Commit Graph

33 Commits

Author SHA1 Message Date
Ganesh Ajjanagadde
a0a47a09b0 lavu/libm: add isfinite fallback
Reviewed-by: Ronald S. Bultje <rsbultje@gmail.com>
Signed-off-by: Ganesh Ajjanagadde <gajjanagadde@gmail.com>
2016-01-13 20:00:19 -05:00
Ganesh Ajjanagadde
73616b1f21 lavu/libm,configure: remove exp10, exp10f detection
Subsequent commit introduces ff_exp10 instead.

Signed-off-by: Ganesh Ajjanagadde <gajjanagadde@gmail.com>
2015-12-25 10:12:49 -08:00
Ganesh Ajjanagadde
5630ed5be6 lavu/libm: misc, minor changes
Addition of comments marking the end of ifdef blocks, correction of an
incorrect (at double precision) M_LN2, removal of an unnecessary undef.

Signed-off-by: Ganesh Ajjanagadde <gajjanagadde@gmail.com>
2015-12-25 09:25:20 -08:00
Ganesh Ajjanagadde
e29db08cf7 lavu/libm: add exp10 support
exp10 is a function available in GNU libm. Looks like no other common
libm has it. This adds support for it to FFmpeg.

There are essentially 2 ways of handling the fallback:
1. Using pow(10, x)
2. Using exp2(M_LOG2_10 * x).

First one represents a Pareto improvement, with no speed or accuracy
regression anywhere, but speed improvement limited to GNU libm.

Second one represents a slight accuracy loss (relative error ~ 1e-13)
for non GNU libm. Speedup of > 2x is obtained on non GNU libm platforms,
~30% on GNU libm. These are "average case numbers", another benefit is
the lack of triggering of the well-known terrible worst case paths
through pow.

Based on reviews, second one chosen. Comment added accordingly.

Reviewed-by: Hendrik Leppkes <h.leppkes@gmail.com>
Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
Reviewed-by: Ronald S. Bultje <rsbultje@gmail.com>
Signed-off-by: Ganesh Ajjanagadde <gajjanagadde@gmail.com>
2015-12-23 09:22:59 -08:00
Ganesh Ajjanagadde
dd68cde28a lavu/libm: add erf hack and make dynaudnorm available everywhere
Source code is from Boost:
http://www.boost.org/doc/libs/1_46_1/boost/math/special_functions/erf.hpp
with appropriate modifications for FFmpeg.

Tested on interval -6 to 6 (beyond which it saturates), +/-NAN, +/-INFINITY
under -fsanitize=undefined on clang to test for possible undefined behavior.

This function turns out to actually be essentially as accurate and faster than the
libm (GNU/BSD's/Mac OS X), and I can think of 3 reasons why upstream
does not use this:
1. They are not aware of it.
2. They are concerned about licensing - this applies especially to GNU
libm.
3. They do not know and/or appreciate the benefits of rational
approximations over polynomial approximations. Boost uses them to great
effect, see e.g swr/resample for bessel derived from them, which is also
similarly superior to libm variants.

First, performance.
sample benchmark (clang -O3, Haswell, GNU/Linux):

3e8 values evenly spaced from 0 to 6
time (libm):
./test  13.39s user 0.00s system 100% cpu 13.376 total
time (boost based):
./test  9.20s user 0.00s system 100% cpu 9.190 total

Second, accuracy.
1e8 eval pts from 0 to 6
maxdiff (absolute): 2.2204460492503131e-16
occuring at point where libm erf is correctly rounded, this is not.

Illustration of superior rounding of this function:
arg   : 0.83999999999999997
erf   : 0.76514271145499457
boost : 0.76514271145499446
real  : 0.76514271145499446

i.e libm is actually incorrectly rounded. Note that this is clear from:
https://github.com/JuliaLang/openlibm/blob/master/src/s_erf.c (the Sun
implementation used by both BSD and GNU libm's), where only 1 ulp is
guaranteed.

Reasons it is not easy/worthwhile to create a "correctly rounded"
variant of this function (i.e 0.5ulp):
1. Upstream libm's don't do it anyway, so we can't guarantee this unless
we force this implementation on all platforms. This is not easy, as the
linker would complain unless measures are taken.
2. Nothing in FFmpeg cares or can care about such things, due to the
above and FFmpeg's nature.
3. Creating a correctly rounded function will in practice need some use of long
double/fma. long double, although C89/C90, unfortunately has problems on
ppc. This needs fixing of toolchain flags/configure. In any case this
will be slower for miniscule gain.

