Tested x86_64 and x86 and ulps updated accordingly.
* math/auto-libm-test-in: Add tests of ccos, ccosh, cexp, clog,
csqrt, ctan and ctanh.
* math/auto-libm-test-out: Regenerated.
* math/libm-test.inc (TEST_COND_x86_64): New macro.
(TEST_COND_x86): Likewise.
(ccos_test_data): Use AUTO_TESTS_c_c.
(ccosh_test_data): Likewise.
(cexp_test_data): Likewise.
(clog_test_data): Likewise.
(csqrt_test_data): Likewise.
(ctan_test_data): Likewise.
(ctan_tonearest_test_data): Likewise.
(ctan_towardzero_test_data): Likewise.
(ctan_downward_test_data): Likewise.
(ctan_upward_test_data): Likewise.
(ctanh_test_data): Likewise.
(ctanh_tonearest_test_data): Likewise.
(ctanh_towardzero_test_data): Likewise.
(ctanh_downward_test_data): Likewise.
(ctanh_upward_test_data): Likewise.
* math/gen-auto-libm-tests.c (func_calc_method): Add value
mpc_c_c.
(func_calc_desc): Add mpc_c_c union field.
(FUNC_mpc_c_c): New macro.
(test_functions): Add cacos, cacosh, casin, casinh, catan, catanh,
ccos, ccosh, cexp, clog, clog10, csin, csinh, csqrt, ctan and
ctanh.
(special_fill_min_subnorm_p120): New function.
(special_real_inputs): Add min_subnorm_p120.
(calc_generic_results): Handle mpc_c_c.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2013-12-20 12:32:44 +00:00
|
|
|
double: 2
|
|
|
|
|
float: 2
|
|
|
|
|
idouble: 2
|
|
|
|
|
ifloat: 2
|
2012-07-04 09:55:26 +00:00
|
|
|
ildouble: 4
|
|
|
|
|
ldouble: 4
|
|
|
|
|
|
|
|
|
|
Function: Imaginary part of "ctanh_towardzero":
|
Move various TEST_c_c tests from libm-test.inc to auto-libm-test-inc.
This patch moves tests of ccos, ccosh, cexp, clog, csqrt, ctan and
ctanh to auto-libm-test-in, adding the required support to
gen-auto-libm-tests. Other TEST_c_c functions aren't moved for now
(although the relevant table entries are put in gen-auto-libm-tests
for it to know how to handle them): clog10 because of a known MPC bug
causing it to hang for at least some pure imaginary inputs (fixed in
SVN, but I'd rather not rely on unreleased versions of MPFR or MPC
even if relying on very recent releases); the inverse trig and
hyperbolic functions because of known slowness in special cases; and
csin / csinh because of observed slowness that I need to investigate
and report to the MPC maintainers. Slowness can be bypassed by moving
to incremental generation (only for new / changed tests) rather than
regenerating the whole of auto-libm-test-out every time, but that
needs implementing. (This patch takes the time for running
gen-auto-libm-tests from about one second to seven, on my system,
which I think is reasonable. The slow functions would make it take
several minutes at least, which seems unreasonable.)
Tested x86_64 and x86 and ulps updated accordingly.
* math/auto-libm-test-in: Add tests of ccos, ccosh, cexp, clog,
csqrt, ctan and ctanh.
* math/auto-libm-test-out: Regenerated.
* math/libm-test.inc (TEST_COND_x86_64): New macro.
(TEST_COND_x86): Likewise.
(ccos_test_data): Use AUTO_TESTS_c_c.
(ccosh_test_data): Likewise.
(cexp_test_data): Likewise.
(clog_test_data): Likewise.
(csqrt_test_data): Likewise.
(ctan_test_data): Likewise.
(ctan_tonearest_test_data): Likewise.
(ctan_towardzero_test_data): Likewise.
(ctan_downward_test_data): Likewise.
(ctan_upward_test_data): Likewise.
(ctanh_test_data): Likewise.
(ctanh_tonearest_test_data): Likewise.
(ctanh_towardzero_test_data): Likewise.
(ctanh_downward_test_data): Likewise.
(ctanh_upward_test_data): Likewise.
* math/gen-auto-libm-tests.c (func_calc_method): Add value
mpc_c_c.
(func_calc_desc): Add mpc_c_c union field.
(FUNC_mpc_c_c): New macro.
(test_functions): Add cacos, cacosh, casin, casinh, catan, catanh,
ccos, ccosh, cexp, clog, clog10, csin, csinh, csqrt, ctan and
ctanh.
(special_fill_min_subnorm_p120): New function.
(special_real_inputs): Add min_subnorm_p120.
(calc_generic_results): Handle mpc_c_c.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2013-12-20 12:32:44 +00:00
|
|
|
double: 5
|
|
|
|
|
float: 3
|
|
|
|
|
idouble: 5
|
|
|
|
|
ifloat: 3
|
|
|
|
|
ildouble: 3
|
|
|
|
|
ldouble: 3
|
2012-07-04 09:55:26 +00:00
|
|
|
|
|
|
|
|
Function: Real part of "ctanh_upward":
|
Move various TEST_c_c tests from libm-test.inc to auto-libm-test-inc.
This patch moves tests of ccos, ccosh, cexp, clog, csqrt, ctan and
ctanh to auto-libm-test-in, adding the required support to
gen-auto-libm-tests. Other TEST_c_c functions aren't moved for now
(although the relevant table entries are put in gen-auto-libm-tests
for it to know how to handle them): clog10 because of a known MPC bug
causing it to hang for at least some pure imaginary inputs (fixed in
SVN, but I'd rather not rely on unreleased versions of MPFR or MPC
even if relying on very recent releases); the inverse trig and
hyperbolic functions because of known slowness in special cases; and
csin / csinh because of observed slowness that I need to investigate
and report to the MPC maintainers. Slowness can be bypassed by moving
to incremental generation (only for new / changed tests) rather than
regenerating the whole of auto-libm-test-out every time, but that
needs implementing. (This patch takes the time for running
gen-auto-libm-tests from about one second to seven, on my system,
which I think is reasonable. The slow functions would make it take
several minutes at least, which seems unreasonable.)
Tested x86_64 and x86 and ulps updated accordingly.
* math/auto-libm-test-in: Add tests of ccos, ccosh, cexp, clog,
csqrt, ctan and ctanh.
* math/auto-libm-test-out: Regenerated.
* math/libm-test.inc (TEST_COND_x86_64): New macro.
(TEST_COND_x86): Likewise.
(ccos_test_data): Use AUTO_TESTS_c_c.
(ccosh_test_data): Likewise.
(cexp_test_data): Likewise.
(clog_test_data): Likewise.
(csqrt_test_data): Likewise.
(ctan_test_data): Likewise.
(ctan_tonearest_test_data): Likewise.
(ctan_towardzero_test_data): Likewise.
(ctan_downward_test_data): Likewise.
(ctan_upward_test_data): Likewise.
(ctanh_test_data): Likewise.
(ctanh_tonearest_test_data): Likewise.
(ctanh_towardzero_test_data): Likewise.
(ctanh_downward_test_data): Likewise.
(ctanh_upward_test_data): Likewise.
* math/gen-auto-libm-tests.c (func_calc_method): Add value
mpc_c_c.
(func_calc_desc): Add mpc_c_c union field.
(FUNC_mpc_c_c): New macro.
(test_functions): Add cacos, cacosh, casin, casinh, catan, catanh,
ccos, ccosh, cexp, clog, clog10, csin, csinh, csqrt, ctan and
ctanh.
(special_fill_min_subnorm_p120): New function.
(special_real_inputs): Add min_subnorm_p120.
(calc_generic_results): Handle mpc_c_c.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2013-12-20 12:32:44 +00:00
|
|
|
double: 2
|
|
|
|
|
float: 2
|
|
|
|
|
idouble: 2
|
|
|
|
|
ifloat: 2
|
|
|
|
|
ildouble: 3
|
|
|
|
|
ldouble: 3
|
2012-07-04 09:55:26 +00:00
|
|
|
|
|
|
|
|
Function: Imaginary part of "ctanh_upward":
|
|
|
|
|
double: 2
|
Move various TEST_c_c tests from libm-test.inc to auto-libm-test-inc.
This patch moves tests of ccos, ccosh, cexp, clog, csqrt, ctan and
ctanh to auto-libm-test-in, adding the required support to
gen-auto-libm-tests. Other TEST_c_c functions aren't moved for now
(although the relevant table entries are put in gen-auto-libm-tests
for it to know how to handle them): clog10 because of a known MPC bug
causing it to hang for at least some pure imaginary inputs (fixed in
SVN, but I'd rather not rely on unreleased versions of MPFR or MPC
even if relying on very recent releases); the inverse trig and
hyperbolic functions because of known slowness in special cases; and
csin / csinh because of observed slowness that I need to investigate
and report to the MPC maintainers. Slowness can be bypassed by moving
to incremental generation (only for new / changed tests) rather than
regenerating the whole of auto-libm-test-out every time, but that
needs implementing. (This patch takes the time for running
gen-auto-libm-tests from about one second to seven, on my system,
which I think is reasonable. The slow functions would make it take
several minutes at least, which seems unreasonable.)
Tested x86_64 and x86 and ulps updated accordingly.
* math/auto-libm-test-in: Add tests of ccos, ccosh, cexp, clog,
csqrt, ctan and ctanh.
* math/auto-libm-test-out: Regenerated.
* math/libm-test.inc (TEST_COND_x86_64): New macro.
(TEST_COND_x86): Likewise.
(ccos_test_data): Use AUTO_TESTS_c_c.
(ccosh_test_data): Likewise.
(cexp_test_data): Likewise.
(clog_test_data): Likewise.
(csqrt_test_data): Likewise.
(ctan_test_data): Likewise.
(ctan_tonearest_test_data): Likewise.
(ctan_towardzero_test_data): Likewise.
(ctan_downward_test_data): Likewise.
(ctan_upward_test_data): Likewise.
(ctanh_test_data): Likewise.
(ctanh_tonearest_test_data): Likewise.
(ctanh_towardzero_test_data): Likewise.
(ctanh_downward_test_data): Likewise.
(ctanh_upward_test_data): Likewise.
* math/gen-auto-libm-tests.c (func_calc_method): Add value
mpc_c_c.
(func_calc_desc): Add mpc_c_c union field.
(FUNC_mpc_c_c): New macro.
(test_functions): Add cacos, cacosh, casin, casinh, catan, catanh,
ccos, ccosh, cexp, clog, clog10, csin, csinh, csqrt, ctan and
ctanh.
(special_fill_min_subnorm_p120): New function.
(special_real_inputs): Add min_subnorm_p120.
(calc_generic_results): Handle mpc_c_c.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2013-12-20 12:32:44 +00:00
|
|
|
float: 3
|
2012-07-04 09:55:26 +00:00
|
|
|
idouble: 2
|
Move various TEST_c_c tests from libm-test.inc to auto-libm-test-inc.
This patch moves tests of ccos, ccosh, cexp, clog, csqrt, ctan and
ctanh to auto-libm-test-in, adding the required support to
gen-auto-libm-tests. Other TEST_c_c functions aren't moved for now
(although the relevant table entries are put in gen-auto-libm-tests
for it to know how to handle them): clog10 because of a known MPC bug
causing it to hang for at least some pure imaginary inputs (fixed in
SVN, but I'd rather not rely on unreleased versions of MPFR or MPC
even if relying on very recent releases); the inverse trig and
hyperbolic functions because of known slowness in special cases; and
csin / csinh because of observed slowness that I need to investigate
and report to the MPC maintainers. Slowness can be bypassed by moving
to incremental generation (only for new / changed tests) rather than
regenerating the whole of auto-libm-test-out every time, but that
needs implementing. (This patch takes the time for running
gen-auto-libm-tests from about one second to seven, on my system,
which I think is reasonable. The slow functions would make it take
several minutes at least, which seems unreasonable.)
Tested x86_64 and x86 and ulps updated accordingly.
* math/auto-libm-test-in: Add tests of ccos, ccosh, cexp, clog,
csqrt, ctan and ctanh.
* math/auto-libm-test-out: Regenerated.
* math/libm-test.inc (TEST_COND_x86_64): New macro.
(TEST_COND_x86): Likewise.
(ccos_test_data): Use AUTO_TESTS_c_c.
(ccosh_test_data): Likewise.
(cexp_test_data): Likewise.
(clog_test_data): Likewise.
(csqrt_test_data): Likewise.
(ctan_test_data): Likewise.
(ctan_tonearest_test_data): Likewise.
(ctan_towardzero_test_data): Likewise.
(ctan_downward_test_data): Likewise.
(ctan_upward_test_data): Likewise.
(ctanh_test_data): Likewise.
(ctanh_tonearest_test_data): Likewise.
(ctanh_towardzero_test_data): Likewise.
(ctanh_downward_test_data): Likewise.
(ctanh_upward_test_data): Likewise.
* math/gen-auto-libm-tests.c (func_calc_method): Add value
mpc_c_c.
(func_calc_desc): Add mpc_c_c union field.
(FUNC_mpc_c_c): New macro.
(test_functions): Add cacos, cacosh, casin, casinh, catan, catanh,
ccos, ccosh, cexp, clog, clog10, csin, csinh, csqrt, ctan and
ctanh.
(special_fill_min_subnorm_p120): New function.
(special_real_inputs): Add min_subnorm_p120.
(calc_generic_results): Handle mpc_c_c.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2013-12-20 12:32:44 +00:00
|
|
|
ifloat: 3
|
|
|
|
|
ildouble: 3
|
|
|
|
|
ldouble: 3
|
2012-07-04 09:55:26 +00:00
|
|
|
|
2002-08-31 08:22:35 +00:00
|
|
|
Function: "erf":
|
2002-12-03 15:39:37 +00:00
|
|
|
double: 1
|
|
|
|
|
idouble: 1
|
2013-12-03 16:22:49 +00:00
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
2002-08-31 08:22:35 +00:00
|
|
|
|
Test most libm functions in all rounding modes.
This patch makes libm-test.inc tests of most functions use ALL_RM_TEST
unless there was some reason to defer that change for a particular
function.
I started out planning to defer the change for pow (bug 16315), cexp /
ccos / ccosh / csin / csinh (likely fallout from exp, bug 16284) and
cpow (exact expectations for signs of exact zero results not wanted).
Testing on x86_64 and x86 showed additional failures for acosh, cacos,
catan, catanh, clog, clog10, jn, log, log10, log1p, log2, tgamma, yn,
so making the change for those functions was deferred as well, pending
investigation to show which of these represent distinct bugs (some
such bugs may already be filed) and appropriate fixing / XFAILing.
Failures include wrong signs of zero results, errors slightly above
the 9ulp bound (in such cases it may make sense for functions to set
round-to-nearest internally to reduce error accumulation), large
errors and incorrect overflow/underflow for the rounding mode (with
consequent missing errno settings in some cases). It's possible some
could be issues with test expectations, though I didn't notice any
that were obviously like that (I added NO_TEST_INLINE for cases that
were failing for ildoubl on x86 and where it seemed reasonable for
them to fail for the fast-math inlines).
There may of course be failures on other architectures for functions
that didn't fail on x86_64 or x86, in which case the usual rule
applies: file a bug (preferably identifying the underlying problem
function, in cases where function A calls function B and a problem
with function B may present in the test results for function A) if not
already in Bugzilla then fix or XFAIL.
Tested x86_64 and x86 and ulps updated accordingly.
* math/libm-test.inc (asinh_test): Use ALL_RM_TEST.
(atan_test): Likewise.
(atanh_test_data): Use NO_TEST_INLINE for two tests.
(atanh_test): Use ALL_RM_TEST.
(atan2_test_data): Likewise.
(cabs_test): Likewise.
(cacosh_test): Likewise.
(carg_test): Likewise.
(casin_test): Likewise.
(casinh_test): Likewise.
(cbrt_test): Likewise.
(csqrt_test): Likewise.
(erf_test): Likewise.
(erfc_test): Likewise.
(pow10_test): Likewise.
(exp2_test): Likewise.
(hypot_test): Likewise.
(j0_test): Likewise.
(j1_test): Likewise.
(lgamma_test): Likewise.
