avcodec/aacenc_quantization: Deduplicate quantization functions

Up until now, there were four copies of
quantize_and_encode_band_cost_(ZERO|[SU]QUAD|[SU]PAIR|ESC|NONE|NOISE|STEREO)
(namely in aaccoder.o, aacenc_is.o, aacenc_ltp.o, aacenc_pred.o).
As 43b378a0d321e3d01f196cec95e13acfac80d464 says, this is meant to
enable more optimizations.

Yet neither GCC nor Clang performed such optimizations: The functions
in the aforementioned files are not optimized according to
the specifics of the calls in the respective file. Therefore
duplicating them is a waste of space; this commit therefore stops doing
so. The remaining copy is placed into aaccoder.c (which is the only
place where the "round to zero" variant of quantize_and_encode_band()
is used, so that this can be completely internal to aaccoder.c).

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

1 files changed, 223 insertions, 0 deletions

diff --git a/libavcodec/aaccoder.c b/libavcodec/aaccoder.c
index 2988247a15..e3b6b2f02c 100644
--- a/libavcodec/aaccoder.c
+++ b/libavcodec/aaccoder.c
@@ -62,6 +62,229 @@
 
 #include "libavcodec/aaccoder_trellis.h"
 
+typedef float (*quantize_and_encode_band_func)(struct AACEncContext *s, PutBitContext *pb,
+                                               const float *in, float *quant, const float *scaled,
+                                               int size, int scale_idx, int cb,
+                                               const float lambda, const float uplim,
+                                               int *bits, float *energy);
+
+/**
+ * Calculate rate distortion cost for quantizing with given codebook
+ *
+ * @return quantization distortion
+ */
+static av_always_inline float quantize_and_encode_band_cost_template(
+                                struct AACEncContext *s,
+                                PutBitContext *pb, const float *in, float *out,
+                                const float *scaled, int size, int scale_idx,
+                                int cb, const float lambda, const float uplim,
+                                int *bits, float *energy, int BT_ZERO, int BT_UNSIGNED,
+                                int BT_PAIR, int BT_ESC, int BT_NOISE, int BT_STEREO,
+                                const float ROUNDING)
+{
+    const int q_idx = POW_SF2_ZERO - scale_idx + SCALE_ONE_POS - SCALE_DIV_512;
+    const float Q   = ff_aac_pow2sf_tab [q_idx];
+    const float Q34 = ff_aac_pow34sf_tab[q_idx];
+    const float IQ  = ff_aac_pow2sf_tab [POW_SF2_ZERO + scale_idx - SCALE_ONE_POS + SCALE_DIV_512];
+    const float CLIPPED_ESCAPE = 165140.0f*IQ;
+    float cost = 0;
+    float qenergy = 0;
+    const int dim = BT_PAIR ? 2 : 4;
+    int resbits = 0;
+    int off;
+
+    if (BT_ZERO || BT_NOISE || BT_STEREO) {
+        for (int i = 0; i < size; i++)
+            cost += in[i]*in[i];
+        if (bits)
+            *bits = 0;
+        if (energy)
+            *energy = qenergy;
+        if (out) {
+            for (int i = 0; i < size; i += dim)
+                for (int j = 0; j < dim; j++)
+                    out[i+j] = 0.0f;
+        }
+        return cost * lambda;
+    }
+    if (!scaled) {
+        s->abs_pow34(s->scoefs, in, size);
+        scaled = s->scoefs;
+    }
+    s->quant_bands(s->qcoefs, in, scaled, size, !BT_UNSIGNED, aac_cb_maxval[cb], Q34, ROUNDING);
+    if (BT_UNSIGNED) {
+        off = 0;
+    } else {
+        off = aac_cb_maxval[cb];
+    }
+    for (int i = 0; i < size; i += dim) {
+        const float *vec;
+        int *quants = s->qcoefs + i;
+        int curidx = 0;
+        int curbits;
+        float quantized, rd = 0.0f;
+        for (int j = 0; j < dim; j++) {
+            curidx *= aac_cb_range[cb];
+            curidx += quants[j] + off;
+        }