Reviewed-by: James Almer <jamrial@gmail.com>
Signed-off-by: Ganesh Ajjanagadde <gajjanagadde@gmail.com>
2015-12-21 09:03:11 -08:00
Ganesh Ajjanagadde
062e3e2382 lavu/libm: add copysign hack
For systems with broken libms.
Tested with NAN, -NAN, INFINITY, -INFINITY, +/-x for regular double x and
combinations of these.

Old versions of MSVC need some UINT64_C hackery.

Reviewed-by: Ronald S. Bultje <rsbultje@gmail.com>
Reviewed-by: Hendrik Leppkes <h.leppkes@gmail.com>
Signed-off-by: Ganesh Ajjanagadde <gajjanagadde@gmail.com>
2015-12-19 14:05:49 -08:00
Ganesh Ajjanagadde
29af74e4e3 avutil/libm: fix isnan compatibility hack
Commit 14ea4151d7 had a bug in that the
conversion of the uint64_t result to an int (the return signature) would
lead to implementation defined behavior, and in this case simply
returned 0 for NAN. A fix via AND'ing the result with 1 does the trick,
simply by ensuring a 0 or 1 return value.

Patch tested with FATE on x86-64, GNU/Linux by forcing the compatibility
code via an ifdef hack suggested by Michael.

Signed-off-by: Ganesh Ajjanagadde <gajjanagadde@gmail.com>
2015-11-24 21:33:13 -05:00
Ganesh Ajjanagadde
275aca8fba configure+libm.h: add hypot emulation
It is known that the naive sqrt(x*x + y*y) approach for computing the
hypotenuse suffers from overflow and accuracy issues, see e.g
http://www.johndcook.com/blog/2010/06/02/whats-so-hard-about-finding-a-hypotenuse/.
This adds hypot support to FFmpeg, a C99 function.

On platforms without hypot, this patch does a reaonable workaround, that
although not as accurate as GNU libm, is readable and does not suffer
from the overflow issue. Improvements can be made separately.

Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: Ganesh Ajjanagadde <gajjanagadde@gmail.com>
2015-11-21 08:51:49 -05:00
Ganesh Ajjanagadde
14ea4151d7 avutil/libm: correct isnan, isinf compat hacks
isnan and isinf are actually macros as per the standard. In particular,
the existing implementation has incorrect signature. Furthermore, this
results in undefined behavior for e.g double values outside float range
as per the standard.

This patch corrects the undefined behavior for all usage within FFmpeg.

Note that long double is not handled as it is not used in FFmpeg.
Furthermore, even if at some point long double gets used, it is likely
not needed to modify the macro in practice for usage in FFmpeg. See
below for analysis.

Getting long double to work strictly per the spec is significantly harder
since a long double may be an IEEE 128 bit quad (very rare), 80 bit
extended precision value (on GCC/Clang), or simply double (on recent Microsoft).
On the other hand, any potential future usage of long double is likely
for precision (when a platform offers extra precision) and not for range, since
the range anyway varies and is not as portable as IEEE 754 single/double
precision. In such cases, the implicit cast to a double is well defined
and isinf and isnan should work as intended.

Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: Ganesh Ajjanagadde <gajjanagadde@gmail.com>
2015-11-21 08:51:49 -05:00
Carl Eugen Hoyos
4436a8f44d Remove fminf() emulation.
The emulation is unused and causes compilation trouble on systems
where fminf() is defined in <math.h> but missing from libm.
This should fix compilation on Debian powerpcspe.
2014-11-08 11:31:11 +01:00
Michael Niedermayer
e374e77292 avutil/libm: fix fminf() emulation build failure due to undefined FFMIN
Found-by: James Almer <jamrial@gmail.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2014-06-06 21:00:11 +02:00
Michael Niedermayer
afce834843 avutil/libm: Replace macro based fminf() by function
This avoids issues when the FFMIN parameter evaluation has side effects

Reviewed-by: Clément Bœsch <u@pkh.me>
Reviewed-by: Derek Buitenhuis <derek.buitenhuis@gmail.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2014-06-06 17:18:14 +02:00
Michael Niedermayer
3133e7fd44 avutil/libm: use FFMIN instead of fmin()
MSVC apparently doesnt support fmin() either

Suggested/Found-by: ubitux, Daemon404, nevcairiel
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2014-06-06 16:02:53 +02:00
Michael Niedermayer
2c1bf3fc96 avutil/libm: fix fminf typo
Found-by: ubitux
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2014-06-06 15:46:03 +02:00
Michael Niedermayer
97508af274 build: add fminf() emulation
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2014-06-06 14:54:57 +02:00
James Almer
e65d8509f0 libm: Add fallback definition for cbrt() using pow()
The function is known to be missing in at least one target (MSVC).