(gamma_test): Likewise.
(sincos_test): Likewise.
(tanh_test): Likewise.
(y0_test): Likewise.
(y1_test): Likewise.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2014-03-21 00:03:38 +00:00
|
|
|
Function: "erf_downward":
|
|
|
|
|
float: 1
|
|
|
|
|
ifloat: 1
|
|
|
|
|
|
|
|
|
|
Function: "erf_towardzero":
|
|
|
|
|
float: 1
|
|
|
|
|
ifloat: 1
|
|
|
|
|
|
|
|
|
|
Function: "erf_upward":
|
|
|
|
|
float: 1
|
|
|
|
|
ifloat: 1
|
|
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
|
|
|
|
|
2001-09-19 10:37:31 +00:00
|
|
|
Function: "erfc":
|
2002-12-03 15:39:37 +00:00
|
|
|
double: 1
|
2012-03-01 21:15:38 +00:00
|
|
|
float: 1
|
2002-12-03 15:39:37 +00:00
|
|
|
idouble: 1
|
2012-03-01 21:15:38 +00:00
|
|
|
ifloat: 1
|
2002-08-31 08:22:35 +00:00
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
|
|
|
|
|
Test most libm functions in all rounding modes.
This patch makes libm-test.inc tests of most functions use ALL_RM_TEST
unless there was some reason to defer that change for a particular
function.
I started out planning to defer the change for pow (bug 16315), cexp /
ccos / ccosh / csin / csinh (likely fallout from exp, bug 16284) and
cpow (exact expectations for signs of exact zero results not wanted).
Testing on x86_64 and x86 showed additional failures for acosh, cacos,
catan, catanh, clog, clog10, jn, log, log10, log1p, log2, tgamma, yn,
so making the change for those functions was deferred as well, pending
investigation to show which of these represent distinct bugs (some
such bugs may already be filed) and appropriate fixing / XFAILing.
Failures include wrong signs of zero results, errors slightly above
the 9ulp bound (in such cases it may make sense for functions to set
round-to-nearest internally to reduce error accumulation), large
errors and incorrect overflow/underflow for the rounding mode (with
consequent missing errno settings in some cases). It's possible some
could be issues with test expectations, though I didn't notice any
that were obviously like that (I added NO_TEST_INLINE for cases that
were failing for ildoubl on x86 and where it seemed reasonable for
them to fail for the fast-math inlines).
There may of course be failures on other architectures for functions
that didn't fail on x86_64 or x86, in which case the usual rule
applies: file a bug (preferably identifying the underlying problem
function, in cases where function A calls function B and a problem
with function B may present in the test results for function A) if not
already in Bugzilla then fix or XFAIL.
Tested x86_64 and x86 and ulps updated accordingly.
* math/libm-test.inc (asinh_test): Use ALL_RM_TEST.
(atan_test): Likewise.
(atanh_test_data): Use NO_TEST_INLINE for two tests.
(atanh_test): Use ALL_RM_TEST.
(atan2_test_data): Likewise.
(cabs_test): Likewise.
(cacosh_test): Likewise.
(carg_test): Likewise.
(casin_test): Likewise.
(casinh_test): Likewise.
(cbrt_test): Likewise.
(csqrt_test): Likewise.
(erf_test): Likewise.
(erfc_test): Likewise.
(pow10_test): Likewise.
(exp2_test): Likewise.
(hypot_test): Likewise.
(j0_test): Likewise.
(j1_test): Likewise.
(lgamma_test): Likewise.
(gamma_test): Likewise.
(sincos_test): Likewise.
(tanh_test): Likewise.
(y0_test): Likewise.
(y1_test): Likewise.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2014-03-21 00:03:38 +00:00
|
|
|
Function: "erfc_downward":
|
|
|
|
|
double: 1
|
|
|
|
|
float: 3
|
|
|
|
|
idouble: 1
|
|
|
|
|
ifloat: 3
|
|
|
|
|
ildouble: 3
|
|
|
|
|
ldouble: 3
|
|
|
|
|
|
|
|
|
|
Function: "erfc_towardzero":
|
|
|
|
|
double: 1
|
|
|
|
|
float: 3
|
|
|
|
|
idouble: 1
|
|
|
|
|
ifloat: 3
|
|
|
|
|
ildouble: 2
|
|
|
|
|
ldouble: 2
|
|
|
|
|
|
|
|
|
|
Function: "erfc_upward":
|
|
|
|
|
double: 2
|
|
|
|
|
float: 2
|
|
|
|
|
idouble: 2
|
|
|
|
|
ifloat: 2
|
|
|
|
|
ildouble: 3
|
|
|
|
|
ldouble: 3
|
|
|
|
|
|
2001-09-19 10:37:31 +00:00
|
|
|
Function: "exp10":
|
2013-07-02 20:00:48 +00:00
|
|
|
double: 1
|
|
|
|
|
idouble: 1
|
|
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
2001-09-19 10:37:31 +00:00
|
|
|
|
Fix x86 / x86_64 expl / expl10l wild results in directed rounding modes (bug 16356).
This patch fixes bug 16356, bad results from x86 / x86_64 expl /
exp10l in directed rounding modes, the most serious of the bugs shown
up by my patch expanding libm test coverage. When I fixed bug 16293,
I thought it was only necessary to set round-to-nearest when using
frndint in expm1 functions, because in other cases the cancellation
error from having the resulting fractional part close to 1 or -1 would
not be significant. However, in expl and exp10l, the way the final
fractional part gets computed (something more complicated than a
simple subtraction, because more precision is needed than you'd get
that way) can result in a value outside the range [-1, 1] when the
argument to frndint was very close to an integer and was rounded the
"wrong" way because of the rounding mode - and the f2xm1 instruction
has undefined results if its argument is outside [-1, 1], so resulting
in the large errors seen. So this patch removes the USE_AS_EXPM1L
conditionals on the round-to-nearest settings, so all of expl, expm1l
and exp10l now get round-to-nearest used for frndint (meaning the
final fractional part can at most be slightly above 0.5 in
magnitude). Associated tests of exp and exp10 are added and testing
of exp10 in directed rounding modes enabled.
Tested x86_64 and x86 and ulps updated accordingly.
* sysdeps/i386/fpu/e_expl.S (IEEE754_EXPL): Also set
round-to-nearest for [!USE_AS_EXPM1L].
* sysdeps/x86_64/fpu/e_expl.S (IEEE754_EXPL): Likewise.
* math/auto-libm-test-in: Do not expect cosh tests to fail. Add
more tests of exp and exp10. Expect some exp10 tests to miss
exceptions or fail in directed rounding modes.
* math/auto-libm-test-out: Regenerated.
* math/libm-test.inc (exp10_tonearest_test_data): New array.
(exp10_test_tonearest): New function.
(exp10_towardzero_test_data): New array.
(exp10_test_towardzero): New function.
(exp10_downward_test_data): New array.
(exp10_test_downward): New function.
(exp10_upward_test_data): New array.
(exp10_test_upward): New function.
(main): Call the new functions.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2013-12-21 13:07:16 +00:00
|
|
|
Function: "exp10_downward":
|
|
|
|
|
double: 1
|
2014-06-25 11:33:22 +00:00
|
|
|
float: 1
|
Fix x86 / x86_64 expl / expl10l wild results in directed rounding modes (bug 16356).
This patch fixes bug 16356, bad results from x86 / x86_64 expl /
exp10l in directed rounding modes, the most serious of the bugs shown
up by my patch expanding libm test coverage. When I fixed bug 16293,
I thought it was only necessary to set round-to-nearest when using
frndint in expm1 functions, because in other cases the cancellation
error from having the resulting fractional part close to 1 or -1 would
not be significant. However, in expl and exp10l, the way the final
fractional part gets computed (something more complicated than a
simple subtraction, because more precision is needed than you'd get
that way) can result in a value outside the range [-1, 1] when the
argument to frndint was very close to an integer and was rounded the
"wrong" way because of the rounding mode - and the f2xm1 instruction
has undefined results if its argument is outside [-1, 1], so resulting
in the large errors seen. So this patch removes the USE_AS_EXPM1L
conditionals on the round-to-nearest settings, so all of expl, expm1l
and exp10l now get round-to-nearest used for frndint (meaning the
final fractional part can at most be slightly above 0.5 in
magnitude). Associated tests of exp and exp10 are added and testing
of exp10 in directed rounding modes enabled.
Tested x86_64 and x86 and ulps updated accordingly.
* sysdeps/i386/fpu/e_expl.S (IEEE754_EXPL): Also set
round-to-nearest for [!USE_AS_EXPM1L].
* sysdeps/x86_64/fpu/e_expl.S (IEEE754_EXPL): Likewise.
* math/auto-libm-test-in: Do not expect cosh tests to fail. Add
more tests of exp and exp10. Expect some exp10 tests to miss
exceptions or fail in directed rounding modes.
* math/auto-libm-test-out: Regenerated.
* math/libm-test.inc (exp10_tonearest_test_data): New array.
(exp10_test_tonearest): New function.
(exp10_towardzero_test_data): New array.
(exp10_test_towardzero): New function.
(exp10_downward_test_data): New array.
(exp10_test_downward): New function.
(exp10_upward_test_data): New array.
(exp10_test_upward): New function.
(main): Call the new functions.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2013-12-21 13:07:16 +00:00
|
|
|
idouble: 1
|
2014-06-25 11:33:22 +00:00
|
|
|
ifloat: 1
|
Fix x86 / x86_64 expl / expl10l wild results in directed rounding modes (bug 16356).
This patch fixes bug 16356, bad results from x86 / x86_64 expl /
exp10l in directed rounding modes, the most serious of the bugs shown
up by my patch expanding libm test coverage. When I fixed bug 16293,
I thought it was only necessary to set round-to-nearest when using
frndint in expm1 functions, because in other cases the cancellation
error from having the resulting fractional part close to 1 or -1 would
not be significant. However, in expl and exp10l, the way the final
fractional part gets computed (something more complicated than a
simple subtraction, because more precision is needed than you'd get
that way) can result in a value outside the range [-1, 1] when the
argument to frndint was very close to an integer and was rounded the
"wrong" way because of the rounding mode - and the f2xm1 instruction
has undefined results if its argument is outside [-1, 1], so resulting
in the large errors seen. So this patch removes the USE_AS_EXPM1L
conditionals on the round-to-nearest settings, so all of expl, expm1l
and exp10l now get round-to-nearest used for frndint (meaning the
final fractional part can at most be slightly above 0.5 in
magnitude). Associated tests of exp and exp10 are added and testing
of exp10 in directed rounding modes enabled.
Tested x86_64 and x86 and ulps updated accordingly.
* sysdeps/i386/fpu/e_expl.S (IEEE754_EXPL): Also set
round-to-nearest for [!USE_AS_EXPM1L].
* sysdeps/x86_64/fpu/e_expl.S (IEEE754_EXPL): Likewise.
* math/auto-libm-test-in: Do not expect cosh tests to fail. Add
more tests of exp and exp10. Expect some exp10 tests to miss
exceptions or fail in directed rounding modes.
* math/auto-libm-test-out: Regenerated.
* math/libm-test.inc (exp10_tonearest_test_data): New array.
(exp10_test_tonearest): New function.
(exp10_towardzero_test_data): New array.
(exp10_test_towardzero): New function.
(exp10_downward_test_data): New array.
(exp10_test_downward): New function.
(exp10_upward_test_data): New array.
(exp10_test_upward): New function.
(main): Call the new functions.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2013-12-21 13:07:16 +00:00
|
|
|
ildouble: 2
|
|
|
|
|
ldouble: 2
|
|
|
|
|
|
|
|
|
|
Function: "exp10_towardzero":
|
|
|
|
|
double: 1
|
2014-06-25 11:33:22 +00:00
|
|
|
float: 1
|
Fix x86 / x86_64 expl / expl10l wild results in directed rounding modes (bug 16356).
This patch fixes bug 16356, bad results from x86 / x86_64 expl /
exp10l in directed rounding modes, the most serious of the bugs shown
up by my patch expanding libm test coverage. When I fixed bug 16293,
I thought it was only necessary to set round-to-nearest when using
frndint in expm1 functions, because in other cases the cancellation
error from having the resulting fractional part close to 1 or -1 would
not be significant. However, in expl and exp10l, the way the final
fractional part gets computed (something more complicated than a
simple subtraction, because more precision is needed than you'd get
that way) can result in a value outside the range [-1, 1] when the
argument to frndint was very close to an integer and was rounded the
"wrong" way because of the rounding mode - and the f2xm1 instruction
has undefined results if its argument is outside [-1, 1], so resulting
in the large errors seen. So this patch removes the USE_AS_EXPM1L
conditionals on the round-to-nearest settings, so all of expl, expm1l
and exp10l now get round-to-nearest used for frndint (meaning the
final fractional part can at most be slightly above 0.5 in
magnitude). Associated tests of exp and exp10 are added and testing
of exp10 in directed rounding modes enabled.
Tested x86_64 and x86 and ulps updated accordingly.
* sysdeps/i386/fpu/e_expl.S (IEEE754_EXPL): Also set
round-to-nearest for [!USE_AS_EXPM1L].
* sysdeps/x86_64/fpu/e_expl.S (IEEE754_EXPL): Likewise.
* math/auto-libm-test-in: Do not expect cosh tests to fail. Add
more tests of exp and exp10. Expect some exp10 tests to miss
exceptions or fail in directed rounding modes.
* math/auto-libm-test-out: Regenerated.
* math/libm-test.inc (exp10_tonearest_test_data): New array.
(exp10_test_tonearest): New function.
(exp10_towardzero_test_data): New array.
(exp10_test_towardzero): New function.
(exp10_downward_test_data): New array.
(exp10_test_downward): New function.
(exp10_upward_test_data): New array.
(exp10_test_upward): New function.
(main): Call the new functions.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2013-12-21 13:07:16 +00:00
|
|
|
idouble: 1
|
2014-06-25 11:33:22 +00:00
|
|
|
ifloat: 1
|
Fix x86 / x86_64 expl / expl10l wild results in directed rounding modes (bug 16356).
This patch fixes bug 16356, bad results from x86 / x86_64 expl /
exp10l in directed rounding modes, the most serious of the bugs shown
up by my patch expanding libm test coverage. When I fixed bug 16293,
I thought it was only necessary to set round-to-nearest when using
frndint in expm1 functions, because in other cases the cancellation
error from having the resulting fractional part close to 1 or -1 would
not be significant. However, in expl and exp10l, the way the final
fractional part gets computed (something more complicated than a
simple subtraction, because more precision is needed than you'd get
that way) can result in a value outside the range [-1, 1] when the
argument to frndint was very close to an integer and was rounded the
"wrong" way because of the rounding mode - and the f2xm1 instruction
has undefined results if its argument is outside [-1, 1], so resulting
in the large errors seen. So this patch removes the USE_AS_EXPM1L
conditionals on the round-to-nearest settings, so all of expl, expm1l
and exp10l now get round-to-nearest used for frndint (meaning the
final fractional part can at most be slightly above 0.5 in
magnitude). Associated tests of exp and exp10 are added and testing
of exp10 in directed rounding modes enabled.
Tested x86_64 and x86 and ulps updated accordingly.
* sysdeps/i386/fpu/e_expl.S (IEEE754_EXPL): Also set
round-to-nearest for [!USE_AS_EXPM1L].
* sysdeps/x86_64/fpu/e_expl.S (IEEE754_EXPL): Likewise.
* math/auto-libm-test-in: Do not expect cosh tests to fail. Add
more tests of exp and exp10. Expect some exp10 tests to miss
exceptions or fail in directed rounding modes.
* math/auto-libm-test-out: Regenerated.
* math/libm-test.inc (exp10_tonearest_test_data): New array.
(exp10_test_tonearest): New function.
(exp10_towardzero_test_data): New array.
(exp10_test_towardzero): New function.
(exp10_downward_test_data): New array.
(exp10_test_downward): New function.
(exp10_upward_test_data): New array.
(exp10_test_upward): New function.
