chris@162: /**
chris@162:  * Copyright (c) 2014, 2015, Enzien Audio Ltd.
chris@162:  *
chris@162:  * Permission to use, copy, modify, and/or distribute this software for any
chris@162:  * purpose with or without fee is hereby granted, provided that the above
chris@162:  * copyright notice and this permission notice appear in all copies.
chris@162:  *
chris@162:  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES WITH
chris@162:  * REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY
chris@162:  * AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY SPECIAL, DIRECT,
chris@162:  * INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM
chris@162:  * LOSS OF USE, DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR
chris@162:  * OTHER TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR
chris@162:  * PERFORMANCE OF THIS SOFTWARE.
chris@162:  */
chris@162: 
chris@162: #include "SignalBiquad.h"
chris@162: 
chris@162: // http://reanimator-web.appspot.com/articles/simdiir
chris@162: // http://musicdsp.org/files/Audio-EQ-Cookbook.txt
chris@162: 
chris@162: hv_size_t sBiquad_init(SignalBiquad *o) {
chris@162: #if HV_SIMD_AVX
chris@162:   o->xm1 = _mm256_setzero_ps();
chris@162:   o->xm2 = _mm256_setzero_ps();
chris@162: #elif HV_SIMD_SSE
chris@162:   o->xm1 = _mm_setzero_ps();
chris@162:   o->xm2 = _mm_setzero_ps();
chris@162: #elif HV_SIMD_NEON
chris@162:   o->xm1 = vdupq_n_f32(0.0f);
chris@162:   o->xm2 = vdupq_n_f32(0.0f);
chris@162: #else // HV_SIMD_NONE
chris@162:   o->x1 = 0.0f;
chris@162:   o->x2 = 0.0f;
chris@162: #endif
chris@162:   o->y1 = 0.0f;
chris@162:   o->y2 = 0.0f;
chris@162:   return 0;
chris@162: }
chris@162: 
chris@162: void __hv_biquad_f(SignalBiquad *o, hv_bInf_t bIn, hv_bInf_t bX0, hv_bInf_t bX1, hv_bInf_t bX2, hv_bInf_t bY1, hv_bInf_t bY2, hv_bOutf_t bOut) {
chris@162: #if HV_SIMD_AVX
chris@162:   __m256 a = _mm256_mul_ps(bIn, bX0);
chris@162:   __m256 b = _mm256_mul_ps(o->xm1, bX1);
chris@162:   __m256 c = _mm256_mul_ps(o->xm2, bX2);
chris@162:   __m256 d = _mm256_add_ps(a, b);
chris@162:   __m256 e = _mm256_add_ps(c, d); // bIn*bX0 + o->x1*bX1 + o->x2*bX2
chris@162:   float y0 = e[0] - o->y1*bY1[0] - o->y2*bY2[0];
chris@162:   float y1 = e[1] - y0*bY1[1] - o->y1*bY2[1];
chris@162:   float y2 = e[2] - y1*bY1[2] - y0*bY2[2];
chris@162:   float y3 = e[3] - y2*bY1[3] - y1*bY2[3];
chris@162:   float y4 = e[4] - y3*bY1[4] - y2*bY2[4];
chris@162:   float y5 = e[5] - y4*bY1[5] - y3*bY2[5];
chris@162:   float y6 = e[6] - y5*bY1[6] - y4*bY2[6];
chris@162:   float y7 = e[7] - y6*bY1[7] - y5*bY2[7];
chris@162: 
chris@162:   o->xm2 = o->xm1;
chris@162:   o->xm1 = bIn;
chris@162:   o->y1 = y7;
chris@162:   o->y2 = y6;
chris@162: 
chris@162:   *bOut = _mm256_set_ps(y7, y6, y5, y4, y3, y2, y1, y0);
chris@162: #elif HV_SIMD_SSE
chris@162:   __m128 a = _mm_mul_ps(bIn, bX0);
chris@162:   __m128 b = _mm_mul_ps(o->xm1, bX1);
chris@162:   __m128 c = _mm_mul_ps(o->xm2, bX2);
chris@162:   __m128 d = _mm_add_ps(a, b);
chris@162:   __m128 e = _mm_add_ps(c, d);
chris@162:   float y0 = e[0] - o->y1*bY1[0] - o->y2*bY2[0];