Signed-off-by: James Almer <jamrial@gmail.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2013-01-22 16:02:31 +01:00
Michael Niedermayer
2dbc93455c Merge commit '80521c1997a23e148edf89e11b939ab8646297ca'
* commit '80521c1997a23e148edf89e11b939ab8646297ca':
  build: allow targets to specify extra objects to link with executables
  swscale: avoid pointless use of compound literals
  libm: add fallbacks for various single-precision functions
  network: use getservbyport() only if available
  network: add fallbacks for INADDR_LOOPBACK and INET_ADDRSTRLEN
  Include sys/time.h before sys/resource.h

Conflicts:
	Makefile
	configure
	libavutil/libm.h

Merged-by: Michael Niedermayer <michaelni@gmx.at>
2012-10-24 12:53:26 +02:00
Mans Rullgard
fab0a8b2c6 libm: add fallbacks for various single-precision functions
Signed-off-by: Mans Rullgard <mans@mansr.com>
2012-10-23 12:00:21 +01:00
Michael Niedermayer
9ca27df52f Merge remote-tracking branch 'qatar/master'
* qatar/master:
  configure: Check for the math function rint
  TechSmith Screen Codec 2 decoder
  rtsp: Add listen mode
  rtsp: Make rtsp_open_transport_ctx() non-static
  rtsp: Move rtsp_read_close
  rtsp: Parse the mode=receive/record parameter in transport lines

Conflicts:
	Changelog
	libavcodec/avcodec.h
	libavcodec/version.h
	libavformat/version.h

Merged-by: Michael Niedermayer <michaelni@gmx.at>
2012-07-11 23:57:11 +02:00
Ronald S. Bultje
183b1c2268 configure: Check for the math function rint
Add a fallback implementation if it doesn't exist.

Signed-off-by: Martin Storsjö <martin@martin.st>
2012-07-11 10:40:11 +03:00
Michael Niedermayer
87df986dcf Merge remote-tracking branch 'qatar/master'
* qatar/master:
  mss1: validate number of changeable palette entries
  mss1: report palette changed when some additional colours were decoded
  x86: fft: replace call to memcpy by a loop
  udp: Support IGMPv3 source specific multicast and source blocking
  dxva2: include dxva.h if found
  libm: Provide fallback definitions for isnan() and isinf()
  tcp: Pass NULL as hostname to getaddrinfo if the string is empty
  tcp: Set AI_PASSIVE when the socket will be used for listening

Conflicts:
	configure
	libavcodec/mss1.c
	libavformat/udp.c

Merged-by: Michael Niedermayer <michaelni@gmx.at>
2012-06-28 01:08:52 +02:00
Martin Storsjö
46df708b45 libm: Provide fallback definitions for isnan() and isinf()
Signed-off-by: Martin Storsjö <martin@martin.st>
2012-06-27 14:04:02 +03:00
Michael Niedermayer
b536e2facf Merge remote-tracking branch 'qatar/master'
* qatar/master:
  avconv: add an assert to silence an uninitialized variable warning.
  avconv: shut up an uninitialized variable warning.
  avfiltergraph: shut up uninitialized variable warning.
  af_join: initialize a variable to shut up gcc warning.
  amix: fix format specifier for AVFilterLink.sample_rate.
  lavfi: make filters less verbose.
  mpc8: read APE tags.
  lavr: x86: fix ff_conv_fltp_to_flt_6ch function prototypes
  libm: provide fallback definition for cbrtf() using powf()
  network: Don't redefine error codes if they already exist in errno.h
  configure: Check for sys/time.h
  network: Include unistd.h from network.h
  avconv: don't include vsrc_buffer.h, which doesn't exist anymore
  lavfi: reorder AVFilterLink fields.
  lavfi: reorder AVFilterContext fields.
  lavfi: reorder AVFilter fields.
  lavfi: reorder AVFilterBufferRef fields.
  lavfi: reorder AVFilterBuffer fields.

Conflicts:
	configure
	libavcodec/v210dec.h
	libavfilter/asrc_anullsrc.c
	libavfilter/avfilter.h
	libavfilter/buffersrc.c
	libavfilter/src_movie.c
	libavfilter/vf_aspect.c
	libavfilter/vf_crop.c
	libavfilter/vf_drawbox.c
	libavfilter/vf_fade.c
	libavfilter/vf_overlay.c
	libavfilter/vf_pad.c
	libavfilter/vf_scale.c
	libavfilter/vsrc_color.c
	libavformat/network.h
	libavutil/libm.h

Merged-by: Michael Niedermayer <michaelni@gmx.at>
2012-06-27 01:58:09 +02:00
Mans Rullgard
153335625c libm: provide fallback definition for cbrtf() using powf()
This adds a fallback for cbrtf() using powf(x, 1/3).  Since
powf() with a non-integer exponent requires a non-negative
base, special handling of negative inputs is needed.