(main): Call the new functions.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2013-12-21 13:07:16 +00:00
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
|
|
|
|
|
|
|
|
|
Function: "exp10_upward":
|
|
|
|
|
double: 1
|
|
|
|
|
float: 1
|
|
|
|
|
idouble: 1
|
|
|
|
|
ifloat: 1
|
|
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
|
|
|
|
|
2012-03-02 15:12:53 +00:00
|
|
|
Function: "exp_downward":
|
2013-12-03 16:22:49 +00:00
|
|
|
double: 1
|
|
|
|
|
idouble: 1
|
2012-03-02 15:12:53 +00:00
|
|
|
|
|
|
|
|
Function: "exp_towardzero":
|
2013-12-03 16:22:49 +00:00
|
|
|
double: 1
|
|
|
|
|
idouble: 1
|
2012-03-02 15:12:53 +00:00
|
|
|
|
|
|
|
|
Function: "exp_upward":
|
2013-12-03 16:22:49 +00:00
|
|
|
double: 1
|
|
|
|
|
idouble: 1
|
2013-12-03 21:49:56 +00:00
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
2012-03-02 15:12:53 +00:00
|
|
|
|
2001-09-19 10:37:31 +00:00
|
|
|
Function: "expm1":
|
|
|
|
|
double: 1
|
|
|
|
|
float: 1
|
|
|
|
|
idouble: 1
|
|
|
|
|
ifloat: 1
|
2015-03-25 00:05:13 +00:00
|
|
|
ildouble: 2
|
|
|
|
|
ldouble: 2
|
2002-08-31 08:22:35 +00:00
|
|
|
|
2013-12-19 13:36:10 +00:00
|
|
|
Function: "expm1_downward":
|
|
|
|
|
double: 1
|
|
|
|
|
float: 1
|
|
|
|
|
idouble: 1
|
|
|
|
|
ifloat: 1
|
2015-03-25 00:05:13 +00:00
|
|
|
ildouble: 4
|
|
|
|
|
ldouble: 4
|
2013-12-19 13:36:10 +00:00
|
|
|
|
|
|
|
|
Function: "expm1_towardzero":
|
|
|
|
|
double: 1
|
|
|
|
|
float: 1
|
|
|
|
|
idouble: 1
|
|
|
|
|
ifloat: 1
|
2015-03-25 00:05:13 +00:00
|
|
|
ildouble: 3
|
|
|
|
|
ldouble: 3
|
2013-12-19 13:36:10 +00:00
|
|
|
|
|
|
|
|
Function: "expm1_upward":
|
|
|
|
|
double: 1
|
|
|
|
|
float: 1
|
|
|
|
|
idouble: 1
|
|
|
|
|
ifloat: 1
|
2015-03-25 00:05:13 +00:00
|
|
|
ildouble: 3
|
|
|
|
|
ldouble: 3
|
2013-12-19 13:36:10 +00:00
|
|
|
|
2001-09-19 10:37:31 +00:00
|
|
|
Function: "gamma":
|
2013-05-24 19:21:22 +00:00
|
|
|
double: 1
|
2014-01-01 14:34:38 +00:00
|
|
|
float: 1
|
2013-05-24 19:21:22 +00:00
|
|
|
idouble: 1
|
2014-01-01 14:34:38 +00:00
|
|
|
ifloat: 1
|
2001-09-19 10:37:31 +00:00
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
|
|
|
|
|
Test most libm functions in all rounding modes.
This patch makes libm-test.inc tests of most functions use ALL_RM_TEST
unless there was some reason to defer that change for a particular
function.
I started out planning to defer the change for pow (bug 16315), cexp /
ccos / ccosh / csin / csinh (likely fallout from exp, bug 16284) and
cpow (exact expectations for signs of exact zero results not wanted).
Testing on x86_64 and x86 showed additional failures for acosh, cacos,
catan, catanh, clog, clog10, jn, log, log10, log1p, log2, tgamma, yn,
so making the change for those functions was deferred as well, pending
investigation to show which of these represent distinct bugs (some
such bugs may already be filed) and appropriate fixing / XFAILing.
Failures include wrong signs of zero results, errors slightly above
the 9ulp bound (in such cases it may make sense for functions to set
round-to-nearest internally to reduce error accumulation), large
errors and incorrect overflow/underflow for the rounding mode (with
consequent missing errno settings in some cases). It's possible some
could be issues with test expectations, though I didn't notice any
that were obviously like that (I added NO_TEST_INLINE for cases that
were failing for ildoubl on x86 and where it seemed reasonable for
them to fail for the fast-math inlines).
There may of course be failures on other architectures for functions
that didn't fail on x86_64 or x86, in which case the usual rule
applies: file a bug (preferably identifying the underlying problem
function, in cases where function A calls function B and a problem
with function B may present in the test results for function A) if not
already in Bugzilla then fix or XFAIL.
Tested x86_64 and x86 and ulps updated accordingly.
* math/libm-test.inc (asinh_test): Use ALL_RM_TEST.
(atan_test): Likewise.
(atanh_test_data): Use NO_TEST_INLINE for two tests.
(atanh_test): Use ALL_RM_TEST.
(atan2_test_data): Likewise.
(cabs_test): Likewise.
(cacosh_test): Likewise.
(carg_test): Likewise.
(casin_test): Likewise.
(casinh_test): Likewise.
(cbrt_test): Likewise.
(csqrt_test): Likewise.
(erf_test): Likewise.
(erfc_test): Likewise.
(pow10_test): Likewise.
(exp2_test): Likewise.
(hypot_test): Likewise.
(j0_test): Likewise.
(j1_test): Likewise.
(lgamma_test): Likewise.
(gamma_test): Likewise.
(sincos_test): Likewise.
(tanh_test): Likewise.
(y0_test): Likewise.
(y1_test): Likewise.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2014-03-21 00:03:38 +00:00
|
|
|
Function: "gamma_downward":
|
|
|
|
|
double: 2
|
|
|
|
|
float: 1
|
|
|
|
|
idouble: 2
|
|
|
|
|
ifloat: 1
|
|
|
|
|
ildouble: 3
|
|
|
|
|
ldouble: 3
|
|
|
|
|
|
|
|
|
|
Function: "gamma_towardzero":
|
|
|
|
|
double: 1
|
|
|
|
|
float: 1
|
|
|
|
|
idouble: 1
|
|
|
|
|
ifloat: 1
|
|
|
|
|
ildouble: 2
|
|
|
|
|
ldouble: 2
|
|
|
|
|
|
|
|
|
|
Function: "gamma_upward":
|
|
|
|
|
double: 1
|
|
|
|
|
float: 3
|
|
|
|
|
idouble: 1
|
|
|
|
|
ifloat: 3
|
|
|
|
|
ildouble: 2
|
|
|
|
|
ldouble: 2
|
|
|
|
|
|
2001-09-19 10:37:31 +00:00
|
|
|
Function: "hypot":
|
2013-12-16 21:18:07 +00:00
|
|
|
double: 1
|
|
|
|
|
idouble: 1
|
|
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
2001-09-19 10:37:31 +00:00
|
|
|
|
Test most libm functions in all rounding modes.
This patch makes libm-test.inc tests of most functions use ALL_RM_TEST
unless there was some reason to defer that change for a particular
function.
I started out planning to defer the change for pow (bug 16315), cexp /
ccos / ccosh / csin / csinh (likely fallout from exp, bug 16284) and
cpow (exact expectations for signs of exact zero results not wanted).
Testing on x86_64 and x86 showed additional failures for acosh, cacos,
catan, catanh, clog, clog10, jn, log, log10, log1p, log2, tgamma, yn,
so making the change for those functions was deferred as well, pending
investigation to show which of these represent distinct bugs (some
such bugs may already be filed) and appropriate fixing / XFAILing.
Failures include wrong signs of zero results, errors slightly above
the 9ulp bound (in such cases it may make sense for functions to set
round-to-nearest internally to reduce error accumulation), large
errors and incorrect overflow/underflow for the rounding mode (with
consequent missing errno settings in some cases). It's possible some
could be issues with test expectations, though I didn't notice any
that were obviously like that (I added NO_TEST_INLINE for cases that
were failing for ildoubl on x86 and where it seemed reasonable for
them to fail for the fast-math inlines).
There may of course be failures on other architectures for functions
that didn't fail on x86_64 or x86, in which case the usual rule
applies: file a bug (preferably identifying the underlying problem
function, in cases where function A calls function B and a problem
with function B may present in the test results for function A) if not
already in Bugzilla then fix or XFAIL.
Tested x86_64 and x86 and ulps updated accordingly.
* math/libm-test.inc (asinh_test): Use ALL_RM_TEST.
(atan_test): Likewise.
(atanh_test_data): Use NO_TEST_INLINE for two tests.
(atanh_test): Use ALL_RM_TEST.
(atan2_test_data): Likewise.
(cabs_test): Likewise.
(cacosh_test): Likewise.
(carg_test): Likewise.
(casin_test): Likewise.
(casinh_test): Likewise.
(cbrt_test): Likewise.
(csqrt_test): Likewise.
(erf_test): Likewise.
(erfc_test): Likewise.
(pow10_test): Likewise.
(exp2_test): Likewise.
(hypot_test): Likewise.
(j0_test): Likewise.
(j1_test): Likewise.
(lgamma_test): Likewise.
(gamma_test): Likewise.
(sincos_test): Likewise.
(tanh_test): Likewise.
(y0_test): Likewise.
(y1_test): Likewise.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2014-03-21 00:03:38 +00:00
|
|
|
Function: "hypot_downward":
|
|
|
|
|
double: 1
|
|
|
|
|
idouble: 1
|
|
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
|
|
|
|
|
|
|
|
|
Function: "hypot_towardzero":
|
|
|
|
|
double: 1
|
|
|
|
|
idouble: 1
|
|
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
|
|
|
|
|
|
|
|
|
Function: "hypot_upward":
|
|
|
|
|
double: 1
|
|
|
|
|
idouble: 1
|
|
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
|
|
|
|
|
2001-09-19 10:37:31 +00:00
|
|
|
Function: "j0":
|
|
|
|
|
double: 2
|
|
|
|
|
float: 2
|
|
|
|
|
idouble: 2
|
|
|
|
|
ifloat: 2
|
2012-03-16 12:28:25 +00:00
|
|
|
ildouble: 2
|
|
|
|
|
ldouble: 2
|
2001-09-19 10:37:31 +00:00
|
|
|
|
Test most libm functions in all rounding modes.
This patch makes libm-test.inc tests of most functions use ALL_RM_TEST
unless there was some reason to defer that change for a particular
function.
I started out planning to defer the change for pow (bug 16315), cexp /
ccos / ccosh / csin / csinh (likely fallout from exp, bug 16284) and
cpow (exact expectations for signs of exact zero results not wanted).
Testing on x86_64 and x86 showed additional failures for acosh, cacos,
catan, catanh, clog, clog10, jn, log, log10, log1p, log2, tgamma, yn,
so making the change for those functions was deferred as well, pending
investigation to show which of these represent distinct bugs (some
such bugs may already be filed) and appropriate fixing / XFAILing.
Failures include wrong signs of zero results, errors slightly above
the 9ulp bound (in such cases it may make sense for functions to set
round-to-nearest internally to reduce error accumulation), large
errors and incorrect overflow/underflow for the rounding mode (with
consequent missing errno settings in some cases). It's possible some
could be issues with test expectations, though I didn't notice any
that were obviously like that (I added NO_TEST_INLINE for cases that
were failing for ildoubl on x86 and where it seemed reasonable for
them to fail for the fast-math inlines).
There may of course be failures on other architectures for functions
that didn't fail on x86_64 or x86, in which case the usual rule
applies: file a bug (preferably identifying the underlying problem
function, in cases where function A calls function B and a problem
with function B may present in the test results for function A) if not
already in Bugzilla then fix or XFAIL.
Tested x86_64 and x86 and ulps updated accordingly.
* math/libm-test.inc (asinh_test): Use ALL_RM_TEST.
(atan_test): Likewise.
(atanh_test_data): Use NO_TEST_INLINE for two tests.
(atanh_test): Use ALL_RM_TEST.
(atan2_test_data): Likewise.
(cabs_test): Likewise.
(cacosh_test): Likewise.
(carg_test): Likewise.
(casin_test): Likewise.
(casinh_test): Likewise.
(cbrt_test): Likewise.
(csqrt_test): Likewise.
(erf_test): Likewise.
(erfc_test): Likewise.
(pow10_test): Likewise.
(exp2_test): Likewise.
(hypot_test): Likewise.
(j0_test): Likewise.
(j1_test): Likewise.
(lgamma_test): Likewise.
(gamma_test): Likewise.
(sincos_test): Likewise.
(tanh_test): Likewise.
(y0_test): Likewise.
(y1_test): Likewise.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2014-03-21 00:03:38 +00:00
|
|
|
Function: "j0_downward":
|
|
|
|
|
double: 2
|
|
|
|
|
float: 4
|
|
|
|
|
idouble: 2
|
|
|
|
|
ifloat: 4
|
|
|
|
|
ildouble: 4
|
|
|
|
|
ldouble: 4
|
|
|
|
|
|
|
|
|
|
Function: "j0_towardzero":
|
|
|
|
|
double: 3
|
|
|
|
|
float: 2
|
|
|
|
|
idouble: 3
|
|
|
|
|
ifloat: 2
|
|
|
|
|
ildouble: 5
|
|
|
|
|
ldouble: 5
|
|
|
|
|
|
|
|
|
|
Function: "j0_upward":
|
|
|
|
|
double: 3
|
|
|
|
|
float: 2
|
|
|
|
|
idouble: 3
|
|
|
|
|
ifloat: 2
|
|
|
|
|
ildouble: 4
|
|
|
|
|
ldouble: 4
|
|
|
|
|
|
2001-09-19 10:37:31 +00:00
|
|
|
Function: "j1":
|
|
|
|
|
double: 1
|
|
|
|
|
float: 2
|
|
|
|
|
idouble: 1
|
|
|
|
|
ifloat: 2
|
2002-08-31 08:22:35 +00:00
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
2001-09-19 10:37:31 +00:00
|
|
|
|
Test most libm functions in all rounding modes.
This patch makes libm-test.inc tests of most functions use ALL_RM_TEST
unless there was some reason to defer that change for a particular
function.
I started out planning to defer the change for pow (bug 16315), cexp /
ccos / ccosh / csin / csinh (likely fallout from exp, bug 16284) and
cpow (exact expectations for signs of exact zero results not wanted).
Testing on x86_64 and x86 showed additional failures for acosh, cacos,
catan, catanh, clog, clog10, jn, log, log10, log1p, log2, tgamma, yn,
so making the change for those functions was deferred as well, pending
investigation to show which of these represent distinct bugs (some
such bugs may already be filed) and appropriate fixing / XFAILing.
Failures include wrong signs of zero results, errors slightly above
the 9ulp bound (in such cases it may make sense for functions to set
round-to-nearest internally to reduce error accumulation), large
errors and incorrect overflow/underflow for the rounding mode (with
consequent missing errno settings in some cases). It's possible some
could be issues with test expectations, though I didn't notice any
that were obviously like that (I added NO_TEST_INLINE for cases that
were failing for ildoubl on x86 and where it seemed reasonable for
them to fail for the fast-math inlines).
There may of course be failures on other architectures for functions
that didn't fail on x86_64 or x86, in which case the usual rule
applies: file a bug (preferably identifying the underlying problem
function, in cases where function A calls function B and a problem
with function B may present in the test results for function A) if not
already in Bugzilla then fix or XFAIL.
Tested x86_64 and x86 and ulps updated accordingly.
* math/libm-test.inc (asinh_test): Use ALL_RM_TEST.
(atan_test): Likewise.
(atanh_test_data): Use NO_TEST_INLINE for two tests.
(atanh_test): Use ALL_RM_TEST.
(atan2_test_data): Likewise.
(cabs_test): Likewise.
(cacosh_test): Likewise.
(carg_test): Likewise.
(casin_test): Likewise.
(casinh_test): Likewise.
(cbrt_test): Likewise.
(csqrt_test): Likewise.
(erf_test): Likewise.
(erfc_test): Likewise.
(pow10_test): Likewise.
(exp2_test): Likewise.
(hypot_test): Likewise.
(j0_test): Likewise.
(j1_test): Likewise.
(lgamma_test): Likewise.
