xsubdec: Convert to the new bitstream reader

[ffmpeg] / libavcodec / sbrdsp.c

diff --git a/libavcodec/sbrdsp.c b/libavcodec/sbrdsp.c
index 7be962ed77cc0e65b8423b7c121e0c70966bb170..029433225f7463e9ba3375f0870bc58fac8f9b5d 100644 (file)
--- a/libavcodec/sbrdsp.c
+++ b/libavcodec/sbrdsp.c
@@ -20,7 +20,9 @@
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  */
 
+#include "config.h"
 #include "libavutil/attributes.h"
+#include "libavutil/intfloat.h"
 #include "sbrdsp.h"
 
 static void sbr_sum64x5_c(float *z)
@@ -34,48 +36,67 @@ static void sbr_sum64x5_c(float *z)
 
 static float sbr_sum_square_c(float (*x)[2], int n)
 {
-    float sum = 0.0f;
+    float sum0 = 0.0f, sum1 = 0.0f;
     int i;
 
-    for (i = 0; i < n; i++)
-        sum += x[i][0] * x[i][0] + x[i][1] * x[i][1];
+    for (i = 0; i < n; i += 2)
+    {
+        sum0 += x[i + 0][0] * x[i + 0][0];
+        sum1 += x[i + 0][1] * x[i + 0][1];
+        sum0 += x[i + 1][0] * x[i + 1][0];
+        sum1 += x[i + 1][1] * x[i + 1][1];
+    }
 
-    return sum;
+    return sum0 + sum1;
 }
 
 static void sbr_neg_odd_64_c(float *x)
 {
+    union av_intfloat32 *xi = (union av_intfloat32*) x;
     int i;
-    for (i = 1; i < 64; i += 2)
-        x[i] = -x[i];
+    for (i = 1; i < 64; i += 4) {
+        xi[i + 0].i ^= 1U << 31;
+        xi[i + 2].i ^= 1U << 31;
+    }
 }
 
 static void sbr_qmf_pre_shuffle_c(float *z)
 {
+    union av_intfloat32 *zi = (union av_intfloat32*) z;
     int k;
-    z[64] = z[0];
-    z[65] = z[1];
-    for (k = 1; k < 32; k++) {
-        z[64+2*k  ] = -z[64 - k];
-        z[64+2*k+1] =  z[ k + 1];
+    zi[64].i = zi[0].i;
+    zi[65].i = zi[1].i;
+    for (k = 1; k < 31; k += 2) {
+        zi[64 + 2 * k + 0].i = zi[64 - k].i ^ (1U << 31);
+        zi[64 + 2 * k + 1].i = zi[ k + 1].i;
+        zi[64 + 2 * k + 2].i = zi[63 - k].i ^ (1U << 31);
+        zi[64 + 2 * k + 3].i = zi[ k + 2].i;
     }
+    zi[64 + 2 * 31 + 0].i = zi[64 - 31].i ^ (1U << 31);
+    zi[64 + 2 * 31 + 1].i = zi[31 +  1].i;
 }
 
 static void sbr_qmf_post_shuffle_c(float W[32][2], const float *z)
 {
+    const union av_intfloat32 *zi = (const union av_intfloat32*) z;
+    union av_intfloat32 *Wi       = (union av_intfloat32*) W;
     int k;
-    for (k = 0; k < 32; k++) {
-        W[k][0] = -z[63-k];
-        W[k][1] = z[k];
+    for (k = 0; k < 32; k += 2) {
+        Wi[2 * k + 0].i = zi[63 - k].i ^ (1U << 31);
+        Wi[2 * k + 1].i = zi[ k + 0].i;
+        Wi[2 * k + 2].i = zi[62 - k].i ^ (1U << 31);
+        Wi[2 * k + 3].i = zi[ k + 1].i;
     }
 }
 
 static void sbr_qmf_deint_neg_c(float *v, const float *src)
 {
+    const union av_intfloat32 *si = (const union av_intfloat32*)src;
+    union av_intfloat32 *vi = (union av_intfloat32*)v;
     int i;
     for (i = 0; i < 32; i++) {
-        v[     i] =  src[63 - 2*i    ];
-        v[63 - i] = -src[63 - 2*i - 1];
+        vi[     i].i = si[63 - 2 * i    ].i;
+        vi[63 - i].i = si[63 - 2 * i - 1].i ^ (1U << 31);
     }
 }
 