+        curbits =  ff_aac_spectral_bits[cb-1][curidx];
+        vec     = &ff_aac_codebook_vectors[cb-1][curidx*dim];
+        if (BT_UNSIGNED) {
+            for (int j = 0; j < dim; j++) {
+                float t = fabsf(in[i+j]);
+                float di;
+                if (BT_ESC && vec[j] == 64.0f) { //FIXME: slow
+                    if (t >= CLIPPED_ESCAPE) {
+                        quantized = CLIPPED_ESCAPE;
+                        curbits += 21;
+                    } else {
+                        int c = av_clip_uintp2(quant(t, Q, ROUNDING), 13);
+                        quantized = c*cbrtf(c)*IQ;
+                        curbits += av_log2(c)*2 - 4 + 1;
+                    }
+                } else {
+                    quantized = vec[j]*IQ;
+                }
+                di = t - quantized;
+                if (out)
+                    out[i+j] = in[i+j] >= 0 ? quantized : -quantized;
+                if (vec[j] != 0.0f)
+                    curbits++;
+                qenergy += quantized*quantized;
+                rd += di*di;
+            }
+        } else {
+            for (int j = 0; j < dim; j++) {
+                quantized = vec[j]*IQ;
+                qenergy += quantized*quantized;
+                if (out)
+                    out[i+j] = quantized;
+                rd += (in[i+j] - quantized)*(in[i+j] - quantized);
+            }
+        }
+        cost    += rd * lambda + curbits;
+        resbits += curbits;
+        if (cost >= uplim)
+            return uplim;
+        if (pb) {
+            put_bits(pb, ff_aac_spectral_bits[cb-1][curidx], ff_aac_spectral_codes[cb-1][curidx]);
+            if (BT_UNSIGNED)
+                for (int j = 0; j < dim; j++)
+                    if (ff_aac_codebook_vectors[cb-1][curidx*dim+j] != 0.0f)
+                        put_bits(pb, 1, in[i+j] < 0.0f);
+            if (BT_ESC) {
+                for (int j = 0; j < 2; j++) {
+                    if (ff_aac_codebook_vectors[cb-1][curidx*2+j] == 64.0f) {
+                        int coef = av_clip_uintp2(quant(fabsf(in[i+j]), Q, ROUNDING), 13);
+                        int len = av_log2(coef);
+
+                        put_bits(pb, len - 4 + 1, (1 << (len - 4 + 1)) - 2);
+                        put_sbits(pb, len, coef);
+                    }
+                }
+            }
+        }
+    }
+
+    if (bits)
+        *bits = resbits;
+    if (energy)
+        *energy = qenergy;
+    return cost;
+}
+
+static inline float quantize_and_encode_band_cost_NONE(struct AACEncContext *s, PutBitContext *pb,
+                                                const float *in, float *quant, const float *scaled,
+                                                int size, int scale_idx, int cb,
+                                                const float lambda, const float uplim,
+                                                int *bits, float *energy) {
+    av_assert0(0);
+    return 0.0f;
+}
+
+#define QUANTIZE_AND_ENCODE_BAND_COST_FUNC(NAME, BT_ZERO, BT_UNSIGNED, BT_PAIR, BT_ESC, BT_NOISE, BT_STEREO, ROUNDING) \
+static float quantize_and_encode_band_cost_ ## NAME(                                         \
+                                struct AACEncContext *s,                                     \
+                                PutBitContext *pb, const float *in, float *quant,            \
+                                const float *scaled, int size, int scale_idx,                \
+                                int cb, const float lambda, const float uplim,               \
+                                int *bits, float *energy) {                                  \
+    return quantize_and_encode_band_cost_template(                                           \