(gamma_test): Likewise.
(sincos_test): Likewise.
(tanh_test): Likewise.
(y0_test): Likewise.
(y1_test): Likewise.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2014-03-21 00:03:38 +00:00
|
|
|
Function: "j1_downward":
|
|
|
|
|
double: 3
|
|
|
|
|
float: 3
|
|
|
|
|
idouble: 3
|
|
|
|
|
ifloat: 3
|
|
|
|
|
ildouble: 4
|
|
|
|
|
ldouble: 4
|
|
|
|
|
|
|
|
|
|
Function: "j1_towardzero":
|
|
|
|
|
double: 3
|
|
|
|
|
float: 2
|
|
|
|
|
idouble: 3
|
|
|
|
|
ifloat: 2
|
|
|
|
|
ildouble: 4
|
|
|
|
|
ldouble: 4
|
|
|
|
|
|
|
|
|
|
Function: "j1_upward":
|
|
|
|
|
double: 3
|
|
|
|
|
float: 5
|
|
|
|
|
idouble: 3
|
|
|
|
|
ifloat: 5
|
|
|
|
|
ildouble: 3
|
|
|
|
|
ldouble: 3
|
|
|
|
|
|
2001-09-19 10:37:31 +00:00
|
|
|
Function: "jn":
|
2002-12-03 15:39:37 +00:00
|
|
|
double: 4
|
2014-01-01 14:34:38 +00:00
|
|
|
float: 4
|
2002-12-03 15:39:37 +00:00
|
|
|
idouble: 4
|
2014-01-01 14:34:38 +00:00
|
|
|
ifloat: 4
|
2013-12-18 17:59:29 +00:00
|
|
|
ildouble: 4
|
|
|
|
|
ldouble: 4
|
2001-09-19 10:37:31 +00:00
|
|
|
|
|
|
|
|
Function: "lgamma":
|
|
|
|
|
double: 1
|
2014-01-01 14:34:38 +00:00
|
|
|
float: 1
|
2001-09-19 10:37:31 +00:00
|
|
|
idouble: 1
|
2014-01-01 14:34:38 +00:00
|
|
|
ifloat: 1
|
2001-09-19 10:37:31 +00:00
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
|
|
|
|
|
Test most libm functions in all rounding modes.
This patch makes libm-test.inc tests of most functions use ALL_RM_TEST
unless there was some reason to defer that change for a particular
function.
I started out planning to defer the change for pow (bug 16315), cexp /
ccos / ccosh / csin / csinh (likely fallout from exp, bug 16284) and
cpow (exact expectations for signs of exact zero results not wanted).
Testing on x86_64 and x86 showed additional failures for acosh, cacos,
catan, catanh, clog, clog10, jn, log, log10, log1p, log2, tgamma, yn,
so making the change for those functions was deferred as well, pending
investigation to show which of these represent distinct bugs (some
such bugs may already be filed) and appropriate fixing / XFAILing.
Failures include wrong signs of zero results, errors slightly above
the 9ulp bound (in such cases it may make sense for functions to set
round-to-nearest internally to reduce error accumulation), large
errors and incorrect overflow/underflow for the rounding mode (with
consequent missing errno settings in some cases). It's possible some
could be issues with test expectations, though I didn't notice any
that were obviously like that (I added NO_TEST_INLINE for cases that
were failing for ildoubl on x86 and where it seemed reasonable for
them to fail for the fast-math inlines).
There may of course be failures on other architectures for functions
that didn't fail on x86_64 or x86, in which case the usual rule
applies: file a bug (preferably identifying the underlying problem
function, in cases where function A calls function B and a problem
with function B may present in the test results for function A) if not
already in Bugzilla then fix or XFAIL.
Tested x86_64 and x86 and ulps updated accordingly.
* math/libm-test.inc (asinh_test): Use ALL_RM_TEST.
(atan_test): Likewise.
(atanh_test_data): Use NO_TEST_INLINE for two tests.
(atanh_test): Use ALL_RM_TEST.
(atan2_test_data): Likewise.
(cabs_test): Likewise.
(cacosh_test): Likewise.
(carg_test): Likewise.
(casin_test): Likewise.
(casinh_test): Likewise.
(cbrt_test): Likewise.
(csqrt_test): Likewise.
(erf_test): Likewise.
(erfc_test): Likewise.
(pow10_test): Likewise.
(exp2_test): Likewise.
(hypot_test): Likewise.
(j0_test): Likewise.
(j1_test): Likewise.
(lgamma_test): Likewise.
(gamma_test): Likewise.
(sincos_test): Likewise.
(tanh_test): Likewise.
(y0_test): Likewise.
(y1_test): Likewise.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2014-03-21 00:03:38 +00:00
|
|
|
Function: "lgamma_downward":
|
|
|
|
|
double: 2
|
|
|
|
|
float: 1
|
|
|
|
|
idouble: 2
|
|
|
|
|
ifloat: 1
|
|
|
|
|
ildouble: 3
|
|
|
|
|
ldouble: 3
|
|
|
|
|
|
|
|
|
|
Function: "lgamma_towardzero":
|
|
|
|
|
double: 1
|
|
|
|
|
float: 1
|
|
|
|
|
idouble: 1
|
|
|
|
|
ifloat: 1
|
|
|
|
|
ildouble: 2
|
|
|
|
|
ldouble: 2
|
|
|
|
|
|
|
|
|
|
Function: "lgamma_upward":
|
|
|
|
|
double: 1
|
|
|
|
|
float: 3
|
|
|
|
|
idouble: 1
|
|
|
|
|
ifloat: 3
|
|
|
|
|
ildouble: 2
|
|
|
|
|
ldouble: 2
|
|
|
|
|
|
2013-12-05 13:54:50 +00:00
|
|
|
Function: "log":
|
|
|
|
|
float: 1
|
|
|
|
|
ifloat: 1
|
|
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
|
|
|
|
|
2002-12-03 15:39:37 +00:00
|
|
|
Function: "log10":
|
2011-09-10 02:16:10 +00:00
|
|
|
double: 1
|
2002-12-06 22:26:57 +00:00
|
|
|
float: 2
|
2011-09-10 02:16:10 +00:00
|
|
|
idouble: 1
|
2002-12-06 22:26:57 +00:00
|
|
|
ifloat: 2
|
2001-09-19 10:37:31 +00:00
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
|
|
|
|
|
2014-05-23 12:07:50 +00:00
|
|
|
Function: "log10_downward":
|
|
|
|
|
double: 1
|
|
|
|
|
float: 1
|
|
|
|
|
idouble: 1
|
|
|
|
|
ifloat: 1
|
|
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
|
|
|
|
|
|
|
|
|
Function: "log10_towardzero":
|
|
|
|
|
double: 1
|
|
|
|
|
float: 1
|
|
|
|
|
idouble: 1
|
|
|
|
|
ifloat: 1
|
|
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
|
|
|
|
|
|
|
|
|
Function: "log10_upward":
|
|
|
|
|
double: 1
|
|
|
|
|
float: 1
|
|
|
|
|
idouble: 1
|
|
|
|
|
ifloat: 1
|
|
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
|
|
|
|
|
2001-09-19 10:37:31 +00:00
|
|
|
Function: "log1p":
|
|
|
|
|
float: 1
|
|
|
|
|
ifloat: 1
|
2013-12-05 13:54:50 +00:00
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
2001-09-19 10:37:31 +00:00
|
|
|
|
Relax gen-auto-libm-tests may-underflow rules, test log1p in all rounding modes.
gen-auto-libm-tests presently allows but does not require underflow
exceptions for results with magnitude in the range (greatest
subnormal, least normal].
In some cases, the magnitude of the exact result is very slightly
above the least normal, but rounding in the implementation results in
it effectively computing an infinite-precision result that is slightly
below the least normal, so raising an underflow exception. This is in
accordance with the documented accuracy goals, but results in
testsuite failures.
This patch changes the logic to allow underflows when the mathematical
result is up to 0.5ulp above the least normal (so in any case where
the round-to-nearest result is the least normal). Ideally underflows
in all these cases would be accepted only when an underflow with the
actual result is consistent with the rounding mode (in FE_TOWARDZERO
mode, a return value of the least normal implies that the
infinite-precision result did not underflow so there should be no
underflow exception, for example), so as to match the documented goals
more precisely - whereas at present the tests for exceptions are
completely independent of the tests of the returned values. (The same
applies to overflow exceptions as well - they too should be checked
for consistency with the result, as in FE_TOWARDZERO mode a result
1ulp below the largest finite value should be inconsistent with an
overflow exception and cause a failure with overflow rather than
simply being considered a 1ulp error when overflow is expected.) But
the present patch at least deals with the cases causing spurious
failures so that (a) certain existing tests no longer need to be
marked as having spurious exceptions (such markings in
auto-libm-test-in end up applying to more cases than just those they
are needed for) and (b) log1p can be tested in all rounding modes
without introducing more such failures. This patch duly moves tests
of log1p to ALL_RM_TEST.
Tested x86_64 and x86 and ulps updated accordingly.
[BZ #16357]
[BZ #16599]
* math/gen-auto-libm-tests.c (fp_format_desc): Add field
min_plus_half.
(fp_formats): Update initializers.
(init_fp_formats): Initialize new field.
(output_for_one_input_case): Allow underflow for results up to
min_plus_half.
* math/libm-test.inc (log1p_test): Use ALL_RM_TEST.
* math/auto-libm-test-in: Don't mark some underflows from asin and
atanh as spurious.
* math/auto-libm-test-out: Regenerated.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2014-03-25 12:26:06 +00:00
|
|
|
Function: "log1p_downward":
|
|
|
|
|
double: 1
|
|
|
|
|
float: 1
|
|
|
|
|
idouble: 1
|
|
|
|
|
ifloat: 1
|
|
|
|
|
ildouble: 2
|
|
|
|
|
ldouble: 2
|
|
|
|
|
|
|
|
|
|
Function: "log1p_towardzero":
|
|
|
|
|
double: 1
|
|
|
|
|
float: 1
|
|
|
|
|
idouble: 1
|
|
|
|
|
ifloat: 1
|
|
|
|
|
ildouble: 2
|
|
|
|
|
ldouble: 2
|
|
|
|
|
|
|
|
|
|
Function: "log1p_upward":
|
|
|
|
|
double: 1
|
|
|
|
|
float: 1
|
|
|
|
|
idouble: 1
|
|
|
|
|
ifloat: 1
|
|
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
|
|
|
|
|
Add more tests of log2.
In testing for x86_64 on an AMD processor, I observed libm test
failures of the form:
testing long double (without inline functions)
Failure: Test: log2_downward (0x2.b7e151628aed4p+0)
Result:
is: 1.44269504088896356633e+00 0xb.8aa3b295c17f67600000p-3
should be: 1.44269504088896356622e+00 0xb.8aa3b295c17f67500000p-3
difference: 1.08420217248550443400e-19 0x8.00000000000000000000p-66
ulp : 1.0000
max.ulp : 0.0000
Maximal error of `log2_downward'
is : 1 ulp
accepted: 0 ulp
These issues arise because the maximum ulps when regenerating on one
processor are not the same as on another processor, so regeneration on
several processors may be needed when updating libm-test-ulps to avoid
failures for some users testing glibc - but such regeneration on
multiple processors is inconvenient. Causes can be: on x86 and, for
x86_64, for long double, variation in results of x87 instructions for
transcendental operations between processors; on x86, variation in
compiler excess precision between compiler versions and
configurations; on any processor where the compiler may contract
expressions using fused multiply-add, variation in what contraction
occurs.
Although it's hard to be sure libm-test-ulps covers all ulps that may
be seen in any configuration for the given architecture, in practice
it helps simply to add wider test coverage to make it more likely
that, when testing on one processor, the ulps seen are the biggest
that can be seen for that function on that processor, and hopefully
they are also the biggest that can be seen for that function in other
configurations for that architecture. Thus, this patch adds some
tests of log2 that increase the ulps I see on x86_64 on an Intel
processor, so that hopefully future from-scratch regenerations on that
processor will produce ulps big enough not to have errors from testing
on AMD processors. These tests were found by randomly generating
inputs and seeing what produced ulps larger than those currently in
libm-test-ulps. Of course such increases also improve the accuracy of
the empirical table of known ulps generated from libm-test-ulps files
that goes in the manual.
Tested for x86_64 and x86 and ulps updated accordingly.
* math/auto-libm-test-in: Add more tests of log2.
* math/auto-libm-test-out: Regenerated.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2015-03-24 23:06:28 +00:00
|
|
|
Function: "log2":
|
|
|
|
|
double: 1
|
|
|
|
|
float: 1
|
|
|
|
|
idouble: 1
|
|
|
|
|
ifloat: 1
|
|
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
|
|
|
|
|
2014-06-10 12:07:15 +00:00
|
|
|
Function: "log2_downward":
|
Add more tests of log2.
In testing for x86_64 on an AMD processor, I observed libm test
failures of the form:
testing long double (without inline functions)
Failure: Test: log2_downward (0x2.b7e151628aed4p+0)
Result:
is: 1.44269504088896356633e+00 0xb.8aa3b295c17f67600000p-3
should be: 1.44269504088896356622e+00 0xb.8aa3b295c17f67500000p-3
difference: 1.08420217248550443400e-19 0x8.00000000000000000000p-66
ulp : 1.0000
max.ulp : 0.0000
Maximal error of `log2_downward'
is : 1 ulp
accepted: 0 ulp
These issues arise because the maximum ulps when regenerating on one
processor are not the same as on another processor, so regeneration on
several processors may be needed when updating libm-test-ulps to avoid
failures for some users testing glibc - but such regeneration on
multiple processors is inconvenient. Causes can be: on x86 and, for
x86_64, for long double, variation in results of x87 instructions for
transcendental operations between processors; on x86, variation in
compiler excess precision between compiler versions and
configurations; on any processor where the compiler may contract
expressions using fused multiply-add, variation in what contraction
occurs.
Although it's hard to be sure libm-test-ulps covers all ulps that may
be seen in any configuration for the given architecture, in practice
it helps simply to add wider test coverage to make it more likely
that, when testing on one processor, the ulps seen are the biggest
that can be seen for that function on that processor, and hopefully
they are also the biggest that can be seen for that function in other
configurations for that architecture. Thus, this patch adds some
tests of log2 that increase the ulps I see on x86_64 on an Intel
processor, so that hopefully future from-scratch regenerations on that
processor will produce ulps big enough not to have errors from testing
on AMD processors. These tests were found by randomly generating
inputs and seeing what produced ulps larger than those currently in
libm-test-ulps. Of course such increases also improve the accuracy of
the empirical table of known ulps generated from libm-test-ulps files
that goes in the manual.
Tested for x86_64 and x86 and ulps updated accordingly.
* math/auto-libm-test-in: Add more tests of log2.
* math/auto-libm-test-out: Regenerated.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2015-03-24 23:06:28 +00:00
|
|
|
double: 3
|
|
|
|
|
float: 3
|
|
|
|
|
idouble: 3
|
|
|
|
|
ifloat: 3
|
|
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
2014-06-10 12:07:15 +00:00
|
|
|
|
|
|
|
|
Function: "log2_towardzero":
|
|
|
|
|
double: 1
|
Add more tests of log2.
In testing for x86_64 on an AMD processor, I observed libm test
failures of the form:
testing long double (without inline functions)
Failure: Test: log2_downward (0x2.b7e151628aed4p+0)
Result:
is: 1.44269504088896356633e+00 0xb.8aa3b295c17f67600000p-3
should be: 1.44269504088896356622e+00 0xb.8aa3b295c17f67500000p-3
difference: 1.08420217248550443400e-19 0x8.00000000000000000000p-66
ulp : 1.0000
max.ulp : 0.0000
Maximal error of `log2_downward'
is : 1 ulp
accepted: 0 ulp
These issues arise because the maximum ulps when regenerating on one
processor are not the same as on another processor, so regeneration on
several processors may be needed when updating libm-test-ulps to avoid
failures for some users testing glibc - but such regeneration on
multiple processors is inconvenient. Causes can be: on x86 and, for
x86_64, for long double, variation in results of x87 instructions for
transcendental operations between processors; on x86, variation in
compiler excess precision between compiler versions and
configurations; on any processor where the compiler may contract
expressions using fused multiply-add, variation in what contraction
occurs.
