The bits we don't use won't be wasted, so it's less important to
get exactly the optimal number of bits below the cap.

The gain*2 when overshooting was too aggressive and the undershoot
case wasn't aggressive enough. This now seems to work reasonably well.

Not sure if it was the original intent, but we now reduce the
loss percentage threshold for pitch scaling as 1/nFramesPerPacket
since only the first frame will have pitch scaling anyway.
As a side effect, this brings back the original behavior of
disabling pitch scaling for 0% loss.

Making LTP scaling depend on the bitrate and whether FEC is on.
The thresholds for scaling 1 and 2 are now independent.

The function copies NLSFs from the stack to the state which for
order 10 means we were copying uninitialized values. That in turn
breaks check-asm when comparing the state under valgrind.

Reviewed by Timothy B. Terriberry.

Thanks Tim

Thanks to Ray Essick

Broken by earlier ARM optimizations.

Gustaf Ullberg authored 6 years ago and Jean-Marc Valin committed 6 years ago

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

Gustaf Ullberg authored 7 years ago and Jean-Marc Valin committed 7 years ago

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

Distinguish source files for the SSE 4.1 instruction
set extension consistently by their filename. This makes
it easier to check the correct flags are being set at
build time.

Signed-off-by: Jonathan Lennox <jonathan@vidyo.com>

Linfeng Zhang authored 8 years ago and Jean-Marc Valin committed 7 years ago

The optimization is bit exact with C function.

This optimization speeds up fixed-point SILK encoder on NEON about
5% to 8%. (Tested on Acer Chromebook, ARMv7 Processor rev 3 (v7l).)

Change-Id: I582f6f3585b7946149e16a2ad3084ebc0ae79a4f

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

- celt/modes.c:430:14: warning: cast from 'const unsigned char *' to
  'opus_int16 *' increases required alignment from 1 to 2 [-Wcast-align]
- 'C[0][1]' may be used uninitialized [-Wmaybe-uninitialized]
- Unused variable/parameter
- Value stored is never read
- MSVC warnings about "possible loss of data" due to type conversions
- MSVC warning C4146: unary minus operator applied to unsigned type
- silk/NLSF_del_dec_quant.c:137:20: warning: array subscript is above
  array bounds [-Warray-bounds] (gcc -O3 false positive)
- src/mlp_train.h:39:20: warning: function declaration isn't a prototype
  [-Wstrict-prototypes]
- Remove SMALL_FOOTPRINT code from SSE 4.1 FIR implementation, matching
  the C implementation.

The clang -Wcast-align warnings with SSE intrinsics are a known
clang issue: https://llvm.org/bugs/show_bug.cgi?id=20670

Linfeng Zhang authored 8 years ago and Jean-Marc Valin committed 8 years ago

The optimization is bit exact with C function.

Change-Id: Ib3bdc26a5a4ebe02e7f24be85104e8e9a2a9a738

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

Reported by Mark Warner.

For large values of maxabs_Q20, silk_MUL( maxabs_Q20, ind + 1 ) could
overflow.

Koen Vos authored 8 years ago and Jean-Marc Valin committed 8 years ago

more conservative scaling of LTP corrs; remove assert that checks for negative residual energy; discard CB entries leading to such negative energies

Koen Vos authored 9 years ago and Jean-Marc Valin committed 8 years ago

The NSQ SSE optimizations are disabled for now because they need to be updated