@@ -88,6 +109,11 @@ static void sbr_qmf_deint_bfly_c(float *v, const float *src0, const float *src1)
     }
 }
 
+
+#if 0
+    /* This code is slower because it multiplies memory accesses.
+     * It is left for educational purposes and because it may offer
+     * a better reference for writing arch-specific DSP functions. */
 static av_always_inline void autocorrelate(const float x[40][2],
                                            float phi[3][2][2], int lag)
 {
@@ -120,6 +146,30 @@ static void sbr_autocorrelate_c(const float x[40][2], float phi[3][2][2])
     autocorrelate(x, phi, 1);
     autocorrelate(x, phi, 2);
 }
+#else
+static void sbr_autocorrelate_c(const float x[40][2], float phi[3][2][2])
+{
+    float real_sum2 = x[0][0] * x[2][0] + x[0][1] * x[2][1];
+    float imag_sum2 = x[0][0] * x[2][1] - x[0][1] * x[2][0];
+    float real_sum1 = 0.0f, imag_sum1 = 0.0f, real_sum0 = 0.0f;
+    int   i;
+    for (i = 1; i < 38; i++) {
+        real_sum0 += x[i][0] * x[i    ][0] + x[i][1] * x[i    ][1];
+        real_sum1 += x[i][0] * x[i + 1][0] + x[i][1] * x[i + 1][1];
+        imag_sum1 += x[i][0] * x[i + 1][1] - x[i][1] * x[i + 1][0];
+        real_sum2 += x[i][0] * x[i + 2][0] + x[i][1] * x[i + 2][1];
+        imag_sum2 += x[i][0] * x[i + 2][1] - x[i][1] * x[i + 2][0];
+    }
+    phi[2 - 2][1][0] = real_sum2;
+    phi[2 - 2][1][1] = imag_sum2;
+    phi[2    ][1][0] = real_sum0 + x[ 0][0] * x[ 0][0] + x[ 0][1] * x[ 0][1];
+    phi[1    ][0][0] = real_sum0 + x[38][0] * x[38][0] + x[38][1] * x[38][1];
+    phi[2 - 1][1][0] = real_sum1 + x[ 0][0] * x[ 1][0] + x[ 0][1] * x[ 1][1];
+    phi[2 - 1][1][1] = imag_sum1 + x[ 0][0] * x[ 1][1] - x[ 0][1] * x[ 1][0];
+    phi[0    ][0][0] = real_sum1 + x[38][0] * x[39][0] + x[38][1] * x[39][1];
+    phi[0    ][0][1] = imag_sum1 + x[38][0] * x[39][1] - x[38][1] * x[39][0];
+}
+#endif
 
 static void sbr_hf_gen_c(float (*X_high)[2], const float (*X_low)[2],
                          const float alpha0[2], const float alpha1[2],
@@ -150,7 +200,7 @@ static void sbr_hf_gen_c(float (*X_high)[2], const float (*X_low)[2],
 }
 
 static void sbr_hf_g_filt_c(float (*Y)[2], const float (*X_high)[40][2],
-                            const float *g_filt, int m_max, int ixh)
+                            const float *g_filt, int m_max, intptr_t ixh)
 {
     int m;
 
@@ -234,4 +284,9 @@ av_cold void ff_sbrdsp_init(SBRDSPContext *s)
     s->hf_apply_noise[1] = sbr_hf_apply_noise_1;
     s->hf_apply_noise[2] = sbr_hf_apply_noise_2;
     s->hf_apply_noise[3] = sbr_hf_apply_noise_3;
+
+    if (ARCH_ARM)
+        ff_sbrdsp_init_arm(s);
+    if (ARCH_X86)
+        ff_sbrdsp_init_x86(s);
 }