search for: xcorr

...*/ }; -#endif +#endif /* HAVE_ARM_NE10 */ # if defined(FIXED_POINT) + +#if defined(OPUS_ARM_MAY_HAVE_NEON) || defined(OPUS_ARM_MAY_HAVE_EDSP) + +#if !defined(OPUS_ARM_PRESUME_NEON) && (!defined(OPUS_ARM_PRESUME_EDSP) || defined(OPUS_ARM_MAY_HAVE_NEON)) + opus_val32 (*const CELT_PITCH_XCORR_IMPL[OPUS_ARCHMASK+1])(const opus_val16 *, const opus_val16 *, opus_val32 *, int , int) = { celt_pitch_xcorr_c, /* ARMv4 */ @@ -104,8 +109,23 @@ opus_val32 (*const CELT_PITCH_XCORR_IMPL[OPUS_ARCHMASK+1])(const opus_val16 *, MAY_HAVE_MEDIA(celt_pitch_xcorr), /* Media */ M...

...+ ((defined(OPUS_ARM_MAY_HAVE_NEON) && !defined(OPUS_ARM_PRESUME_NEON)) || \ > + (defined(OPUS_ARM_MAY_HAVE_MEDIA) && !defined(OPUS_ARM_PRESUME_MEDIA)) || \ > + (defined(OPUS_ARM_MAY_HAVE_EDSP) && !defined(OPUS_ARM_PRESUME_EDSP))) > opus_val32 (*const CELT_PITCH_XCORR_IMPL[OPUS_ARCHMASK+1])(const opus_val16 *, > const opus_val16 *, opus_val32 *, int , int) = { Maybe I'm missing something, but... > -/*Is run-time CPU detection enabled on this platform?*/ > -# if defined(OPUS_HAVE_RTCD) && (defined(OPUS_ARM_ASM) \ > - || (defined(...

...ibrary. It brings with it lots of complications (to name one: wouldn't the .pc files need to be updated?). Please use the same mechanism that the SSE intrinsics use to add CFLAGS to the compilation of specific object files, e.g., $(SSE_OBJ): CFLAGS += -msse4.1 > +void (*const CELT_PITCH_XCORR_IMPL[OPUS_ARCHMASK+1])(const opus_val16 *, > + const opus_val16 *, opus_val32 *, int , int, int) = { > + celt_pitch_xcorr_c, /* ARMv4 */ > + celt_pitch_xcorr_c, /* EDSP */ > + celt_pitch_xcorr_c, /* Media */ > +#if defined(OPUS_ARM_NEON_INTR) > + celt_...

..._node/Per_002dObject-Flags.html which specifically advocates against per object flags. I can follow your feedback if you still think per-object flag is the right thing to do for libopus. Please let me know either way. > > $(SSE_OBJ): CFLAGS += -msse4.1 > >> +void (*const CELT_PITCH_XCORR_IMPL[OPUS_ARCHMASK+1])(const opus_val16 *, >> + const opus_val16 *, opus_val32 *, int , int, int) = { >> + celt_pitch_xcorr_c, /* ARMv4 */ >> + celt_pitch_xcorr_c, /* EDSP */ >> + celt_pitch_xcorr_c, /* Media...

...m engineers working on NE10 [1] that it would be better to use NE10 [1] for FFT optimizations for opus use cases. However, these FFT patches are currently in review and haven't been integrated into NE10 yet. While the FFT functions in NE10 are getting baked, I wanted to optimize the celt_pitch_xcorr (floating point only) and use it to introduce ARM NEON intrinsics into libopus project. This will also lay the foundation when FFT routines in NE10 become available. I would like to know the following as I DID NOT fully use-case test this celt_pitch_xcorr optimization. Did only unit tests. a. Si...

Optimize celt_pitch_xcorr function (for fixed point). Even though same code in theory should work for ARMv7 as well, turning this on only for aarch64 at the moment since there is a fixed point asm implementation for ARMv7 neon. Signed-off-by: Viswanath Puttagunta <viswanath.puttagunta at linaro.org> --- celt/arm/cel...