+                                s, pb, in, quant, scaled, size, scale_idx,                   \
+                                BT_ESC ? ESC_BT : cb, lambda, uplim, bits, energy,           \
+                                BT_ZERO, BT_UNSIGNED, BT_PAIR, BT_ESC, BT_NOISE, BT_STEREO,  \
+                                ROUNDING);                                                   \
+}
+
+QUANTIZE_AND_ENCODE_BAND_COST_FUNC(ZERO,  1, 0, 0, 0, 0, 0, ROUND_STANDARD)
+QUANTIZE_AND_ENCODE_BAND_COST_FUNC(SQUAD, 0, 0, 0, 0, 0, 0, ROUND_STANDARD)
+QUANTIZE_AND_ENCODE_BAND_COST_FUNC(UQUAD, 0, 1, 0, 0, 0, 0, ROUND_STANDARD)
+QUANTIZE_AND_ENCODE_BAND_COST_FUNC(SPAIR, 0, 0, 1, 0, 0, 0, ROUND_STANDARD)
+QUANTIZE_AND_ENCODE_BAND_COST_FUNC(UPAIR, 0, 1, 1, 0, 0, 0, ROUND_STANDARD)
+QUANTIZE_AND_ENCODE_BAND_COST_FUNC(ESC,   0, 1, 1, 1, 0, 0, ROUND_STANDARD)
+QUANTIZE_AND_ENCODE_BAND_COST_FUNC(ESC_RTZ, 0, 1, 1, 1, 0, 0, ROUND_TO_ZERO)
+QUANTIZE_AND_ENCODE_BAND_COST_FUNC(NOISE, 0, 0, 0, 0, 1, 0, ROUND_STANDARD)
+QUANTIZE_AND_ENCODE_BAND_COST_FUNC(STEREO,0, 0, 0, 0, 0, 1, ROUND_STANDARD)
+
+static quantize_and_encode_band_func quantize_and_encode_band_cost_arr[] =
+{
+    quantize_and_encode_band_cost_ZERO,
+    quantize_and_encode_band_cost_SQUAD,
+    quantize_and_encode_band_cost_SQUAD,
+    quantize_and_encode_band_cost_UQUAD,
+    quantize_and_encode_band_cost_UQUAD,
+    quantize_and_encode_band_cost_SPAIR,
+    quantize_and_encode_band_cost_SPAIR,
+    quantize_and_encode_band_cost_UPAIR,
+    quantize_and_encode_band_cost_UPAIR,
+    quantize_and_encode_band_cost_UPAIR,
+    quantize_and_encode_band_cost_UPAIR,
+    quantize_and_encode_band_cost_ESC,
+    quantize_and_encode_band_cost_NONE,     /* CB 12 doesn't exist */
+    quantize_and_encode_band_cost_NOISE,
+    quantize_and_encode_band_cost_STEREO,
+    quantize_and_encode_band_cost_STEREO,
+};
+
+static quantize_and_encode_band_func quantize_and_encode_band_cost_rtz_arr[] =
+{
+    quantize_and_encode_band_cost_ZERO,
+    quantize_and_encode_band_cost_SQUAD,
+    quantize_and_encode_band_cost_SQUAD,
+    quantize_and_encode_band_cost_UQUAD,
+    quantize_and_encode_band_cost_UQUAD,
+    quantize_and_encode_band_cost_SPAIR,
+    quantize_and_encode_band_cost_SPAIR,
+    quantize_and_encode_band_cost_UPAIR,
+    quantize_and_encode_band_cost_UPAIR,
+    quantize_and_encode_band_cost_UPAIR,
+    quantize_and_encode_band_cost_UPAIR,
+    quantize_and_encode_band_cost_ESC_RTZ,
+    quantize_and_encode_band_cost_NONE,     /* CB 12 doesn't exist */
+    quantize_and_encode_band_cost_NOISE,
+    quantize_and_encode_band_cost_STEREO,
+    quantize_and_encode_band_cost_STEREO,
+};
+
+float ff_quantize_and_encode_band_cost(struct AACEncContext *s, PutBitContext *pb,
+                                       const float *in, float *quant, const float *scaled,
+                                       int size, int scale_idx, int cb,
+                                       const float lambda, const float uplim,
+                                       int *bits, float *energy)
+{
+    return quantize_and_encode_band_cost_arr[cb](s, pb, in, quant, scaled, size,
+                                                 scale_idx, cb, lambda, uplim,
+                                                 bits, energy);
+}
+
+static inline void quantize_and_encode_band(struct AACEncContext *s, PutBitContext *pb,
+                                            const float *in, float *out, int size, int scale_idx,
+                                            int cb, const float lambda, int rtz)
+{
+    (rtz ? quantize_and_encode_band_cost_rtz_arr : quantize_and_encode_band_cost_arr)[cb](s, pb, in, out, NULL, size, scale_idx, cb,
+                                     lambda, INFINITY, NULL, NULL);
+}
+
 /**
  * structure used in optimal codebook search
  */