Although it's hard to be sure libm-test-ulps covers all ulps that may
be seen in any configuration for the given architecture, in practice
it helps simply to add wider test coverage to make it more likely
that, when testing on one processor, the ulps seen are the biggest
that can be seen for that function on that processor, and hopefully
they are also the biggest that can be seen for that function in other
configurations for that architecture. Thus, this patch adds some
tests of log2 that increase the ulps I see on x86_64 on an Intel
processor, so that hopefully future from-scratch regenerations on that
processor will produce ulps big enough not to have errors from testing
on AMD processors. These tests were found by randomly generating
inputs and seeing what produced ulps larger than those currently in
libm-test-ulps. Of course such increases also improve the accuracy of
the empirical table of known ulps generated from libm-test-ulps files
that goes in the manual.
Tested for x86_64 and x86 and ulps updated accordingly.
* math/auto-libm-test-in: Add more tests of log2.
* math/auto-libm-test-out: Regenerated.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2015-03-24 23:06:28 +00:00
|
|
|
float: 2
|
2014-06-10 12:07:15 +00:00
|
|
|
idouble: 1
|
Add more tests of log2.
In testing for x86_64 on an AMD processor, I observed libm test
failures of the form:
testing long double (without inline functions)
Failure: Test: log2_downward (0x2.b7e151628aed4p+0)
Result:
is: 1.44269504088896356633e+00 0xb.8aa3b295c17f67600000p-3
should be: 1.44269504088896356622e+00 0xb.8aa3b295c17f67500000p-3
difference: 1.08420217248550443400e-19 0x8.00000000000000000000p-66
ulp : 1.0000
max.ulp : 0.0000
Maximal error of `log2_downward'
is : 1 ulp
accepted: 0 ulp
These issues arise because the maximum ulps when regenerating on one
processor are not the same as on another processor, so regeneration on
several processors may be needed when updating libm-test-ulps to avoid
failures for some users testing glibc - but such regeneration on
multiple processors is inconvenient. Causes can be: on x86 and, for
x86_64, for long double, variation in results of x87 instructions for
transcendental operations between processors; on x86, variation in
compiler excess precision between compiler versions and
configurations; on any processor where the compiler may contract
expressions using fused multiply-add, variation in what contraction
occurs.
Although it's hard to be sure libm-test-ulps covers all ulps that may
be seen in any configuration for the given architecture, in practice
it helps simply to add wider test coverage to make it more likely
that, when testing on one processor, the ulps seen are the biggest
that can be seen for that function on that processor, and hopefully
they are also the biggest that can be seen for that function in other
configurations for that architecture. Thus, this patch adds some
tests of log2 that increase the ulps I see on x86_64 on an Intel
processor, so that hopefully future from-scratch regenerations on that
processor will produce ulps big enough not to have errors from testing
on AMD processors. These tests were found by randomly generating
inputs and seeing what produced ulps larger than those currently in
libm-test-ulps. Of course such increases also improve the accuracy of
the empirical table of known ulps generated from libm-test-ulps files
that goes in the manual.
Tested for x86_64 and x86 and ulps updated accordingly.
* math/auto-libm-test-in: Add more tests of log2.
* math/auto-libm-test-out: Regenerated.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2015-03-24 23:06:28 +00:00
|
|
|
ifloat: 2
|
2014-06-10 12:07:15 +00:00
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
|
|
|
|
|
|
|
|
|
Function: "log2_upward":
|
Add more tests of log2.
In testing for x86_64 on an AMD processor, I observed libm test
failures of the form:
testing long double (without inline functions)
Failure: Test: log2_downward (0x2.b7e151628aed4p+0)
Result:
is: 1.44269504088896356633e+00 0xb.8aa3b295c17f67600000p-3
should be: 1.44269504088896356622e+00 0xb.8aa3b295c17f67500000p-3
difference: 1.08420217248550443400e-19 0x8.00000000000000000000p-66
ulp : 1.0000
max.ulp : 0.0000
Maximal error of `log2_downward'
is : 1 ulp
accepted: 0 ulp
These issues arise because the maximum ulps when regenerating on one
processor are not the same as on another processor, so regeneration on
several processors may be needed when updating libm-test-ulps to avoid
failures for some users testing glibc - but such regeneration on
multiple processors is inconvenient. Causes can be: on x86 and, for
x86_64, for long double, variation in results of x87 instructions for
transcendental operations between processors; on x86, variation in
compiler excess precision between compiler versions and
configurations; on any processor where the compiler may contract
expressions using fused multiply-add, variation in what contraction
occurs.
Although it's hard to be sure libm-test-ulps covers all ulps that may
be seen in any configuration for the given architecture, in practice
it helps simply to add wider test coverage to make it more likely
that, when testing on one processor, the ulps seen are the biggest
that can be seen for that function on that processor, and hopefully
they are also the biggest that can be seen for that function in other
configurations for that architecture. Thus, this patch adds some
tests of log2 that increase the ulps I see on x86_64 on an Intel
processor, so that hopefully future from-scratch regenerations on that
processor will produce ulps big enough not to have errors from testing
on AMD processors. These tests were found by randomly generating
inputs and seeing what produced ulps larger than those currently in
libm-test-ulps. Of course such increases also improve the accuracy of
the empirical table of known ulps generated from libm-test-ulps files
that goes in the manual.
Tested for x86_64 and x86 and ulps updated accordingly.
* math/auto-libm-test-in: Add more tests of log2.
* math/auto-libm-test-out: Regenerated.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2015-03-24 23:06:28 +00:00
|
|
|
double: 3
|
2014-06-10 12:07:15 +00:00
|
|
|
float: 2
|
Add more tests of log2.
In testing for x86_64 on an AMD processor, I observed libm test
failures of the form:
testing long double (without inline functions)
Failure: Test: log2_downward (0x2.b7e151628aed4p+0)
Result:
is: 1.44269504088896356633e+00 0xb.8aa3b295c17f67600000p-3
should be: 1.44269504088896356622e+00 0xb.8aa3b295c17f67500000p-3
difference: 1.08420217248550443400e-19 0x8.00000000000000000000p-66
ulp : 1.0000
max.ulp : 0.0000
Maximal error of `log2_downward'
is : 1 ulp
accepted: 0 ulp
These issues arise because the maximum ulps when regenerating on one
processor are not the same as on another processor, so regeneration on
several processors may be needed when updating libm-test-ulps to avoid
failures for some users testing glibc - but such regeneration on
multiple processors is inconvenient. Causes can be: on x86 and, for
x86_64, for long double, variation in results of x87 instructions for
transcendental operations between processors; on x86, variation in
compiler excess precision between compiler versions and
configurations; on any processor where the compiler may contract
expressions using fused multiply-add, variation in what contraction
occurs.
Although it's hard to be sure libm-test-ulps covers all ulps that may
be seen in any configuration for the given architecture, in practice
it helps simply to add wider test coverage to make it more likely
that, when testing on one processor, the ulps seen are the biggest
that can be seen for that function on that processor, and hopefully
they are also the biggest that can be seen for that function in other
configurations for that architecture. Thus, this patch adds some
tests of log2 that increase the ulps I see on x86_64 on an Intel
processor, so that hopefully future from-scratch regenerations on that
processor will produce ulps big enough not to have errors from testing
on AMD processors. These tests were found by randomly generating
inputs and seeing what produced ulps larger than those currently in
libm-test-ulps. Of course such increases also improve the accuracy of
the empirical table of known ulps generated from libm-test-ulps files
that goes in the manual.
Tested for x86_64 and x86 and ulps updated accordingly.
* math/auto-libm-test-in: Add more tests of log2.
* math/auto-libm-test-out: Regenerated.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2015-03-24 23:06:28 +00:00
|
|
|
idouble: 3
|
2014-06-10 12:07:15 +00:00
|
|
|
ifloat: 2
|
|
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
|
|
|
|
|
2014-03-21 18:13:58 +00:00
|
|
|
Function: "log_downward":
|
|
|
|
|
float: 1
|
|
|
|
|
ifloat: 1
|
|
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
|
|
|
|
|
|
|
|
|
Function: "log_towardzero":
|
|
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
|
|
|
|
|
|
|
|
|
Function: "log_upward":
|
|
|
|
|
float: 1
|
|
|
|
|
ifloat: 1
|
|
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
|
|
|
|
|
2012-04-09 09:42:05 +00:00
|
|
|
Function: "pow":
|
|
|
|
|
float: 1
|
|
|
|
|
ifloat: 1
|
2012-11-28 13:40:54 +00:00
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
2012-04-09 09:42:05 +00:00
|
|
|
|
2013-05-24 20:33:14 +00:00
|
|
|
Function: "pow10":
|
|
|
|
|
double: 1
|
|
|
|
|
idouble: 1
|
|
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
|
|
|
|
|
Test most libm functions in all rounding modes.
This patch makes libm-test.inc tests of most functions use ALL_RM_TEST
unless there was some reason to defer that change for a particular
function.
I started out planning to defer the change for pow (bug 16315), cexp /
ccos / ccosh / csin / csinh (likely fallout from exp, bug 16284) and
cpow (exact expectations for signs of exact zero results not wanted).
Testing on x86_64 and x86 showed additional failures for acosh, cacos,
catan, catanh, clog, clog10, jn, log, log10, log1p, log2, tgamma, yn,
so making the change for those functions was deferred as well, pending
investigation to show which of these represent distinct bugs (some
such bugs may already be filed) and appropriate fixing / XFAILing.
Failures include wrong signs of zero results, errors slightly above
the 9ulp bound (in such cases it may make sense for functions to set
round-to-nearest internally to reduce error accumulation), large
errors and incorrect overflow/underflow for the rounding mode (with
consequent missing errno settings in some cases). It's possible some
could be issues with test expectations, though I didn't notice any
that were obviously like that (I added NO_TEST_INLINE for cases that
were failing for ildoubl on x86 and where it seemed reasonable for
them to fail for the fast-math inlines).
There may of course be failures on other architectures for functions
that didn't fail on x86_64 or x86, in which case the usual rule
applies: file a bug (preferably identifying the underlying problem
function, in cases where function A calls function B and a problem
with function B may present in the test results for function A) if not
already in Bugzilla then fix or XFAIL.
Tested x86_64 and x86 and ulps updated accordingly.
* math/libm-test.inc (asinh_test): Use ALL_RM_TEST.
(atan_test): Likewise.
(atanh_test_data): Use NO_TEST_INLINE for two tests.
(atanh_test): Use ALL_RM_TEST.
(atan2_test_data): Likewise.
(cabs_test): Likewise.
(cacosh_test): Likewise.
(carg_test): Likewise.
(casin_test): Likewise.
(casinh_test): Likewise.
(cbrt_test): Likewise.
(csqrt_test): Likewise.
(erf_test): Likewise.
(erfc_test): Likewise.
(pow10_test): Likewise.
(exp2_test): Likewise.
(hypot_test): Likewise.
(j0_test): Likewise.
(j1_test): Likewise.
(lgamma_test): Likewise.
(gamma_test): Likewise.
(sincos_test): Likewise.
(tanh_test): Likewise.
(y0_test): Likewise.
(y1_test): Likewise.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2014-03-21 00:03:38 +00:00
|
|
|
Function: "pow10_downward":
|
|
|
|
|
double: 1
|
2014-06-25 11:33:22 +00:00
|
|
|
float: 1
|
Test most libm functions in all rounding modes.
This patch makes libm-test.inc tests of most functions use ALL_RM_TEST
unless there was some reason to defer that change for a particular
function.
I started out planning to defer the change for pow (bug 16315), cexp /
ccos / ccosh / csin / csinh (likely fallout from exp, bug 16284) and
cpow (exact expectations for signs of exact zero results not wanted).
Testing on x86_64 and x86 showed additional failures for acosh, cacos,
catan, catanh, clog, clog10, jn, log, log10, log1p, log2, tgamma, yn,
so making the change for those functions was deferred as well, pending
investigation to show which of these represent distinct bugs (some
such bugs may already be filed) and appropriate fixing / XFAILing.
Failures include wrong signs of zero results, errors slightly above
the 9ulp bound (in such cases it may make sense for functions to set
round-to-nearest internally to reduce error accumulation), large
errors and incorrect overflow/underflow for the rounding mode (with
consequent missing errno settings in some cases). It's possible some
could be issues with test expectations, though I didn't notice any
that were obviously like that (I added NO_TEST_INLINE for cases that
were failing for ildoubl on x86 and where it seemed reasonable for
them to fail for the fast-math inlines).
There may of course be failures on other architectures for functions
that didn't fail on x86_64 or x86, in which case the usual rule
applies: file a bug (preferably identifying the underlying problem
function, in cases where function A calls function B and a problem
with function B may present in the test results for function A) if not
already in Bugzilla then fix or XFAIL.
Tested x86_64 and x86 and ulps updated accordingly.
* math/libm-test.inc (asinh_test): Use ALL_RM_TEST.
(atan_test): Likewise.
(atanh_test_data): Use NO_TEST_INLINE for two tests.
(atanh_test): Use ALL_RM_TEST.
(atan2_test_data): Likewise.
(cabs_test): Likewise.
(cacosh_test): Likewise.
(carg_test): Likewise.
(casin_test): Likewise.
(casinh_test): Likewise.
(cbrt_test): Likewise.
(csqrt_test): Likewise.
(erf_test): Likewise.
(erfc_test): Likewise.
(pow10_test): Likewise.
(exp2_test): Likewise.
(hypot_test): Likewise.
(j0_test): Likewise.
(j1_test): Likewise.
(lgamma_test): Likewise.
(gamma_test): Likewise.
(sincos_test): Likewise.
(tanh_test): Likewise.
(y0_test): Likewise.
(y1_test): Likewise.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2014-03-21 00:03:38 +00:00
|
|
|
idouble: 1
|
2014-06-25 11:33:22 +00:00
|
|
|
ifloat: 1
|
Test most libm functions in all rounding modes.
This patch makes libm-test.inc tests of most functions use ALL_RM_TEST
unless there was some reason to defer that change for a particular
function.
I started out planning to defer the change for pow (bug 16315), cexp /
ccos / ccosh / csin / csinh (likely fallout from exp, bug 16284) and
cpow (exact expectations for signs of exact zero results not wanted).
Testing on x86_64 and x86 showed additional failures for acosh, cacos,
catan, catanh, clog, clog10, jn, log, log10, log1p, log2, tgamma, yn,
so making the change for those functions was deferred as well, pending
investigation to show which of these represent distinct bugs (some
such bugs may already be filed) and appropriate fixing / XFAILing.
Failures include wrong signs of zero results, errors slightly above
the 9ulp bound (in such cases it may make sense for functions to set
round-to-nearest internally to reduce error accumulation), large
errors and incorrect overflow/underflow for the rounding mode (with
consequent missing errno settings in some cases). It's possible some
could be issues with test expectations, though I didn't notice any
that were obviously like that (I added NO_TEST_INLINE for cases that
were failing for ildoubl on x86 and where it seemed reasonable for
them to fail for the fast-math inlines).
There may of course be failures on other architectures for functions
that didn't fail on x86_64 or x86, in which case the usual rule
applies: file a bug (preferably identifying the underlying problem
function, in cases where function A calls function B and a problem
with function B may present in the test results for function A) if not
already in Bugzilla then fix or XFAIL.
Tested x86_64 and x86 and ulps updated accordingly.
* math/libm-test.inc (asinh_test): Use ALL_RM_TEST.
(atan_test): Likewise.
(atanh_test_data): Use NO_TEST_INLINE for two tests.
(atanh_test): Use ALL_RM_TEST.
(atan2_test_data): Likewise.
(cabs_test): Likewise.
(cacosh_test): Likewise.
(carg_test): Likewise.
(casin_test): Likewise.
(casinh_test): Likewise.
(cbrt_test): Likewise.
(csqrt_test): Likewise.
(erf_test): Likewise.
(erfc_test): Likewise.
(pow10_test): Likewise.
(exp2_test): Likewise.
(hypot_test): Likewise.
(j0_test): Likewise.
(j1_test): Likewise.