Optimize celt_pitch_xcorr function (for fixed point). Even though same code in theory should work for ARMv7 as well, turning this on only for aarch64 at the moment since there is a fixed point asm implementation for ARMv7 neon. Signed-off-by: Viswanath Puttagunta <viswanath.puttagunta at linaro.org> --- celt/arm/cel...

Optimize celt_pitch_xcorr function (for floating point) using ARM NEON intrinsics for SoCs that have NEON VFP unit. As initial step, targeting ARMv7 NEON (VFP3+) based SoCs. To enable this optimization, use --enable-arm-neon-intrinsics configure option. This flag is not enabled by default. Compile time and runtime checks...

...19 +++++++++++++++++++ celt/pitch.h | 19 ------------------- 2 files changed, 19 insertions(+), 19 deletions(-) diff --git a/celt/arm/pitch_arm.h b/celt/arm/pitch_arm.h index d5c9408..fe76f8d 100644 --- a/celt/arm/pitch_arm.h +++ b/celt/arm/pitch_arm.h @@ -75,4 +75,23 @@ void celt_pitch_xcorr_float_neon(const opus_val16 *_x, const opus_val16 *_y, #endif #endif /* end !FIXED_POINT */ + +/*Is run-time CPU detection enabled on this platform?*/ +# if defined(OPUS_HAVE_RTCD) && \ + (defined(OPUS_ARM_ASM) || defined(OPUS_ARM_MAY_HAVE_NEON_INTR)) +extern +# if defined(FIXED_POINT...

...const opus_val16 * OPUS_RESTRICT x_lp, opus_val16 * OPUS_RESTR for (j=0;j<len>>1;j++) sum += SHR32(MULT16_16(x_lp[j],y[i+j]), shift); #else - sum = celt_inner_prod_c(x_lp, y+i, len>>1); + sum = celt_inner_prod(x_lp, y+i, len>>1, arch); #endif xcorr[i] = MAX32(-1, sum); #ifdef FIXED_POINT diff --git a/src/opus_multistream_encoder.c b/src/opus_multistream_encoder.c index c07132f..6ecea5d 100644 --- a/src/opus_multistream_encoder.c +++ b/src/opus_multistream_encoder.c @@ -295,7 +295,7 @@ void surround_analysis(const CELTMode *celt_mode, const v...

This sequence of patches provides arm64 support for Opus. Tested on iOS, Android, and Ubuntu 14.04. The patch sequence was written on top of Viswanath Puttagunta's Ne10 patches, but all but the second ("Reorganize pitch_arm.h") should, I think, apply independently of it. It does depends on my previous intrinsics configury reorganization, however. Comments welcome. With this and

..._ptr; opus_int i, j, k, lag_counter, lag_low, lag_high; opus_int nb_cbk_search, delta, idx, cbk_size; silk_float scratch_mem[ SCRATCH_SIZE ]; @@ -529,7 +529,6 @@ static void silk_P_Ana_calc_corr_st3( silk_assert(lag_high-lag_low+1 <= SCRATCH_SIZE); celt_pitch_xcorr( target_ptr, target_ptr - start_lag - lag_high, xcorr, sf_length, lag_high - lag_low + 1 ); for( j = lag_low; j <= lag_high; j++ ) { - basis_ptr = target_ptr - ( start_lag + j ); silk_assert( lag_counter < SCRATCH_SIZE ); scratch_mem[ lag_counter...

Hi, Optimizes celt_pitch_xcorr for ARM NEON floating point. Changes from RFCv3: - celt_neon_intr.c - removed warnings due to not having constant pointers - Put simpler loop to take care of corner cases. Unrolling using intrinsics was not really mapping well to what was done in celt_pitch_xcorr_arm.s - Makefile.am...

...phai ,matrix(1,t,1))#kronecker of alphai cormat<-matrix(c(1,rho,rho,1),nrow=2,ncol=2)#correlation matrix cormat.chold <- chol(cormat)#choleski transformation of correlation matrix akrox <- cbind(akro,x) ax <- akrox%*%cormat.chold ai <- as.matrix(ax[,1]) pData$alphai<-as.vector(ai) xcorr <- as.matrix(ax[,2:(1+ncol(x))]) pData$xcorrei<-as.vector(xcorr) pData$yi <- 5 + pData$alphai + 5* pData$xcorrei + rnorm(nt) ##########################panel data frame################################## library(plm) pData <- pdata.frame(pData, c("id", "time")) pData...

