search for: opus_int32

...CE OR OTHERWISE) +ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE +POSSIBILITY OF SUCH DAMAGE. +***********************************************************************/ + +#ifndef SILK_SIGPROC_FIX_ARM5E_H +#define SILK_SIGPROC_FIX_ARM5E_H + +#undef silk_MLA +static inline opus_int32 silk_MLA(opus_int32 a, opus_int32 b, opus_int32 c) +{ + opus_int32 res; + __asm__( + "mla %0, %1, %2, %3;\n" + : "=&r"(res) + : "r"(b), "r"(c), "r"(a) + ); + return res; +} + +#undef silk_SMULTT +static inline opus_int32 s...

...ff --git a/silk/NSQ.c b/silk/NSQ.c index d8513dc..ec81f3b 100644 --- a/silk/NSQ.c +++ b/silk/NSQ.c @@ -205,7 +205,7 @@ void silk_noise_shape_quantizer( int arch /* I Architecture */ ) { - opus_int i, j; + opus_int i; opus_int32 LTP_pred_Q13, LPC_pred_Q10, n_AR_Q12, n_LTP_Q13; opus_int32 n_LF_Q12, r_Q10, rr_Q10, q1_Q0, q1_Q10, q2_Q10, rd1_Q20, rd2_Q20; opus_int32 exc_Q14, LPC_exc_Q14, xq_Q14, Gain_Q10; @@ -248,21 +248,7 @@ void silk_noise_shape_quantizer( /* Noise shape feedback */ silk_...

...ff --git a/silk/NSQ.c b/silk/NSQ.c index d8513dc..ec81f3b 100644 --- a/silk/NSQ.c +++ b/silk/NSQ.c @@ -205,7 +205,7 @@ void silk_noise_shape_quantizer( int arch /* I Architecture */ ) { - opus_int i, j; + opus_int i; opus_int32 LTP_pred_Q13, LPC_pred_Q10, n_AR_Q12, n_LTP_Q13; opus_int32 n_LF_Q12, r_Q10, rr_Q10, q1_Q0, q1_Q10, q2_Q10, rd1_Q20, rd2_Q20; opus_int32 exc_Q14, LPC_exc_Q14, xq_Q14, Gain_Q10; @@ -248,21 +248,7 @@ void silk_noise_shape_quantizer( /* Noise shape feedback */ silk_...

--- silk/macros.h | 10 +++++----- 1 file changed, 5 insertions(+), 5 deletions(-) diff --git a/silk/macros.h b/silk/macros.h index 1ba614a..e1e05b9 100644 --- a/silk/macros.h +++ b/silk/macros.h @@ -48,14 +48,14 @@ POSSIBILITY OF SUCH DAMAGE. /* (a32 * (opus_int32)((opus_int16)(b32))) >> 16 output have to be 32bit int */ #if OPUS_FAST_INT64 -#define silk_SMULWB(a32, b32) (((a32) * (opus_int64)((opus_int16)(b32))) >> 16) +#define silk_SMULWB(a32, b32) ((opus_int32)(((a32) * (opus_int64)((opus_int16)(b32))) >> 16)) #el...

...RDER ], /* I Noise shaping coefs */ + const opus_int HarmShapeGain_Q14[ MAX_NB_SUBFR ], /* I Long term shaping coefs */ + const opus_int Tilt_Q14[ MAX_NB_SUBFR ], /* I Spectral tilt */ + const opus_int32 LF_shp_Q14[ MAX_NB_SUBFR ], /* I Low frequency shaping coefs */ + const opus_int32 Gains_Q16[ MAX_NB_SUBFR ], /* I Quantization step sizes */ + const opus_int pitchL[ MAX_NB_SUBFR ], /* I...

NSQ_LPC_BUF_LENGTH is independent of DECISION_DELAY. --- silk/define.h | 4 ---- 1 file changed, 4 deletions(-) diff --git a/silk/define.h b/silk/define.h index 781cfdc..1286048 100644 --- a/silk/define.h +++ b/silk/define.h @@ -173,11 +173,7 @@ extern "C" #define MAX_MATRIX_SIZE MAX_LPC_ORDER /* Max of LPC Order and LTP order */ -#if( MAX_LPC_ORDER >

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

Following Tim's comments, here are my reworked patches for the Neon intrinsic function patches of of my Aarch64 patchset, i.e. replacing patches 5-8 of the v2 series. Patches 1-4 and 9-18 of the old series still apply unmodified. The one new (as opposed to changed) patch is the first one in this series, to add named constants for the ARM architecture variants. There are also some minor code

Here are my aarch64 patches rebased to the current tip of Opus master. They're largely the same as my previous patch set, with the addition of the final one (the Neon fixed-point implementation of xcorr_kernel). This replaces Viswanath's Neon fixed-point celt_pitch_xcorr, since xcorr_kernel is used in celt_fir and celt_iir as well. These have been tested for correctness under qemu

...plementation of xcorr_kernel. Enable intrinsics by default. Clean up some intrinsics-related wording in configure. Move OPUS_FAST_INT64 definition to celt/arch.h. Add OPUS_FAST_INT64 flavors of celt/fixed_generic.h macros. Explicitly cast results of silk OPUS_FAST_INT64 macros back to opus_int32. Add OPUS_FAST_INT64 definition of silk_SMULWT. Clean up formatting of configure output for ARM intrinsics detection. Add configure check for Aarch64-specific Neon intrinsics. Add Aarch64 intrinsics for saturated add/subtract. Add Aarch64 intrinsic for SIG2WORD16. Makefile.am...