(lgamma_test): Likewise.
(gamma_test): Likewise.
(sincos_test): Likewise.
(tanh_test): Likewise.
(y0_test): Likewise.
(y1_test): Likewise.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2014-03-21 00:03:38 +00:00
|
|
|
ildouble: 2
|
|
|
|
|
ldouble: 2
|
|
|
|
|
|
|
|
|
|
Function: "pow10_towardzero":
|
|
|
|
|
double: 1
|
2014-06-25 11:33:22 +00:00
|
|
|
float: 1
|
Test most libm functions in all rounding modes.
This patch makes libm-test.inc tests of most functions use ALL_RM_TEST
unless there was some reason to defer that change for a particular
function.
I started out planning to defer the change for pow (bug 16315), cexp /
ccos / ccosh / csin / csinh (likely fallout from exp, bug 16284) and
cpow (exact expectations for signs of exact zero results not wanted).
Testing on x86_64 and x86 showed additional failures for acosh, cacos,
catan, catanh, clog, clog10, jn, log, log10, log1p, log2, tgamma, yn,
so making the change for those functions was deferred as well, pending
investigation to show which of these represent distinct bugs (some
such bugs may already be filed) and appropriate fixing / XFAILing.
Failures include wrong signs of zero results, errors slightly above
the 9ulp bound (in such cases it may make sense for functions to set
round-to-nearest internally to reduce error accumulation), large
errors and incorrect overflow/underflow for the rounding mode (with
consequent missing errno settings in some cases). It's possible some
could be issues with test expectations, though I didn't notice any
that were obviously like that (I added NO_TEST_INLINE for cases that
were failing for ildoubl on x86 and where it seemed reasonable for
them to fail for the fast-math inlines).
There may of course be failures on other architectures for functions
that didn't fail on x86_64 or x86, in which case the usual rule
applies: file a bug (preferably identifying the underlying problem
function, in cases where function A calls function B and a problem
with function B may present in the test results for function A) if not
already in Bugzilla then fix or XFAIL.
Tested x86_64 and x86 and ulps updated accordingly.
* math/libm-test.inc (asinh_test): Use ALL_RM_TEST.
(atan_test): Likewise.
(atanh_test_data): Use NO_TEST_INLINE for two tests.
(atanh_test): Use ALL_RM_TEST.
(atan2_test_data): Likewise.
(cabs_test): Likewise.
(cacosh_test): Likewise.
(carg_test): Likewise.
(casin_test): Likewise.
(casinh_test): Likewise.
(cbrt_test): Likewise.
(csqrt_test): Likewise.
(erf_test): Likewise.
(erfc_test): Likewise.
(pow10_test): Likewise.
(exp2_test): Likewise.
(hypot_test): Likewise.
(j0_test): Likewise.
(j1_test): Likewise.
(lgamma_test): Likewise.
(gamma_test): Likewise.
(sincos_test): Likewise.
(tanh_test): Likewise.
(y0_test): Likewise.
(y1_test): Likewise.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2014-03-21 00:03:38 +00:00
|
|
|
idouble: 1
|
2014-06-25 11:33:22 +00:00
|
|
|
ifloat: 1
|
Test most libm functions in all rounding modes.
This patch makes libm-test.inc tests of most functions use ALL_RM_TEST
unless there was some reason to defer that change for a particular
function.
I started out planning to defer the change for pow (bug 16315), cexp /
ccos / ccosh / csin / csinh (likely fallout from exp, bug 16284) and
cpow (exact expectations for signs of exact zero results not wanted).
Testing on x86_64 and x86 showed additional failures for acosh, cacos,
catan, catanh, clog, clog10, jn, log, log10, log1p, log2, tgamma, yn,
so making the change for those functions was deferred as well, pending
investigation to show which of these represent distinct bugs (some
such bugs may already be filed) and appropriate fixing / XFAILing.
Failures include wrong signs of zero results, errors slightly above
the 9ulp bound (in such cases it may make sense for functions to set
round-to-nearest internally to reduce error accumulation), large
errors and incorrect overflow/underflow for the rounding mode (with
consequent missing errno settings in some cases). It's possible some
could be issues with test expectations, though I didn't notice any
that were obviously like that (I added NO_TEST_INLINE for cases that
were failing for ildoubl on x86 and where it seemed reasonable for
them to fail for the fast-math inlines).
There may of course be failures on other architectures for functions
that didn't fail on x86_64 or x86, in which case the usual rule
applies: file a bug (preferably identifying the underlying problem
function, in cases where function A calls function B and a problem
with function B may present in the test results for function A) if not
already in Bugzilla then fix or XFAIL.
Tested x86_64 and x86 and ulps updated accordingly.
* math/libm-test.inc (asinh_test): Use ALL_RM_TEST.
(atan_test): Likewise.
(atanh_test_data): Use NO_TEST_INLINE for two tests.
(atanh_test): Use ALL_RM_TEST.
(atan2_test_data): Likewise.
(cabs_test): Likewise.
(cacosh_test): Likewise.
(carg_test): Likewise.
(casin_test): Likewise.
(casinh_test): Likewise.
(cbrt_test): Likewise.
(csqrt_test): Likewise.
(erf_test): Likewise.
(erfc_test): Likewise.
(pow10_test): Likewise.
(exp2_test): Likewise.
(hypot_test): Likewise.
(j0_test): Likewise.
(j1_test): Likewise.
(lgamma_test): Likewise.
(gamma_test): Likewise.
(sincos_test): Likewise.
(tanh_test): Likewise.
(y0_test): Likewise.
(y1_test): Likewise.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2014-03-21 00:03:38 +00:00
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
|
|
|
|
|
|
|
|
|
Function: "pow10_upward":
|
|
|
|
|
double: 1
|
|
|
|
|
float: 1
|
|
|
|
|
idouble: 1
|
|
|
|
|
ifloat: 1
|
|
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
|
|
|
|
|
2012-03-05 12:22:46 +00:00
|
|
|
Function: "pow_downward":
|
2014-06-23 20:12:33 +00:00
|
|
|
double: 1
|
2012-03-05 12:22:46 +00:00
|
|
|
float: 1
|
2014-06-23 20:12:33 +00:00
|
|
|
idouble: 1
|
2012-03-05 12:22:46 +00:00
|
|
|
ifloat: 1
|
2014-06-23 20:12:33 +00:00
|
|
|
ildouble: 4
|
|
|
|
|
ldouble: 4
|
2012-03-05 12:22:46 +00:00
|
|
|
|
|
|
|
|
Function: "pow_towardzero":
|
2014-06-23 20:12:33 +00:00
|
|
|
double: 1
|
2012-03-05 12:22:46 +00:00
|
|
|
float: 1
|
2014-06-23 20:12:33 +00:00
|
|
|
idouble: 1
|
2012-03-05 12:22:46 +00:00
|
|
|
ifloat: 1
|
2014-06-23 20:12:33 +00:00
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
2012-03-05 12:22:46 +00:00
|
|
|
|
|
|
|
|
Function: "pow_upward":
|
2014-06-23 20:12:33 +00:00
|
|
|
double: 1
|
2012-03-05 12:22:46 +00:00
|
|
|
float: 1
|
2014-06-23 20:12:33 +00:00
|
|
|
idouble: 1
|
2012-03-05 12:22:46 +00:00
|
|
|
ifloat: 1
|
2014-06-23 20:12:33 +00:00
|
|
|
ildouble: 2
|
|
|
|
|
ldouble: 2
|
2012-03-05 12:22:46 +00:00
|
|
|
|
2013-12-05 13:54:50 +00:00
|
|
|
Function: "sin":
|
|
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
|
|
|
|
|
2012-03-02 20:51:39 +00:00
|
|
|
Function: "sin_downward":
|
2013-12-05 13:54:50 +00:00
|
|
|
double: 1
|
|
|
|
|
idouble: 1
|
|
|
|
|
ildouble: 2
|
|
|
|
|
ldouble: 2
|
|
|
|
|
|
2012-03-02 20:51:39 +00:00
|
|
|
Function: "sin_towardzero":
|
2013-12-05 13:54:50 +00:00
|
|
|
double: 1
|
|
|
|
|
idouble: 1
|
2012-03-02 20:51:39 +00:00
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
|
|
|
|
|
|
|
|
|
Function: "sin_upward":
|
2013-12-05 13:54:50 +00:00
|
|
|
double: 1
|
|
|
|
|
idouble: 1
|
|
|
|
|
ildouble: 2
|
|
|
|
|
ldouble: 2
|
2012-03-02 20:51:39 +00:00
|
|
|
|
2001-09-19 10:37:31 +00:00
|
|
|
Function: "sincos":
|
2013-07-05 12:58:20 +00:00
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
|
|
|
|
|
Test most libm functions in all rounding modes.
This patch makes libm-test.inc tests of most functions use ALL_RM_TEST
unless there was some reason to defer that change for a particular
function.
I started out planning to defer the change for pow (bug 16315), cexp /
ccos / ccosh / csin / csinh (likely fallout from exp, bug 16284) and
cpow (exact expectations for signs of exact zero results not wanted).
Testing on x86_64 and x86 showed additional failures for acosh, cacos,
catan, catanh, clog, clog10, jn, log, log10, log1p, log2, tgamma, yn,
so making the change for those functions was deferred as well, pending
investigation to show which of these represent distinct bugs (some
such bugs may already be filed) and appropriate fixing / XFAILing.
Failures include wrong signs of zero results, errors slightly above
the 9ulp bound (in such cases it may make sense for functions to set
round-to-nearest internally to reduce error accumulation), large
errors and incorrect overflow/underflow for the rounding mode (with
consequent missing errno settings in some cases). It's possible some
could be issues with test expectations, though I didn't notice any
that were obviously like that (I added NO_TEST_INLINE for cases that
were failing for ildoubl on x86 and where it seemed reasonable for
them to fail for the fast-math inlines).
There may of course be failures on other architectures for functions
that didn't fail on x86_64 or x86, in which case the usual rule
applies: file a bug (preferably identifying the underlying problem
function, in cases where function A calls function B and a problem
with function B may present in the test results for function A) if not
already in Bugzilla then fix or XFAIL.
Tested x86_64 and x86 and ulps updated accordingly.
* math/libm-test.inc (asinh_test): Use ALL_RM_TEST.
(atan_test): Likewise.
(atanh_test_data): Use NO_TEST_INLINE for two tests.
(atanh_test): Use ALL_RM_TEST.
(atan2_test_data): Likewise.
(cabs_test): Likewise.
(cacosh_test): Likewise.
(carg_test): Likewise.
(casin_test): Likewise.
(casinh_test): Likewise.
(cbrt_test): Likewise.
(csqrt_test): Likewise.
(erf_test): Likewise.
(erfc_test): Likewise.
(pow10_test): Likewise.
(exp2_test): Likewise.
(hypot_test): Likewise.
(j0_test): Likewise.
(j1_test): Likewise.
(lgamma_test): Likewise.
(gamma_test): Likewise.
(sincos_test): Likewise.
(tanh_test): Likewise.
(y0_test): Likewise.
(y1_test): Likewise.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2014-03-21 00:03:38 +00:00
|
|
|
Function: "sincos_downward":
|
|
|
|
|
double: 1
|
|
|
|
|
idouble: 1
|
|
|
|
|
ildouble: 2
|
|
|
|
|
ldouble: 2
|
|
|
|
|
|
|
|
|
|
Function: "sincos_towardzero":
|
|
|
|
|
double: 1
|
|
|
|
|
idouble: 1
|
|
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
|
|
|
|
|
|
|
|
|
Function: "sincos_upward":
|
|
|
|
|
double: 1
|
|
|
|
|
idouble: 1
|
|
|
|
|
ildouble: 2
|
|
|
|
|
ldouble: 2
|
|
|
|
|
|
2015-03-24 23:48:04 +00:00
|
|
|
Function: "sinh":
|
2013-12-05 13:54:50 +00:00
|
|
|
double: 1
|
2015-03-24 23:48:04 +00:00
|
|
|
float: 1
|
2013-12-05 13:54:50 +00:00
|
|
|
idouble: 1
|
2015-03-24 23:48:04 +00:00
|
|
|
ifloat: 1
|
2014-01-01 14:34:38 +00:00
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
2012-03-05 12:20:24 +00:00
|
|
|
|
2015-03-24 23:48:04 +00:00
|
|
|
Function: "sinh_downward":
|
|
|
|
|
double: 2
|
|
|
|
|
float: 3
|
|
|
|
|
idouble: 2
|
|
|
|
|
ifloat: 3
|
|
|
|
|
ildouble: 4
|
|
|
|
|
ldouble: 4
|
|
|
|
|
|
2012-03-05 12:20:24 +00:00
|
|
|
Function: "sinh_towardzero":
|
2015-03-24 23:48:04 +00:00
|
|
|
double: 2
|
|
|
|
|
float: 2
|
|
|
|
|
idouble: 2
|
|
|
|
|
ifloat: 2
|
|
|
|
|
ildouble: 3
|
|
|
|
|
ldouble: 3
|
2012-03-05 12:20:24 +00:00
|
|
|
|
|
|
|
|
Function: "sinh_upward":
|
2015-03-24 23:48:04 +00:00
|
|
|
double: 2
|
|
|
|
|
float: 3
|
|
|
|
|
idouble: 2
|
|
|
|
|
ifloat: 3
|
|
|
|
|
ildouble: 4
|
|
|
|
|
ldouble: 4
|
2012-03-05 12:20:24 +00:00
|
|
|
|
2001-09-19 10:37:31 +00:00
|
|
|
Function: "tan":
|
2012-03-16 20:05:04 +00:00
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
2001-09-19 10:37:31 +00:00
|
|
|
|
2012-03-02 20:51:39 +00:00
|
|
|
Function: "tan_downward":
|
2013-12-05 13:54:50 +00:00
|
|
|
double: 1
|
|
|
|
|
float: 2
|
|
|
|
|
idouble: 1
|
|
|
|
|
ifloat: 2
|
2012-03-02 20:51:39 +00:00
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
|
|
|
|
|
|
|
|
|
Function: "tan_towardzero":
|
2013-12-05 13:54:50 +00:00
|
|
|
double: 1
|
2012-03-02 20:51:39 +00:00
|
|
|
float: 1
|
2013-12-05 13:54:50 +00:00
|
|
|
idouble: 1
|
2012-03-02 20:51:39 +00:00
|
|
|
ifloat: 1
|
2013-07-02 20:00:48 +00:00
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
2012-03-02 20:51:39 +00:00
|
|
|
|
|
|
|
|
Function: "tan_upward":
|
2013-12-05 13:54:50 +00:00
|
|
|
double: 1
|
2012-03-02 20:51:39 +00:00
|
|
|
float: 1
|
2013-12-05 13:54:50 +00:00
|
|
|
idouble: 1
|
2012-03-02 20:51:39 +00:00
|
|
|
ifloat: 1
|
|
|
|
|
ildouble: 2
|
|
|
|
|
ldouble: 2
|
|
|
|
|
|
Test most libm functions in all rounding modes.
This patch makes libm-test.inc tests of most functions use ALL_RM_TEST
unless there was some reason to defer that change for a particular
function.
I started out planning to defer the change for pow (bug 16315), cexp /
ccos / ccosh / csin / csinh (likely fallout from exp, bug 16284) and
cpow (exact expectations for signs of exact zero results not wanted).
Testing on x86_64 and x86 showed additional failures for acosh, cacos,
catan, catanh, clog, clog10, jn, log, log10, log1p, log2, tgamma, yn,
so making the change for those functions was deferred as well, pending
investigation to show which of these represent distinct bugs (some
such bugs may already be filed) and appropriate fixing / XFAILing.
Failures include wrong signs of zero results, errors slightly above
the 9ulp bound (in such cases it may make sense for functions to set
round-to-nearest internally to reduce error accumulation), large
errors and incorrect overflow/underflow for the rounding mode (with
consequent missing errno settings in some cases). It's possible some
could be issues with test expectations, though I didn't notice any
that were obviously like that (I added NO_TEST_INLINE for cases that
were failing for ildoubl on x86 and where it seemed reasonable for
them to fail for the fast-math inlines).
