Viswanath Puttagunta authored 9 years ago and Jean-Marc Valin committed 9 years ago

Optimize opus encode (float only) usecase using ARM NE10
library. Mainly effects opus_fft and ctl_mdct_forward
and related functions.

This optimization can be used for ARM CPUs that have NEON
VFP unit. This patch only enables optimizations for ARMv7.

Official ARM NE10 library page available at
http://projectne10.github.io/Ne10/



To enable this optimization, use
--enable-intrinsics --with-NE10=<install_prefix>
or
--enable-intrinsics --with-NE10-libraries=<NE10_lib_dir> --with-NE10-includes=<NE10_includes_dir>

Compile time checks made during configure process to make sure
optimization option available only when compiler supports NEON
instrinsics.

Runtime checks made to make sure optimized functions only called
on appropriate hardware.

Signed-off-by: Timothy B. Terriberry <tterribe@xiph.org>

This should suppress our current issues with unused parameters,
 unused variables, and set-but-not-used variables.

There is also no trailing whitespace.

xiangmingzhu authored 10 years ago and Jean-Marc Valin committed 10 years ago

1. Only for fixed point on x86 platform (32bit and 64bit, uses SIMD
   intrinsics up to SSE4.2)
2. Use "configure --enable-fixed-point --enable-intrinsics" to enable
   optimization, default is disabled.
3. Official test cases are verified and passed.

Signed-off-by: Timothy B. Terriberry <tterribe@xiph.org>

Thanks to Kevin O'Connor for reporting this.

This is a bit faster at -O2 because memcpy()/memmove()/memset() are
vectorized. The code is also cleaner.

NaN detection should now be able to catch values that would create NaNs
further down.

This also splits the rate adjustment between SILK and CELT (previous
code only changed SILK)

Still needs more tuning

Timothy B. Terriberry authored 11 years ago and Jean-Marc Valin committed 11 years ago

Optimizing celt_pitch_xcorr()/xcorr_kernel() which also speeds up
FIRs, IIRs and auto-correlations

Signed-off-by: Jean-Marc Valin <jmvalin@jmvalin.ca>

Works by turning off pitch and energy prediction in CELT, while
setting first_frame_after_reset in SILK to disable pitch and LSF interpolation
and reduce LPC gain.

The problem was that forcing CELT-mode for low bitrate CBR was done too late,
after the encoder had decided to use SILK. This was causing redundancy
to be allocated because the encoder didn't realize it was going to keep
using CELT.

Padding is now handled by the repacketizer.

This fixes an actual error in the downmix (using the float version even
for the int API), as well as a bunch of conversion warnings.

Stereo mode, stereo width, min bandwidth, VBR damping

The LFE encoder would first decide to switch to SILK, and allocated
redundancy bytes. Then, it would force CELT because of LFE, but keep
the redundancy allocation, and end up with too few bytes, that would
end up being padded to the desired size for CBR.

Even when using SILK/hybrid.

The previous version simply produced zeros for fixed-point.

Default complexity is now 9.

Should fix delayed decision on surround, but not well tested.

This should make the scaling the same for fixed and float. It changes the float
scaling too by normalizing by the number of channels, which matters for
bandwidth detection.

Code is still float, but at least tonality esitmation seems to work.
Speech/music analysis is still disabled.

I checked that it's not *completely* busted, but it probably needs more testing.