...OTHERWISE) +ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE +POSSIBILITY OF SUCH DAMAGE. +***********************************************************************/ + +#ifndef SILK_SIGPROC_FIX_ARM64_H +#define SILK_SIGPROC_FIX_ARM64_H + +#undef silk_MLA +static OPUS_INLINE opus_int32 silk_MLA_arm64(opus_int32 a, opus_int32 b, + opus_int32 c) +{ + opus_int32 res; + __asm__( + "#silk_MLA\n\t" + "madd %w0, %w1, %w2, %w3\n\t" + : "=&r"(res) + : "r"(b), "r"(c), "r"(a) + ); + return res; +} +#defi...

...(__GNUC__) && (defined(_STDINT_H) || defined(_STDINT_H_)) || defined (HAVE_STDINT_H)) -#include <stdint.h> +# include <stdint.h> + typedef int8_t opus_int8; + typedef uint8_t opus_uint8; typedef int16_t opus_int16; typedef uint16_t opus_uint16; typedef int32_t opus_int32; typedef uint32_t opus_uint32; + typedef int64_t opus_int64; + typedef uint64_t opus_uint64; + #elif defined(_WIN32) # if defined(__CYGWIN__) -# include <_G_config.h> - typedef _G_int32_t opus_int32; - typedef _G_uint32_t opus_uint32; - typedef _G_int16 opus_int...

...SSIBILITY OF SUCH DAMAGE. #define opus_unlikely(x) (!!(x)) #endif -/* Set this if opus_int64 is a native type of the CPU. */ -#define OPUS_FAST_INT64 (defined(__x86_64__) || defined(__LP64__) || defined(_WIN64)) - /* This is an OPUS_INLINE header file for general platform. */ /* (a32 * (opus_int32)((opus_int16)(b32))) >> 16 output have to be 32bit int */ -- 2.4.9 (Apple Git-60)

Jonathan Lennox wrote: > +opus_int32 silk_noise_shape_quantizer_short_prediction_neon(const opus_int32 *buf32, const opus_int32 *coef32) > +{ > + int32x4_t coef0 = vld1q_s32(coef32); > + int32x4_t coef1 = vld1q_s32(coef32 + 4); > + int32x4_t coef2 = vld1q_s32(coef32 + 8); > + int32x4_t coef3 = vld1q_s32(coef...

...ows * cols * sizeof(opus_int16)" may > not have the necessary alignment. There may be alignment padding > after the struct and also after the opus_int16 array and this function > is assumed to return a size that includes both. Also the > multiplication and return value should use opus_int32 since int may > not be large enough for order 11 to 14. A complication is that > align() currently takes and returns "int". > > We specify in the IETF doc that we are limited to a max matrix size of 65025 bytes (and with 2 bytes per cell, that limits us to only order 11 for a...

...opus_multistream_encoder.c b/src/opus_multistream_encoder.c index 3efab53..6f3eb53 100644 --- a/src/opus_multistream_encoder.c +++ b/src/opus_multistream_encoder.c @@ -454,7 +454,9 @@ static int opus_multistream_encode_native unsigned char tmp_data[MS_FRAME_TMP]; OpusRepacketizer rp; opus_int32 complexity; +#ifndef FIXED_POINT AnalysisInfo analysis_info; +#endif const CELTMode *celt_mode; opus_int32 bitrates[256]; opus_val16 bandLogE[42]; @@ -629,27 +631,6 @@ static int opus_multistream_encode_native } -#if !defined(DISABLE_FLOAT_API) -static void opus_copy_channel...

--- configure.ac | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/configure.ac b/configure.ac index f52d2c2..e1a6e9b 100644 --- a/configure.ac +++ b/configure.ac @@ -190,7 +190,7 @@ AC_ARG_ENABLE([rtcd], [enable_rtcd=yes]) AC_ARG_ENABLE([intrinsics], - [AS_HELP_STRING([--disable-intrinsics], [Disable intrinsics optimizations for ARM(float) X86(fixed)])],, +

This patch adds a macro abstracting the condition under which the silk math macros use opus_int64-based calculations rather than opus_int32. No substantive change, but will make it easier to adjust if additional such platforms are found in the future. --- silk/macros.h | 13 ++++++++----- 1 file changed, 8 insertions(+), 5 deletions(-) diff --git a/silk/macros.h b/silk/macros.h index 2f24950..bc30303 100644 --- a/silk/macros.h +++ b...

...(for the whole encoder) than the original C code using 32-bit multiplications (for both strides and the same stride 2 unrolling). void silk_biquad_alt_c_MulSingleAQ28( const opus_int16 *in, /* I input signal */ const opus_int32 *B_Q28, /* I MA coefficients [3] */ const opus_int32 *A_Q28, /* I AR coefficients [2] */ opus_int32 *S, /* I/O State vector...

...ion filter order */ + opus_int predictLPCOrder, /* I Prediction filter order */ + int arch /* I Architecture */ ) { opus_int i, j; @@ -207,6 +211,9 @@ void silk_noise_shape_quantizer( opus_int32 exc_Q14, LPC_exc_Q14, xq_Q14, Gain_Q10; opus_int32 tmp1, tmp2, sLF_AR_shp_Q14; opus_int32 *psLPC_Q14, *shp_lag_ptr, *pred_lag_ptr; +#ifdef OPUS_ARM_MAY_HAVE_NEON_INTR + opus_int32 a_Q12_rev[16]; +#endif shp_lag_ptr = &NSQ->sLTP_shp_Q14[ NSQ->sLTP_shp_buf_idx -...