There may of course be failures on other architectures for functions
that didn't fail on x86_64 or x86, in which case the usual rule
applies: file a bug (preferably identifying the underlying problem
function, in cases where function A calls function B and a problem
with function B may present in the test results for function A) if not
already in Bugzilla then fix or XFAIL.
Tested x86_64 and x86 and ulps updated accordingly.
* math/libm-test.inc (asinh_test): Use ALL_RM_TEST.
(atan_test): Likewise.
(atanh_test_data): Use NO_TEST_INLINE for two tests.
(atanh_test): Use ALL_RM_TEST.
(atan2_test_data): Likewise.
(cabs_test): Likewise.
(cacosh_test): Likewise.
(carg_test): Likewise.
(casin_test): Likewise.
(casinh_test): Likewise.
(cbrt_test): Likewise.
(csqrt_test): Likewise.
(erf_test): Likewise.
(erfc_test): Likewise.
(pow10_test): Likewise.
(exp2_test): Likewise.
(hypot_test): Likewise.
(j0_test): Likewise.
(j1_test): Likewise.
(lgamma_test): Likewise.
(gamma_test): Likewise.
(sincos_test): Likewise.
(tanh_test): Likewise.
(y0_test): Likewise.
(y1_test): Likewise.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2014-03-21 00:03:38 +00:00
|
|
|
Function: "tanh_downward":
|
|
|
|
|
double: 1
|
|
|
|
|
float: 1
|
|
|
|
|
idouble: 1
|
|
|
|
|
ifloat: 1
|
|
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
|
|
|
|
|
|
|
|
|
Function: "tanh_towardzero":
|
|
|
|
|
double: 1
|
|
|
|
|
float: 1
|
|
|
|
|
idouble: 1
|
|
|
|
|
ifloat: 1
|
|
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
|
|
|
|
|
|
|
|
|
Function: "tanh_upward":
|
|
|
|
|
double: 1
|
|
|
|
|
float: 1
|
|
|
|
|
idouble: 1
|
|
|
|
|
ifloat: 1
|
|
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
|
|
|
|
|
2001-09-19 10:37:31 +00:00
|
|
|
Function: "tgamma":
|
2013-05-08 11:58:18 +00:00
|
|
|
double: 4
|
|
|
|
|
float: 3
|
|
|
|
|
idouble: 4
|
|
|
|
|
ifloat: 3
|
|
|
|
|
ildouble: 4
|
|
|
|
|
ldouble: 4
|
2002-08-31 08:22:35 +00:00
|
|
|
|
2001-09-19 10:37:31 +00:00
|
|
|
Function: "y0":
|
|
|
|
|
double: 2
|
|
|
|
|
float: 1
|
|
|
|
|
idouble: 2
|
|
|
|
|
ifloat: 1
|
2002-08-31 08:22:35 +00:00
|
|
|
ildouble: 1
|
|
|
|
|
ldouble: 1
|
2001-09-19 10:37:31 +00:00
|
|
|
|
Test most libm functions in all rounding modes.
This patch makes libm-test.inc tests of most functions use ALL_RM_TEST
unless there was some reason to defer that change for a particular
function.
I started out planning to defer the change for pow (bug 16315), cexp /
ccos / ccosh / csin / csinh (likely fallout from exp, bug 16284) and
cpow (exact expectations for signs of exact zero results not wanted).
Testing on x86_64 and x86 showed additional failures for acosh, cacos,
catan, catanh, clog, clog10, jn, log, log10, log1p, log2, tgamma, yn,
so making the change for those functions was deferred as well, pending
investigation to show which of these represent distinct bugs (some
such bugs may already be filed) and appropriate fixing / XFAILing.
Failures include wrong signs of zero results, errors slightly above
the 9ulp bound (in such cases it may make sense for functions to set
round-to-nearest internally to reduce error accumulation), large
errors and incorrect overflow/underflow for the rounding mode (with
consequent missing errno settings in some cases). It's possible some
could be issues with test expectations, though I didn't notice any
that were obviously like that (I added NO_TEST_INLINE for cases that
were failing for ildoubl on x86 and where it seemed reasonable for
them to fail for the fast-math inlines).
There may of course be failures on other architectures for functions
that didn't fail on x86_64 or x86, in which case the usual rule
applies: file a bug (preferably identifying the underlying problem
function, in cases where function A calls function B and a problem
with function B may present in the test results for function A) if not
already in Bugzilla then fix or XFAIL.
Tested x86_64 and x86 and ulps updated accordingly.
* math/libm-test.inc (asinh_test): Use ALL_RM_TEST.
(atan_test): Likewise.
(atanh_test_data): Use NO_TEST_INLINE for two tests.
(atanh_test): Use ALL_RM_TEST.
(atan2_test_data): Likewise.
(cabs_test): Likewise.
(cacosh_test): Likewise.
(carg_test): Likewise.
(casin_test): Likewise.
(casinh_test): Likewise.
(cbrt_test): Likewise.
(csqrt_test): Likewise.
(erf_test): Likewise.
(erfc_test): Likewise.
(pow10_test): Likewise.
(exp2_test): Likewise.
(hypot_test): Likewise.
(j0_test): Likewise.
(j1_test): Likewise.
(lgamma_test): Likewise.
(gamma_test): Likewise.
(sincos_test): Likewise.
(tanh_test): Likewise.
(y0_test): Likewise.
(y1_test): Likewise.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2014-03-21 00:03:38 +00:00
|
|
|
Function: "y0_downward":
|
|
|
|
|
double: 3
|
|
|
|
|
float: 4
|
|
|
|
|
idouble: 3
|
|
|
|
|
ifloat: 4
|
|
|
|
|
ildouble: 5
|
|
|
|
|
ldouble: 5
|
|
|
|
|
|
|
|
|
|
Function: "y0_towardzero":
|
|
|
|
|
double: 3
|
|
|
|
|
float: 3
|
|
|
|
|
idouble: 3
|
|
|
|
|
ifloat: 3
|
|
|
|
|
ildouble: 5
|
|
|
|
|
ldouble: 5
|
|
|
|
|
|
|
|
|
|
Function: "y0_upward":
|
|
|
|
|
double: 3
|
|
|
|
|
float: 5
|
|
|
|
|
idouble: 3
|
|
|
|
|
ifloat: 5
|
|
|
|
|
ildouble: 3
|
|
|
|
|
ldouble: 3
|
|
|
|
|
|
2001-09-19 10:37:31 +00:00
|
|
|
Function: "y1":
|
|
|
|
|
double: 3
|
|
|
|
|
float: 2
|
|
|
|
|
idouble: 3
|
|
|
|
|
ifloat: 2
|
2012-03-16 20:08:02 +00:00
|
|
|
ildouble: 2
|
|
|
|
|
ldouble: 2
|
2001-09-19 10:37:31 +00:00
|
|
|
|
Test most libm functions in all rounding modes.
This patch makes libm-test.inc tests of most functions use ALL_RM_TEST
unless there was some reason to defer that change for a particular
function.
I started out planning to defer the change for pow (bug 16315), cexp /
ccos / ccosh / csin / csinh (likely fallout from exp, bug 16284) and
cpow (exact expectations for signs of exact zero results not wanted).
Testing on x86_64 and x86 showed additional failures for acosh, cacos,
catan, catanh, clog, clog10, jn, log, log10, log1p, log2, tgamma, yn,
so making the change for those functions was deferred as well, pending
investigation to show which of these represent distinct bugs (some
such bugs may already be filed) and appropriate fixing / XFAILing.
Failures include wrong signs of zero results, errors slightly above
the 9ulp bound (in such cases it may make sense for functions to set
round-to-nearest internally to reduce error accumulation), large
errors and incorrect overflow/underflow for the rounding mode (with
consequent missing errno settings in some cases). It's possible some
could be issues with test expectations, though I didn't notice any
that were obviously like that (I added NO_TEST_INLINE for cases that
were failing for ildoubl on x86 and where it seemed reasonable for
them to fail for the fast-math inlines).
There may of course be failures on other architectures for functions
that didn't fail on x86_64 or x86, in which case the usual rule
applies: file a bug (preferably identifying the underlying problem
function, in cases where function A calls function B and a problem
with function B may present in the test results for function A) if not
already in Bugzilla then fix or XFAIL.
Tested x86_64 and x86 and ulps updated accordingly.
* math/libm-test.inc (asinh_test): Use ALL_RM_TEST.
(atan_test): Likewise.
(atanh_test_data): Use NO_TEST_INLINE for two tests.
(atanh_test): Use ALL_RM_TEST.
(atan2_test_data): Likewise.
(cabs_test): Likewise.
(cacosh_test): Likewise.
(carg_test): Likewise.
(casin_test): Likewise.
(casinh_test): Likewise.
(cbrt_test): Likewise.
(csqrt_test): Likewise.
(erf_test): Likewise.
(erfc_test): Likewise.
(pow10_test): Likewise.
(exp2_test): Likewise.
(hypot_test): Likewise.
(j0_test): Likewise.
(j1_test): Likewise.
(lgamma_test): Likewise.
(gamma_test): Likewise.
(sincos_test): Likewise.
(tanh_test): Likewise.
(y0_test): Likewise.
(y1_test): Likewise.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2014-03-21 00:03:38 +00:00
|
|
|
Function: "y1_downward":
|
|
|
|
|
double: 3
|
Regenerate x86_64, x86 ulps from scratch.
The x86_64 and x86 libm-test-ulps files hadn't been regenerated from
scratch for some time, as evidenced by the presence of entries for
*_tonearest functions (those tests duplicated the
default-rounding-mode tests, and such duplicates are no longer run).
The aarch64, alpha, hppa, ia64, m68k, microblaze, powerpc, s390, sh,
sparc, tile files similarly could do with from-scratch regeneration as
evidenced by the presence of such entries. (Truncate the existing
file then run "make regen-ulps" and move the resulting file into
place.)
This patch regenerates the x86_64 and x86 files from scratch. It's
likely some of the reduced / removed ulps will need restoring because
they appear on processors or compiler versions other than the one I
tested on, but in such cases I'd like to first see if I can generate
new tests that show such ulps on the Intel processor I'm testing on,
to reduce the effects from different people using different processors
and compilers to regenerate the ulps.
* sysdeps/i386/fpu/libm-test-ulps: Regenerated.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2015-03-24 23:30:19 +00:00
|
|
|
float: 2
|
Test most libm functions in all rounding modes.
This patch makes libm-test.inc tests of most functions use ALL_RM_TEST
unless there was some reason to defer that change for a particular
function.
I started out planning to defer the change for pow (bug 16315), cexp /
ccos / ccosh / csin / csinh (likely fallout from exp, bug 16284) and
cpow (exact expectations for signs of exact zero results not wanted).
Testing on x86_64 and x86 showed additional failures for acosh, cacos,
catan, catanh, clog, clog10, jn, log, log10, log1p, log2, tgamma, yn,
so making the change for those functions was deferred as well, pending
investigation to show which of these represent distinct bugs (some
such bugs may already be filed) and appropriate fixing / XFAILing.
Failures include wrong signs of zero results, errors slightly above
the 9ulp bound (in such cases it may make sense for functions to set
round-to-nearest internally to reduce error accumulation), large
errors and incorrect overflow/underflow for the rounding mode (with
consequent missing errno settings in some cases). It's possible some
could be issues with test expectations, though I didn't notice any
that were obviously like that (I added NO_TEST_INLINE for cases that
were failing for ildoubl on x86 and where it seemed reasonable for
them to fail for the fast-math inlines).
There may of course be failures on other architectures for functions
that didn't fail on x86_64 or x86, in which case the usual rule
applies: file a bug (preferably identifying the underlying problem
function, in cases where function A calls function B and a problem
with function B may present in the test results for function A) if not
already in Bugzilla then fix or XFAIL.
Tested x86_64 and x86 and ulps updated accordingly.
* math/libm-test.inc (asinh_test): Use ALL_RM_TEST.
(atan_test): Likewise.
(atanh_test_data): Use NO_TEST_INLINE for two tests.
(atanh_test): Use ALL_RM_TEST.
(atan2_test_data): Likewise.
(cabs_test): Likewise.
(cacosh_test): Likewise.
(carg_test): Likewise.
(casin_test): Likewise.
(casinh_test): Likewise.
(cbrt_test): Likewise.
(csqrt_test): Likewise.
(erf_test): Likewise.
(erfc_test): Likewise.
(pow10_test): Likewise.
(exp2_test): Likewise.
(hypot_test): Likewise.
(j0_test): Likewise.
(j1_test): Likewise.
(lgamma_test): Likewise.
(gamma_test): Likewise.
(sincos_test): Likewise.
(tanh_test): Likewise.
(y0_test): Likewise.
(y1_test): Likewise.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2014-03-21 00:03:38 +00:00
|
|
|
idouble: 3
|
Regenerate x86_64, x86 ulps from scratch.
The x86_64 and x86 libm-test-ulps files hadn't been regenerated from
scratch for some time, as evidenced by the presence of entries for
*_tonearest functions (those tests duplicated the
default-rounding-mode tests, and such duplicates are no longer run).
The aarch64, alpha, hppa, ia64, m68k, microblaze, powerpc, s390, sh,
sparc, tile files similarly could do with from-scratch regeneration as
evidenced by the presence of such entries. (Truncate the existing
file then run "make regen-ulps" and move the resulting file into
place.)
This patch regenerates the x86_64 and x86 files from scratch. It's
likely some of the reduced / removed ulps will need restoring because
they appear on processors or compiler versions other than the one I
tested on, but in such cases I'd like to first see if I can generate
new tests that show such ulps on the Intel processor I'm testing on,
to reduce the effects from different people using different processors
and compilers to regenerate the ulps.
* sysdeps/i386/fpu/libm-test-ulps: Regenerated.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2015-03-24 23:30:19 +00:00
|
|
|
ifloat: 2
|
Test most libm functions in all rounding modes.
This patch makes libm-test.inc tests of most functions use ALL_RM_TEST
unless there was some reason to defer that change for a particular
function.
I started out planning to defer the change for pow (bug 16315), cexp /
ccos / ccosh / csin / csinh (likely fallout from exp, bug 16284) and
cpow (exact expectations for signs of exact zero results not wanted).
Testing on x86_64 and x86 showed additional failures for acosh, cacos,
catan, catanh, clog, clog10, jn, log, log10, log1p, log2, tgamma, yn,
so making the change for those functions was deferred as well, pending
investigation to show which of these represent distinct bugs (some
such bugs may already be filed) and appropriate fixing / XFAILing.
Failures include wrong signs of zero results, errors slightly above
the 9ulp bound (in such cases it may make sense for functions to set
round-to-nearest internally to reduce error accumulation), large
errors and incorrect overflow/underflow for the rounding mode (with
consequent missing errno settings in some cases). It's possible some
could be issues with test expectations, though I didn't notice any
that were obviously like that (I added NO_TEST_INLINE for cases that
were failing for ildoubl on x86 and where it seemed reasonable for
them to fail for the fast-math inlines).
There may of course be failures on other architectures for functions
that didn't fail on x86_64 or x86, in which case the usual rule
applies: file a bug (preferably identifying the underlying problem
function, in cases where function A calls function B and a problem
with function B may present in the test results for function A) if not
already in Bugzilla then fix or XFAIL.
Tested x86_64 and x86 and ulps updated accordingly.
* math/libm-test.inc (asinh_test): Use ALL_RM_TEST.
(atan_test): Likewise.
(atanh_test_data): Use NO_TEST_INLINE for two tests.
(atanh_test): Use ALL_RM_TEST.
(atan2_test_data): Likewise.
(cabs_test): Likewise.
(cacosh_test): Likewise.
(carg_test): Likewise.
(casin_test): Likewise.
(casinh_test): Likewise.
(cbrt_test): Likewise.
(csqrt_test): Likewise.
(erf_test): Likewise.
(erfc_test): Likewise.
(pow10_test): Likewise.
(exp2_test): Likewise.
(hypot_test): Likewise.
(j0_test): Likewise.
(j1_test): Likewise.
(lgamma_test): Likewise.
(gamma_test): Likewise.
(sincos_test): Likewise.
(tanh_test): Likewise.
(y0_test): Likewise.