chris@162:   float y1 = e[1] - y0*bY1[1] - o->y1*bY2[1];
chris@162:   float y2 = e[2] - y1*bY1[2] - y0*bY2[2];
chris@162:   float y3 = e[3] - y2*bY1[3] - y1*bY2[3];
chris@162: 
chris@162:   o->xm2 = o->xm1;
chris@162:   o->xm1 = bIn;
chris@162:   o->y1 = y3;
chris@162:   o->y2 = y2;
chris@162: 
chris@162:   *bOut = _mm_set_ps(y3, y2, y1, y0);
chris@162: #elif HV_SIMD_NEON
chris@162:   float32x4_t a = vmulq_f32(bIn, bX0);
chris@162:   float32x4_t b = vmulq_f32(o->xm1, bX1);
chris@162:   float32x4_t c = vmulq_f32(o->xm2, bX2);
chris@162:   float32x4_t d = vaddq_f32(a, b);
chris@162:   float32x4_t e = vaddq_f32(c, d);
chris@162:   float y0 = e[0] - o->y1*bY1[0] - o->y2*bY2[0];
chris@162:   float y1 = e[1] - y0*bY1[1] - o->y1*bY2[1];
chris@162:   float y2 = e[2] - y1*bY1[2] - y0*bY2[2];
chris@162:   float y3 = e[3] - y2*bY1[3] - y1*bY2[3];
chris@162: 
chris@162:   o->xm2 = o->xm1;
chris@162:   o->xm1 = bIn;
chris@162:   o->y1 = y3;
chris@162:   o->y2 = y2;
chris@162: 
chris@162:   *bOut = (float32x4_t) {y0, y1, y2, y3};
chris@162: #else
chris@162:   const float y = bIn*bX0 + o->x1*bX1 + o->x2*bX2 - o->y1*bY1 - o->y2*bY2;
chris@162:   o->x2 = o->x1; o->x1 = bIn;
chris@162:   o->y2 = o->y1; o->y1 = y;
chris@162:   *bOut = y;
chris@162: #endif
chris@162: }
chris@162: 
chris@162: static void sBiquad_k_updateCoefficients(SignalBiquad_k *const o) {
chris@162:   // calculate all filter coefficients in the double domain
chris@162: #if HV_SIMD_AVX || HV_SIMD_SSE || HV_SIMD_NEON
chris@162:   double b0 = (double) o->b0;
chris@162:   double b1 = (double) o->b1;
chris@162:   double b2 = (double) o->b2;
chris@162:   double a1 = (double) -o->a1;
chris@162:   double a2 = (double) -o->a2;
chris@162: 
chris@162:   double coeffs[4][8] =
chris@162:   {
chris@162:     { 0,  0,  0,  b0, b1, b2, a1, a2 },
chris@162:     { 0,  0,  b0, b1, b2, 0,  a2, 0  },
chris@162:     { 0,  b0, b1, b2, 0,  0,  0,  0  },
chris@162:     { b0, b1, b2, 0,  0,  0,  0,  0  },
chris@162:   };
chris@162: 
chris@162:   for (int i = 0; i < 8; i++) {
chris@162:     coeffs[1][i] += a1*coeffs[0][i];
chris@162:     coeffs[2][i] += a1*coeffs[1][i] + a2*coeffs[0][i];
chris@162:     coeffs[3][i] += a1*coeffs[2][i] + a2*coeffs[1][i];
chris@162:   }
chris@162: 
chris@162: #if HV_SIMD_AVX || HV_SIMD_SSE
chris@162:   o->coeff_xp3 = _mm_set_ps((float) coeffs[3][0], (float) coeffs[2][0], (float) coeffs[1][0], (float) coeffs[0][0]);
chris@162:   o->coeff_xp2 = _mm_set_ps((float) coeffs[3][1], (float) coeffs[2][1], (float) coeffs[1][1], (float) coeffs[0][1]);
chris@162:   o->coeff_xp1 = _mm_set_ps((float) coeffs[3][2], (float) coeffs[2][2], (float) coeffs[1][2], (float) coeffs[0][2]);
chris@162:   o->coeff_x0 =  _mm_set_ps((float) coeffs[3][3], (float) coeffs[2][3], (float) coeffs[1][3], (float) coeffs[0][3]);
chris@162:   o->coeff_xm1 = _mm_set_ps((float) coeffs[3][4], (float) coeffs[2][4], (float) coeffs[1][4], (float) coeffs[0][4]);
chris@162:   o->coeff_xm2 = _mm_set_ps((float) coeffs[3][5], (float) coeffs[2][5], (float) coeffs[1][5], (float) coeffs[0][5]);
