[FFmpeg-devel] [PATCH 1/2] Add some dsputil functions useful for AAC decoder

[Mans Rullgard]

---
 libavcodec/dsputil.c |   83 ++++++++++++++++++++++++++++++++++++++++++++++++++
 libavcodec/dsputil.h |   50 ++++++++++++++++++++++++++++++
 2 files changed, 133 insertions(+), 0 deletions(-)

diff --git a/libavcodec/dsputil.c b/libavcodec/dsputil.c
index 773d47f..f32ba50 100644
--- a/libavcodec/dsputil.c
+++ b/libavcodec/dsputil.c
@@ -4089,6 +4089,79 @@ void ff_vector_fmul_window_c(float *dst, const float *src0, const float *src1, c
     }
 }
 
+static void vector_fmul_scalar_c(float *dst, const float *src, float mul,
+                                 int len)
+{
+    int i;
+    for (i = 0; i < len; i++)
+        dst[i] = src[i] * mul;
+}
+
+static void vector_fmul_sv_scalar_2_c(float *dst, const float *src,
+                                      const float **sv, float mul, int len)
+{
+    int i;
+    for (i = 0; i < len; i += 2, sv++) {
+        dst[i  ] = src[i  ] * sv[0][0] * mul;
+        dst[i+1] = src[i+1] * sv[0][1] * mul;
+    }
+}
+
+static void vector_fmul_sv_scalar_4_c(float *dst, const float *src,
+                                      const float **sv, float mul, int len)
+{
+    int i;
+    for (i = 0; i < len; i += 4, sv++) {
+        dst[i  ] = src[i  ] * sv[0][0] * mul;
+        dst[i+1] = src[i+1] * sv[0][1] * mul;
+        dst[i+2] = src[i+2] * sv[0][2] * mul;
+        dst[i+3] = src[i+3] * sv[0][3] * mul;
+    }
+}
+
+static void sv_fmul_scalar_2_c(float *dst, const float **sv, float mul,
+                               int len)
+{
+    int i;
+    for (i = 0; i < len; i += 2, sv++) {
+        dst[i  ] = sv[0][0] * mul;
+        dst[i+1] = sv[0][1] * mul;
+    }
+}
+
+static void sv_fmul_scalar_4_c(float *dst, const float **sv, float mul,
+                               int len)
+{
+    int i;
+    for (i = 0; i < len; i += 4, sv++) {
+        dst[i  ] = sv[0][0] * mul;
+        dst[i+1] = sv[0][1] * mul;
+        dst[i+2] = sv[0][2] * mul;
+        dst[i+3] = sv[0][3] * mul;
+    }
+}
+
+static void butterflies_float_c(float *v1, float *v2, int len)
+{
+    int i;
+    for (i = 0; i < len; i++) {
+        float t = v1[i] - v2[i];
+        v1[i] += v2[i];
+        v2[i] = t;
+    }
+}
+
+static float scalarproduct_float_c(const float *v1, const float *v2, int len)
+{
+    float p = 0.0;
+    int i;
+
+    for (i = 0; i < len; i++)
+        p += v1[i] * v2[i];
+
+    return p;
+}
+
 static void int32_to_float_fmul_scalar_c(float *dst, const int *src, float mul, int len){
     int i;
     for(i=0; i<len; i++)
@@ -4722,6 +4795,16 @@ void dsputil_init(DSPContext* c, AVCodecContext *avctx)
     c->add_int16 = add_int16_c;
     c->sub_int16 = sub_int16_c;
     c->scalarproduct_int16 = scalarproduct_int16_c;
+    c->scalarproduct_float = scalarproduct_float_c;
+    c->butterflies_float = butterflies_float_c;
+
+    c->vector_fmul_scalar = vector_fmul_scalar_c;
+
+    c->vector_fmul_sv_scalar[0] = vector_fmul_sv_scalar_2_c;
+    c->vector_fmul_sv_scalar[1] = vector_fmul_sv_scalar_4_c;
+
+    c->sv_fmul_scalar[0] = sv_fmul_scalar_2_c;
+    c->sv_fmul_scalar[1] = sv_fmul_scalar_4_c;
 
     c->shrink[0]= ff_img_copy_plane;
     c->shrink[1]= ff_shrink22;
diff --git a/libavcodec/dsputil.h b/libavcodec/dsputil.h
index 88a0ee8..eb3c208 100644
--- a/libavcodec/dsputil.h
+++ b/libavcodec/dsputil.h
@@ -397,6 +397,56 @@ typedef struct DSPContext {
     /* assume len is a multiple of 8, and arrays are 16-byte aligned */
     void (*int32_to_float_fmul_scalar)(float *dst, const int *src, float mul, int len);
     void (*vector_clipf)(float *dst /* align 16 */, const float *src /* align 16 */, float min, float max, int len /* align 16 */);
+    /**
+     * Multiply a vector of floats by a scalar float.  Source and
+     * destination vectors must overlap exactly or not at all.
+     * @param dst result vector, 16-byte aligned
+     * @param src input vector, 16-byte aligned
+     * @param mul scalar value
+     * @param len length of vector, multiple of 4
+     */
+    void (*vector_fmul_scalar)(float *dst, const float *src, float mul,
+                               int len);
+    /**
+     * Multiply a vector of floats by concatenated short vectors of
+     * floats and by a scalar float.  Source and destination vectors
+     * must overlap exactly or not at all.
+     * [0]: short vectors of length 2, 8-byte aligned
+     * [1]: short vectors of length 4, 16-byte aligned
+     * @param dst output vector, 16-byte aligned
+     * @param src input vector, 16-byte aligned
+     * @param sv  array of pointers to short vectors
+     * @param mul scalar value
+     * @param len number of elements in src and dst, multiple of 4
+     */
+    void (*vector_fmul_sv_scalar[2])(float *dst, const float *src,
+                                     const float **sv, float mul, int len);
+    /**
+     * Multiply short vectors of floats by a scalar float, store
+     * concatenated result.
+     * [0]: short vectors of length 2, 8-byte aligned
+     * [1]: short vectors of length 4, 16-byte aligned
+     * @param dst output vector, 16-byte aligned
+     * @param sv  array of pointers to short vectors
+     * @param mul scalar value
+     * @param len number of output elements, multiple of 4
+     */
+    void (*sv_fmul_scalar[2])(float *dst, const float **sv,
+                              float mul, int len);
+    /**
+     * Calculate the scalar product of two vectors of floats.
+     * @param v1  first vector, 16-byte aligned
+     * @param v2  second vector, 16-byte aligned
+     * @param len length of vectors, multiple of 4
+     */
+    float (*scalarproduct_float)(const float *v1, const float *v2, int len);
+    /**
+     * Calculate the sum and difference of two vectors of floats.
+     * @param v1  first input vector, sum output, 16-byte aligned
+     * @param v2  second input vector, difference output, 16-byte aligned
+     * @param len length of vectors, multiple of 4
+     */
+    void (*butterflies_float)(float *v1, float *v2, int len);
 
     /* C version: convert floats from the range [384.0,386.0] to ints in [-32768,32767]
      * simd versions: convert floats from [-32768.0,32767.0] without rescaling and arrays are 16byte aligned */
-- 
1.6.4.4