(y1_test): Likewise.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2014-03-21 00:03:38 +00:00
|
|
|
ildouble: 7
|
|
|
|
|
ldouble: 7
|
|
|
|
|
|
|
|
|
|
Function: "y1_towardzero":
|
|
|
|
|
double: 3
|
Regenerate x86_64, x86 ulps from scratch.
The x86_64 and x86 libm-test-ulps files hadn't been regenerated from
scratch for some time, as evidenced by the presence of entries for
*_tonearest functions (those tests duplicated the
default-rounding-mode tests, and such duplicates are no longer run).
The aarch64, alpha, hppa, ia64, m68k, microblaze, powerpc, s390, sh,
sparc, tile files similarly could do with from-scratch regeneration as
evidenced by the presence of such entries. (Truncate the existing
file then run "make regen-ulps" and move the resulting file into
place.)
This patch regenerates the x86_64 and x86 files from scratch. It's
likely some of the reduced / removed ulps will need restoring because
they appear on processors or compiler versions other than the one I
tested on, but in such cases I'd like to first see if I can generate
new tests that show such ulps on the Intel processor I'm testing on,
to reduce the effects from different people using different processors
and compilers to regenerate the ulps.
* sysdeps/i386/fpu/libm-test-ulps: Regenerated.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2015-03-24 23:30:19 +00:00
|
|
|
float: 2
|
Test most libm functions in all rounding modes.
This patch makes libm-test.inc tests of most functions use ALL_RM_TEST
unless there was some reason to defer that change for a particular
function.
I started out planning to defer the change for pow (bug 16315), cexp /
ccos / ccosh / csin / csinh (likely fallout from exp, bug 16284) and
cpow (exact expectations for signs of exact zero results not wanted).
Testing on x86_64 and x86 showed additional failures for acosh, cacos,
catan, catanh, clog, clog10, jn, log, log10, log1p, log2, tgamma, yn,
so making the change for those functions was deferred as well, pending
investigation to show which of these represent distinct bugs (some
such bugs may already be filed) and appropriate fixing / XFAILing.
Failures include wrong signs of zero results, errors slightly above
the 9ulp bound (in such cases it may make sense for functions to set
round-to-nearest internally to reduce error accumulation), large
errors and incorrect overflow/underflow for the rounding mode (with
consequent missing errno settings in some cases). It's possible some
could be issues with test expectations, though I didn't notice any
that were obviously like that (I added NO_TEST_INLINE for cases that
were failing for ildoubl on x86 and where it seemed reasonable for
them to fail for the fast-math inlines).
There may of course be failures on other architectures for functions
that didn't fail on x86_64 or x86, in which case the usual rule
applies: file a bug (preferably identifying the underlying problem
function, in cases where function A calls function B and a problem
with function B may present in the test results for function A) if not
already in Bugzilla then fix or XFAIL.
Tested x86_64 and x86 and ulps updated accordingly.
* math/libm-test.inc (asinh_test): Use ALL_RM_TEST.
(atan_test): Likewise.
(atanh_test_data): Use NO_TEST_INLINE for two tests.
(atanh_test): Use ALL_RM_TEST.
(atan2_test_data): Likewise.
(cabs_test): Likewise.
(cacosh_test): Likewise.
(carg_test): Likewise.
(casin_test): Likewise.
(casinh_test): Likewise.
(cbrt_test): Likewise.
(csqrt_test): Likewise.
(erf_test): Likewise.
(erfc_test): Likewise.
(pow10_test): Likewise.
(exp2_test): Likewise.
(hypot_test): Likewise.
(j0_test): Likewise.
(j1_test): Likewise.
(lgamma_test): Likewise.
(gamma_test): Likewise.
(sincos_test): Likewise.
(tanh_test): Likewise.
(y0_test): Likewise.
(y1_test): Likewise.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2014-03-21 00:03:38 +00:00
|
|
|
idouble: 3
|
Regenerate x86_64, x86 ulps from scratch.
The x86_64 and x86 libm-test-ulps files hadn't been regenerated from
scratch for some time, as evidenced by the presence of entries for
*_tonearest functions (those tests duplicated the
default-rounding-mode tests, and such duplicates are no longer run).
The aarch64, alpha, hppa, ia64, m68k, microblaze, powerpc, s390, sh,
sparc, tile files similarly could do with from-scratch regeneration as
evidenced by the presence of such entries. (Truncate the existing
file then run "make regen-ulps" and move the resulting file into
place.)
This patch regenerates the x86_64 and x86 files from scratch. It's
likely some of the reduced / removed ulps will need restoring because
they appear on processors or compiler versions other than the one I
tested on, but in such cases I'd like to first see if I can generate
new tests that show such ulps on the Intel processor I'm testing on,
to reduce the effects from different people using different processors
and compilers to regenerate the ulps.
* sysdeps/i386/fpu/libm-test-ulps: Regenerated.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2015-03-24 23:30:19 +00:00
|
|
|
ifloat: 2
|
Test most libm functions in all rounding modes.
This patch makes libm-test.inc tests of most functions use ALL_RM_TEST
unless there was some reason to defer that change for a particular
function.
I started out planning to defer the change for pow (bug 16315), cexp /
ccos / ccosh / csin / csinh (likely fallout from exp, bug 16284) and
cpow (exact expectations for signs of exact zero results not wanted).
Testing on x86_64 and x86 showed additional failures for acosh, cacos,
catan, catanh, clog, clog10, jn, log, log10, log1p, log2, tgamma, yn,
so making the change for those functions was deferred as well, pending
investigation to show which of these represent distinct bugs (some
such bugs may already be filed) and appropriate fixing / XFAILing.
Failures include wrong signs of zero results, errors slightly above
the 9ulp bound (in such cases it may make sense for functions to set
round-to-nearest internally to reduce error accumulation), large
errors and incorrect overflow/underflow for the rounding mode (with
consequent missing errno settings in some cases). It's possible some
could be issues with test expectations, though I didn't notice any
that were obviously like that (I added NO_TEST_INLINE for cases that
were failing for ildoubl on x86 and where it seemed reasonable for
them to fail for the fast-math inlines).
There may of course be failures on other architectures for functions
that didn't fail on x86_64 or x86, in which case the usual rule
applies: file a bug (preferably identifying the underlying problem
function, in cases where function A calls function B and a problem
with function B may present in the test results for function A) if not
already in Bugzilla then fix or XFAIL.
Tested x86_64 and x86 and ulps updated accordingly.
* math/libm-test.inc (asinh_test): Use ALL_RM_TEST.
(atan_test): Likewise.
(atanh_test_data): Use NO_TEST_INLINE for two tests.
(atanh_test): Use ALL_RM_TEST.
(atan2_test_data): Likewise.
(cabs_test): Likewise.
(cacosh_test): Likewise.
(carg_test): Likewise.
(casin_test): Likewise.
(casinh_test): Likewise.
(cbrt_test): Likewise.
(csqrt_test): Likewise.
(erf_test): Likewise.
(erfc_test): Likewise.
(pow10_test): Likewise.
(exp2_test): Likewise.
(hypot_test): Likewise.
(j0_test): Likewise.
(j1_test): Likewise.
(lgamma_test): Likewise.
(gamma_test): Likewise.
(sincos_test): Likewise.
(tanh_test): Likewise.
(y0_test): Likewise.
(y1_test): Likewise.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2014-03-21 00:03:38 +00:00
|
|
|
ildouble: 5
|
|
|
|
|
ldouble: 5
|
|
|
|
|
|
|
|
|
|
Function: "y1_upward":
|
|
|
|
|
double: 7
|
Regenerate x86_64, x86 ulps from scratch.
The x86_64 and x86 libm-test-ulps files hadn't been regenerated from
scratch for some time, as evidenced by the presence of entries for
*_tonearest functions (those tests duplicated the
default-rounding-mode tests, and such duplicates are no longer run).
The aarch64, alpha, hppa, ia64, m68k, microblaze, powerpc, s390, sh,
sparc, tile files similarly could do with from-scratch regeneration as
evidenced by the presence of such entries. (Truncate the existing
file then run "make regen-ulps" and move the resulting file into
place.)
This patch regenerates the x86_64 and x86 files from scratch. It's
likely some of the reduced / removed ulps will need restoring because
they appear on processors or compiler versions other than the one I
tested on, but in such cases I'd like to first see if I can generate
new tests that show such ulps on the Intel processor I'm testing on,
to reduce the effects from different people using different processors
and compilers to regenerate the ulps.
* sysdeps/i386/fpu/libm-test-ulps: Regenerated.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2015-03-24 23:30:19 +00:00
|
|
|
float: 2
|
Test most libm functions in all rounding modes.
This patch makes libm-test.inc tests of most functions use ALL_RM_TEST
unless there was some reason to defer that change for a particular
function.
I started out planning to defer the change for pow (bug 16315), cexp /
ccos / ccosh / csin / csinh (likely fallout from exp, bug 16284) and
cpow (exact expectations for signs of exact zero results not wanted).
Testing on x86_64 and x86 showed additional failures for acosh, cacos,
catan, catanh, clog, clog10, jn, log, log10, log1p, log2, tgamma, yn,
so making the change for those functions was deferred as well, pending
investigation to show which of these represent distinct bugs (some
such bugs may already be filed) and appropriate fixing / XFAILing.
Failures include wrong signs of zero results, errors slightly above
the 9ulp bound (in such cases it may make sense for functions to set
round-to-nearest internally to reduce error accumulation), large
errors and incorrect overflow/underflow for the rounding mode (with
consequent missing errno settings in some cases). It's possible some
could be issues with test expectations, though I didn't notice any
that were obviously like that (I added NO_TEST_INLINE for cases that
were failing for ildoubl on x86 and where it seemed reasonable for
them to fail for the fast-math inlines).
There may of course be failures on other architectures for functions
that didn't fail on x86_64 or x86, in which case the usual rule
applies: file a bug (preferably identifying the underlying problem
function, in cases where function A calls function B and a problem
with function B may present in the test results for function A) if not
already in Bugzilla then fix or XFAIL.
Tested x86_64 and x86 and ulps updated accordingly.
* math/libm-test.inc (asinh_test): Use ALL_RM_TEST.
(atan_test): Likewise.
(atanh_test_data): Use NO_TEST_INLINE for two tests.
(atanh_test): Use ALL_RM_TEST.
(atan2_test_data): Likewise.
(cabs_test): Likewise.
(cacosh_test): Likewise.
(carg_test): Likewise.
(casin_test): Likewise.
(casinh_test): Likewise.
(cbrt_test): Likewise.
(csqrt_test): Likewise.
(erf_test): Likewise.
(erfc_test): Likewise.
(pow10_test): Likewise.
(exp2_test): Likewise.
(hypot_test): Likewise.
(j0_test): Likewise.
(j1_test): Likewise.
(lgamma_test): Likewise.
(gamma_test): Likewise.
(sincos_test): Likewise.
(tanh_test): Likewise.
(y0_test): Likewise.
(y1_test): Likewise.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2014-03-21 00:03:38 +00:00
|
|
|
idouble: 7
|
Regenerate x86_64, x86 ulps from scratch.
The x86_64 and x86 libm-test-ulps files hadn't been regenerated from
scratch for some time, as evidenced by the presence of entries for
*_tonearest functions (those tests duplicated the
default-rounding-mode tests, and such duplicates are no longer run).
The aarch64, alpha, hppa, ia64, m68k, microblaze, powerpc, s390, sh,
sparc, tile files similarly could do with from-scratch regeneration as
evidenced by the presence of such entries. (Truncate the existing
file then run "make regen-ulps" and move the resulting file into
place.)
This patch regenerates the x86_64 and x86 files from scratch. It's
likely some of the reduced / removed ulps will need restoring because
they appear on processors or compiler versions other than the one I
tested on, but in such cases I'd like to first see if I can generate
new tests that show such ulps on the Intel processor I'm testing on,
to reduce the effects from different people using different processors
and compilers to regenerate the ulps.
* sysdeps/i386/fpu/libm-test-ulps: Regenerated.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2015-03-24 23:30:19 +00:00
|
|
|
ifloat: 2
|
Test most libm functions in all rounding modes.
This patch makes libm-test.inc tests of most functions use ALL_RM_TEST
unless there was some reason to defer that change for a particular
function.
I started out planning to defer the change for pow (bug 16315), cexp /
ccos / ccosh / csin / csinh (likely fallout from exp, bug 16284) and
cpow (exact expectations for signs of exact zero results not wanted).
Testing on x86_64 and x86 showed additional failures for acosh, cacos,
catan, catanh, clog, clog10, jn, log, log10, log1p, log2, tgamma, yn,
so making the change for those functions was deferred as well, pending
investigation to show which of these represent distinct bugs (some
such bugs may already be filed) and appropriate fixing / XFAILing.
Failures include wrong signs of zero results, errors slightly above
the 9ulp bound (in such cases it may make sense for functions to set
round-to-nearest internally to reduce error accumulation), large
errors and incorrect overflow/underflow for the rounding mode (with
consequent missing errno settings in some cases). It's possible some
could be issues with test expectations, though I didn't notice any
that were obviously like that (I added NO_TEST_INLINE for cases that
were failing for ildoubl on x86 and where it seemed reasonable for
them to fail for the fast-math inlines).
There may of course be failures on other architectures for functions
that didn't fail on x86_64 or x86, in which case the usual rule
applies: file a bug (preferably identifying the underlying problem
function, in cases where function A calls function B and a problem
with function B may present in the test results for function A) if not
already in Bugzilla then fix or XFAIL.
Tested x86_64 and x86 and ulps updated accordingly.
* math/libm-test.inc (asinh_test): Use ALL_RM_TEST.
(atan_test): Likewise.
(atanh_test_data): Use NO_TEST_INLINE for two tests.
(atanh_test): Use ALL_RM_TEST.
(atan2_test_data): Likewise.
(cabs_test): Likewise.
(cacosh_test): Likewise.
(carg_test): Likewise.
(casin_test): Likewise.
(casinh_test): Likewise.
(cbrt_test): Likewise.
(csqrt_test): Likewise.
(erf_test): Likewise.
(erfc_test): Likewise.
(pow10_test): Likewise.
(exp2_test): Likewise.
(hypot_test): Likewise.
(j0_test): Likewise.
(j1_test): Likewise.
(lgamma_test): Likewise.
(gamma_test): Likewise.
(sincos_test): Likewise.
(tanh_test): Likewise.
(y0_test): Likewise.
(y1_test): Likewise.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
2014-03-21 00:03:38 +00:00
|
|
|
ildouble: 7
|
|
|
|
|
ldouble: 7
|
|
|
|
|
|
2001-09-19 10:37:31 +00:00
|
|
|
Function: "yn":
|
|
|
|
|
double: 3
|
2013-09-02 14:51:24 +00:00
|
|
|
float: 3
|
2001-09-19 10:37:31 +00:00
|
|
|
idouble: 3
|
2013-09-02 14:51:24 +00:00
|
|
|
ifloat: 3
|
2002-08-31 08:22:35 +00:00
|
|
|
ildouble: 4
|
|
|
|
|
ldouble: 4
|
2001-09-19 10:37:31 +00:00
|
|
|
|
2014-06-27 14:52:13 +00:00
|
|
|
Function: "yn_downward":
|
|
|
|
|
double: 3
|
|
|
|
|
float: 4
|
|
|
|
|
idouble: 3
|
|
|
|
|
ifloat: 4
|
|
|
|
|
ildouble: 5
|
|
|
|
|
ldouble: 5
|
|
|
|
|
|
|
|
|
|
Function: "yn_towardzero":
|
|
|
|
|
double: 3
|
|
|
|
|
float: 3
|
|
|
|
|
idouble: 3
|
|
|
|
|
ifloat: 3
|
|
|
|
|
ildouble: 5
|
|
|
|
|
ldouble: 5
|
|
|
|
|
|
|
|
|
|
Function: "yn_upward":
|
|
|
|
|
double: 4
|
|
|
|
|
float: 5
|
|
|
|
|
idouble: 4
|
|
|
|
|
ifloat: 5
|
|
|
|
|
ildouble: 4
|
|
|
|
|
ldouble: 4
|
|
|
|
|
|
2001-09-19 10:37:31 +00:00
|
|
|
# end of automatic generation
|