chris@162:   o->coeff_ym1 = _mm_set_ps((float) coeffs[3][6], (float) coeffs[2][6], (float) coeffs[1][6], (float) coeffs[0][6]);
chris@162:   o->coeff_ym2 = _mm_set_ps((float) coeffs[3][7], (float) coeffs[2][7], (float) coeffs[1][7], (float) coeffs[0][7]);
chris@162: #else // HV_SIMD_NEON
chris@162:   o->coeff_xp3 = (float32x4_t) {(float) coeffs[0][0], (float) coeffs[1][0], (float) coeffs[2][0], (float) coeffs[3][0]};
chris@162:   o->coeff_xp2 = (float32x4_t) {(float) coeffs[0][1], (float) coeffs[1][1], (float) coeffs[2][1], (float) coeffs[3][1]};
chris@162:   o->coeff_xp1 = (float32x4_t) {(float) coeffs[0][2], (float) coeffs[1][2], (float) coeffs[2][2], (float) coeffs[3][2]};
chris@162:   o->coeff_x0 =  (float32x4_t) {(float) coeffs[0][3], (float) coeffs[1][3], (float) coeffs[2][3], (float) coeffs[3][3]};
chris@162:   o->coeff_xm1 = (float32x4_t) {(float) coeffs[0][4], (float) coeffs[1][4], (float) coeffs[2][4], (float) coeffs[3][4]};
chris@162:   o->coeff_xm2 = (float32x4_t) {(float) coeffs[0][5], (float) coeffs[1][5], (float) coeffs[2][5], (float) coeffs[3][5]};
chris@162:   o->coeff_ym1 = (float32x4_t) {(float) coeffs[0][6], (float) coeffs[1][6], (float) coeffs[2][6], (float) coeffs[3][6]};
chris@162:   o->coeff_ym2 = (float32x4_t) {(float) coeffs[0][7], (float) coeffs[1][7], (float) coeffs[2][7], (float) coeffs[3][7]};
chris@162: #endif
chris@162: #endif
chris@162:   // NOTE(mhroth): not necessary to calculate any coefficients for HV_SIMD_NONE case
chris@162: }
chris@162: 
chris@162: hv_size_t sBiquad_k_init(SignalBiquad_k *o, float b0, float b1, float b2, float a1, float a2) {
chris@162:   // initialise filter coefficients
chris@162:   o->b0 = b0;
chris@162:   o->b1 = b1;
chris@162:   o->b2 = b2;
chris@162:   o->a1 = a1;
chris@162:   o->a2 = a2;
chris@162:   sBiquad_k_updateCoefficients(o);
chris@162: 
chris@162:   // clear filter state
chris@162: #if HV_SIMD_AVX || HV_SIMD_SSE
chris@162:   o->xm1 = _mm_setzero_ps();
chris@162:   o->xm2 = _mm_setzero_ps();
chris@162:   o->ym1 = _mm_setzero_ps();
chris@162:   o->ym2 = _mm_setzero_ps();
chris@162: #elif HV_SIMD_NEON
chris@162:   o->xm1 = vdupq_n_f32(0.0f);
chris@162:   o->xm2 = vdupq_n_f32(0.0f);
chris@162:   o->ym1 = vdupq_n_f32(0.0f);
chris@162:   o->ym2 = vdupq_n_f32(0.0f);
chris@162: #else // HV_SIMD_NONE
chris@162:   o->xm1 = 0.0f;
chris@162:   o->xm2 = 0.0f;
chris@162:   o->ym1 = 0.0f;
chris@162:   o->ym2 = 0.0f;
chris@162: #endif
chris@162:   return 0;
chris@162: }
chris@162: 
chris@162: void sBiquad_k_onMessage(SignalBiquad_k *o, int letIn, const HvMessage *const m) {
chris@162:   if (msg_isFloat(m,0)) {
chris@162:     switch (letIn) {
chris@162:       case 1: o->b0 = msg_getFloat(m,0); break;
chris@162:       case 2: o->b1 = msg_getFloat(m,0); break;
chris@162:       case 3: o->b2 = msg_getFloat(m,0); break;
chris@162:       case 4: o->a1 = msg_getFloat(m,0); break;
chris@162:       case 5: o->a2 = msg_getFloat(m,0); break;
chris@162:       default: return;
chris@162:     }
chris@162:     sBiquad_k_updateCoefficients(o);
chris@162:   }
chris@162: }