diff options
author | Nils Wallménius <nils@rockbox.org> | 2013-05-20 22:25:57 +0200 |
---|---|---|
committer | Nils Wallménius <nils@rockbox.org> | 2013-08-31 08:30:51 +0200 |
commit | 580b307fd791c0997a8831bc800bba87797bfb7e (patch) | |
tree | 807846056f06fd944a750ce41217a877910ebd59 /lib/rbcodec/codecs/libopus/celt/vq.h | |
parent | 74761b70acd96cecc0d35450dd56a98ad9ee7d3d (diff) | |
download | rockbox-580b307fd791c0997a8831bc800bba87797bfb7e.tar.gz rockbox-580b307fd791c0997a8831bc800bba87797bfb7e.zip |
Sync opus codec to upstream git
Sync opus codec to upstream commit
02fed471a4568852d6618e041c4f2af0d7730ee2 (August 30 2013)
This brings in a lot of optimizations but also makes the diff
between our codec and the upstream much smaller as most of our
optimizations have been upstreamed or supeceded.
Speedups across the board for CELT mode files:
64kbps 128kbps
H300 9.82MHz 15.48MHz
c200 4.86MHz 9.63MHz
fuze v1 10.32MHz 15.92MHz
For the silk mode test file (16kbps) arm targets get a speedup
of about 2MHz while the H300 is 7.8MHz slower, likely because it's
now using the pseudostack more rather than the real stack which
is in iram. Patches to get around that are upcomming.
Change-Id: Ifecf963e461c51ac42e09dac1e91bc4bc3b12fa3
Diffstat (limited to 'lib/rbcodec/codecs/libopus/celt/vq.h')
-rw-r--r-- | lib/rbcodec/codecs/libopus/celt/vq.h | 7 |
1 files changed, 2 insertions, 5 deletions
diff --git a/lib/rbcodec/codecs/libopus/celt/vq.h b/lib/rbcodec/codecs/libopus/celt/vq.h index 1ceeeeb268..ffdc69cdc4 100644 --- a/lib/rbcodec/codecs/libopus/celt/vq.h +++ b/lib/rbcodec/codecs/libopus/celt/vq.h | |||
@@ -40,11 +40,9 @@ | |||
40 | /** Algebraic pulse-vector quantiser. The signal x is replaced by the sum of | 40 | /** Algebraic pulse-vector quantiser. The signal x is replaced by the sum of |
41 | * the pitch and a combination of pulses such that its norm is still equal | 41 | * the pitch and a combination of pulses such that its norm is still equal |
42 | * to 1. This is the function that will typically require the most CPU. | 42 | * to 1. This is the function that will typically require the most CPU. |
43 | * @param x Residual signal to quantise/encode (returns quantised version) | 43 | * @param X Residual signal to quantise/encode (returns quantised version) |
44 | * @param W Perceptual weight to use when optimising (currently unused) | ||
45 | * @param N Number of samples to encode | 44 | * @param N Number of samples to encode |
46 | * @param K Number of pulses to use | 45 | * @param K Number of pulses to use |
47 | * @param p Pitch vector (it is assumed that p+x is a unit vector) | ||
48 | * @param enc Entropy encoder state | 46 | * @param enc Entropy encoder state |
49 | * @ret A mask indicating which blocks in the band received pulses | 47 | * @ret A mask indicating which blocks in the band received pulses |
50 | */ | 48 | */ |
@@ -56,10 +54,9 @@ unsigned alg_quant(celt_norm *X, int N, int K, int spread, int B, | |||
56 | ); | 54 | ); |
57 | 55 | ||
58 | /** Algebraic pulse decoder | 56 | /** Algebraic pulse decoder |
59 | * @param x Decoded normalised spectrum (returned) | 57 | * @param X Decoded normalised spectrum (returned) |
60 | * @param N Number of samples to decode | 58 | * @param N Number of samples to decode |
61 | * @param K Number of pulses to use | 59 | * @param K Number of pulses to use |
62 | * @param p Pitch vector (automatically added to x) | ||
63 | * @param dec Entropy decoder state | 60 | * @param dec Entropy decoder state |
64 | * @ret A mask indicating which blocks in the band received pulses | 61 | * @ret A mask indicating which blocks in the band received pulses |
65 | */ | 62 | */ |