Signed-off-by: Mans Rullgard <mans@mansr.com>
2012-06-26 15:56:38 +01:00
Nedeljko Babic
3827a86eac Optimization of AMR NB and WB decoders for MIPS
AMR NB and WB decoders are optimized for MIPS architecture.
Appropriate Makefiles are changed accordingly.

Cnfigure script is changed in order to support optimizations.
 Optimizations are enabled by default when compiling is done for
  mips architecture.
 Appropriate cflags are automatically set.
 Support for several mips CPUs is added in configure script.

New ffmpeg options are added for disabling optimizations.

The FFMPEG option --disable-mipsfpu disables MIPS floating point
 optimizations.
The FFMPEG option --disable-mips32r2 disables MIPS32R2
 optimizations.
The FFMPEG option --disable-mipsdspr1 disables MIPS DSP ASE R1
 optimizations.
The FFMPEG option --disable-mipsdspr2 disables MIPS DSP ASE R2
 optimizations.

Signed-off-by: Nedeljko Babic <nbabic@mips.com>
Reviewed-by: Vitor Sessak <vitor1001@gmail.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2012-06-11 21:12:39 +02:00
Mans Rullgard
ab2539708a Check for cbrtf() and provide fallback
(cherry picked from commit bfb37d7db1c6394099f6b177e10d6e769e678f6a)

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2011-10-22 19:49:27 +02:00
Michael Niedermayer
8772156be0 Merge remote branch 'qatar/master'
* qatar/master:
  APIChanges: document git revision for CODEC_CAP_SLICE_THREADS addition.
  Introduce slice threads flag.
  FATE: allow forcing thread-type when doing threaded fate runs.
  Use av_log_ask_for_sample() where appropriate.
  error: sort, pack, and align error code and string definitions
  The stabilization period after version bumps should be one month, not one week.
  applehttp: Expose the stream bitrate via metadata
  doc: Add some initial docs on the applehttp demuxer
  Provide a fallback version of the libm function trunc
  libavdevice: Define _XOPEN_SOURCE for usleep
  lavc: provide deprecated avcodec_thread_init until next major version
  lavc: provide the opt.h header until the next bump
  error: change AVERROR_EOF value
  error: remove AVERROR_NUMEXPECTED
  error: add error code AVERROR_OPTION_NOT_FOUND, and use it in opt.c

Conflicts:
	libavcodec/h264.c
	libavutil/error.c
	libavutil/error.h

Merged-by: Michael Niedermayer <michaelni@gmx.at>
2011-04-22 03:54:30 +02:00
Martin Storsjö
23d3931a6a Provide a fallback version of the libm function trunc
This fixes compilation on DOS.

Signed-off-by: Martin Storsjö <martin@martin.st>
2011-04-21 14:13:09 +03:00
Mans Rullgard
2912e87a6c Replace FFmpeg with Libav in licence headers
Signed-off-by: Mans Rullgard <mans@mansr.com>
2011-03-19 13:33:20 +00:00
Måns Rullgård
07876e25c8 Workaround for missing llrintf()
Originally committed as revision 22954 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-04-23 16:28:10 +00:00
Diego Biurrun
ba87f0801d Remove explicit filename from Doxygen @file commands.
Passing an explicit filename to this command is only necessary if the
documentation in the @file block refers to a file different from the
one the block resides in.

Originally committed as revision 22921 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-04-20 14:45:34 +00:00
Måns Rullgård
297bfb2fe6 libm.h needs attributes.h
Originally committed as revision 22405 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-03-09 16:17:59 +00:00
Måns Rullgård
335ee1aadd Move libm replacements to new header libm.h
ffmpeg.c uses lrintf(), which is missing on some systems.  Previously
it picked up the replacement via libavutil/internal.h due to
HAVE_AV_CONFIG_H being erroneously defined.

Moving these replacements to a separate header enables ffmpeg.c to
use them without being exposed to internal interfaces.

This use of a non-public header is justified by the header in question
not being part of the internal interface either.  It should rather be
considered as part of the build system, which is shared between the
libraries and the applications.

This header cannot be installed since the tested conditions depend on
the compiler.

Originally committed as revision 22399 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-03-09 15:10:23 +00:00