From 9b7ec42403073ee887efc531c153e6b1b6c15bab Mon Sep 17 00:00:00 2001 From: Nils Wallménius Date: Sun, 19 Jan 2014 16:31:59 +0100 Subject: Sync to upstream libopus Sync to commit bb4b6885a139644cf3ac14e7deda9f633ec2d93c This brings in a bunch of optimizations to decode speed and memory usage. Allocations are switched from using the pseudostack to using the real stack. Enabled hacks to reduce stack usage. This should fix crashes on sansa clip, although some files will not play due to failing allocations in the codec buffer. Speeds up decoding of the following test files: H300 (cf) C200 (arm7tdmi) ipod classic (arm9e) 16 kbps (silk) 14.28 MHz 4.00 MHz 2.61 MHz 64 kbps (celt) 4.09 MHz 8.08 MHz 6.24 MHz 128 kbps (celt) 1.93 MHz 8.83 MHz 6.53 MHz Change-Id: I851733a8a5824b61feb363a173091bc7e6629b58 --- lib/rbcodec/codecs/libopus/celt/cwrs.c | 40 ++++++++++++++++++++++++---------- 1 file changed, 29 insertions(+), 11 deletions(-) (limited to 'lib/rbcodec/codecs/libopus/celt/cwrs.c') diff --git a/lib/rbcodec/codecs/libopus/celt/cwrs.c b/lib/rbcodec/codecs/libopus/celt/cwrs.c index 03b86985d2..921100fdc2 100644 --- a/lib/rbcodec/codecs/libopus/celt/cwrs.c +++ b/lib/rbcodec/codecs/libopus/celt/cwrs.c @@ -460,10 +460,12 @@ void encode_pulses(const int *_y,int _n,int _k,ec_enc *_enc){ ec_enc_uint(_enc,icwrs(_n,_y),CELT_PVQ_V(_n,_k)); } -static void cwrsi(int _n,int _k,opus_uint32 _i,int *_y){ +static opus_val32 cwrsi(int _n,int _k,opus_uint32 _i,int *_y){ opus_uint32 p; int s; int k0; + opus_int16 val; + opus_val32 yy=0; celt_assert(_k>0); celt_assert(_n>1); while(_n>2){ @@ -487,7 +489,9 @@ static void cwrsi(int _n,int _k,opus_uint32 _i,int *_y){ } else for(p=row[_k];p>_i;p=row[_k])_k--; _i-=p; - *_y++=(k0-_k+s)^s; + val=(k0-_k+s)^s; + *_y++=val; + yy=MAC16_16(yy,val,val); } /*Lots of dimensions case:*/ else{ @@ -507,7 +511,9 @@ static void cwrsi(int _n,int _k,opus_uint32 _i,int *_y){ do p=CELT_PVQ_U_ROW[--_k][_n]; while(p>_i); _i-=p; - *_y++=(k0-_k+s)^s; + val=(k0-_k+s)^s; + *_y++=val; + yy=MAC16_16(yy,val,val); } } _n--; @@ -519,14 +525,19 @@ static void cwrsi(int _n,int _k,opus_uint32 _i,int *_y){ k0=_k; _k=(_i+1)>>1; if(_k)_i-=2*_k-1; - *_y++=(k0-_k+s)^s; + val=(k0-_k+s)^s; + *_y++=val; + yy=MAC16_16(yy,val,val); /*_n==1*/ s=-(int)_i; - *_y=(_k+s)^s; + val=(_k+s)^s; + *_y=val; + yy=MAC16_16(yy,val,val); + return yy; } -void decode_pulses(int *_y,int _n,int _k,ec_dec *_dec){ - cwrsi(_n,_k,ec_dec_uint(_dec,CELT_PVQ_V(_n,_k)),_y); +opus_val32 decode_pulses(int *_y,int _n,int _k,ec_dec *_dec){ + return cwrsi(_n,_k,ec_dec_uint(_dec,CELT_PVQ_V(_n,_k)),_y); } #else /* SMALL_FOOTPRINT */ @@ -591,8 +602,10 @@ static opus_uint32 ncwrs_urow(unsigned _n,unsigned _k,opus_uint32 *_u){ _y: Returns the vector of pulses. _u: Must contain entries [0..._k+1] of row _n of U() on input. Its contents will be destructively modified.*/ -static void cwrsi(int _n,int _k,opus_uint32 _i,int *_y,opus_uint32 *_u){ +static opus_val32 cwrsi(int _n,int _k,opus_uint32 _i,int *_y,opus_uint32 *_u){ int j; + opus_int16 val; + opus_val32 yy=0; celt_assert(_n>0); j=0; do{ @@ -607,10 +620,13 @@ static void cwrsi(int _n,int _k,opus_uint32 _i,int *_y,opus_uint32 *_u){ while(p>_i)p=_u[--_k]; _i-=p; yj-=_k; - _y[j]=(yj+s)^s; + val=(yj+s)^s; + _y[j]=val; + yy=MAC16_16(yy,val,val); uprev(_u,_k+2,0); } while(++j<_n); + return yy; } /*Returns the index of the given combination of K elements chosen from a set @@ -685,13 +701,15 @@ void encode_pulses(const int *_y,int _n,int _k,ec_enc *_enc){ RESTORE_STACK; } -void decode_pulses(int *_y,int _n,int _k,ec_dec *_dec){ +opus_val32 decode_pulses(int *_y,int _n,int _k,ec_dec *_dec){ VARDECL(opus_uint32,u); + int ret; SAVE_STACK; celt_assert(_k>0); ALLOC(u,_k+2U,opus_uint32); - cwrsi(_n,_k,ec_dec_uint(_dec,ncwrs_urow(_n,_k,u)),_y,u); + ret = cwrsi(_n,_k,ec_dec_uint(_dec,ncwrs_urow(_n,_k,u)),_y,u); RESTORE_STACK; + return ret; } #endif /* SMALL_FOOTPRINT */ -- cgit v1.2.3