...phai ,matrix(1,t,1))#kronecker of alphai cormat<-matrix(c(1,rho,rho,1),nrow=2,ncol=2)#correlation matrix cormat.chold <- chol(cormat)#choleski transformation of correlation matrix akrox <- cbind(akro,x) ax <- akrox%*%cormat.chold ai <- as.matrix(ax[,1]) pData$alphai<-as.vector(ai) xcorr <- as.matrix(ax[,2:(1+ncol(x))]) pData$xcorrei<-as.vector(xcorr) pData$yi <- 5 + pData$alphai + 5* pData$xcorrei + rnorm(nt) ##########################panel data frame################################## library(plm) pData <- pdata.frame(pData, c("id", "time")) pData...

...phai ,matrix(1,t,1))#kronecker of alphai cormat<-matrix(c(1,rho,rho,1),nrow=2,ncol=2)#correlation matrix cormat.chold <- chol(cormat)#choleski transformation of correlation matrix akrox <- cbind(akro,x) ax <- akrox%*%cormat.chold ai <- as.matrix(ax[,1]) pData$alphai<-as.vector(ai) xcorr <- as.matrix(ax[,2:(1+ncol(x))]) pData$xcorrei<-as.vector(xcorr) pData$yi <- 5 + pData$alphai + 5* pData$xcorrei + rnorm(nt) ##########################panel data frame################################## library(plm) pData <- pdata.frame(pData, c("id", "time")) pData...

...T) && \ + ((defined(OPUS_ARM_MAY_HAVE_NEON) && !defined(OPUS_ARM_PRESUME_NEON)) || \ + (defined(OPUS_ARM_MAY_HAVE_MEDIA) && !defined(OPUS_ARM_PRESUME_MEDIA)) || \ + (defined(OPUS_ARM_MAY_HAVE_EDSP) && !defined(OPUS_ARM_PRESUME_EDSP))) opus_val32 (*const CELT_PITCH_XCORR_IMPL[OPUS_ARCHMASK+1])(const opus_val16 *, const opus_val16 *, opus_val32 *, int , int) = { celt_pitch_xcorr_c, /* ARMv4 */ diff --git a/celt/arm/pitch_arm.h b/celt/arm/pitch_arm.h index eaf61c9..bd41774 100644 --- a/celt/arm/pitch_arm.h +++ b/celt/arm/pitch_arm.h @@ -46,7 +46...

...(OPUS_ARM_MAY_HAVE_NEON) && !defined(OPUS_ARM_PRESUME_NEON)) || \ >> + (defined(OPUS_ARM_MAY_HAVE_MEDIA) && !defined(OPUS_ARM_PRESUME_MEDIA)) || \ >> + (defined(OPUS_ARM_MAY_HAVE_EDSP) && !defined(OPUS_ARM_PRESUME_EDSP))) >> opus_val32 (*const CELT_PITCH_XCORR_IMPL[OPUS_ARCHMASK+1])(const opus_val16 *, >> const opus_val16 *, opus_val32 *, int , int) = { > > Maybe I'm missing something, but... > >> -/*Is run-time CPU detection enabled on this platform?*/ >> -# if defined(OPUS_HAVE_RTCD) && (defined(OPUS_ARM_AS...

...akrox <- cbind(akro,x) ax <- akrox%*%cormat.chold ai <- as.matrix(ax[,1]) pData$alphai<-as.vector(ai) xcorr <- as.matrix(ax[,2:(1+ncol(x))]) pData$xcorrei<-as.vector(xcorr) pData$yi <- 5 + pData$alphai + 5* pData$xcorrei + rnorm(nt) ##########################panel data frame################################## library(plm) pData <- pdata.frame(pData, c("id", "time")) p...

Optimize celt_pitch_xcorr function (for floating point) using ARM NEON intrinsics for SoCs that have NEON VFP unit. To enable this optimization, use --enable-intrinsics configure option. Compile time and runtime checks are also supported to make sure this optimization is only enabled when the compiler supports neon intrin...