dca: NEON optimised high freq VQ decoding

Signed-off-by: Mans Rullgard <mans@mansr.com>

libavcodec/arm/dca.h

+/*
+ * Copyright (c) 2011 Mans Rullgard <mans@mansr.com>
+ *
+ * This file is part of Libav.
+ *
+ * Libav is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License as published by the Free Software Foundation; either
+ * version 2.1 of the License, or (at your option) any later version.
+ *
+ * Libav is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public
+ * License along with Libav; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+
+#ifndef AVCODEC_ARM_DCA_H
+#define AVCODEC_ARM_DCA_H
+
+#include <stdint.h>
+#include "config.h"
+
+#if HAVE_NEON && HAVE_INLINE_ASM
+
+#define int8x8_fmul_int32 int8x8_fmul_int32
+static inline void int8x8_fmul_int32(float *dst, const int8_t *src, int scale)
+{
+    __asm__ ("vcvt.f32.s32 %2,  %2,  #4         \n"
+             "vld1.8       {d0},     [%1,:64]   \n"
+             "vmovl.s8     q0,  d0              \n"
+             "vmovl.s16    q1,  d1              \n"
+             "vmovl.s16    q0,  d0              \n"
+             "vcvt.f32.s32 q0,  q0              \n"
+             "vcvt.f32.s32 q1,  q1              \n"
+             "vmul.f32     q0,  q0,  %y2        \n"
+             "vmul.f32     q1,  q1,  %y2        \n"
+             "vst1.32      {q0-q1},  [%m0,:128] \n"
+             : "=Um"(*(float (*)[8])dst)
+             : "r"(src), "x"(scale)
+             : "d0", "d1", "d2", "d3");
+}
+
+#endif
+
+#endif /* AVCODEC_ARM_DCA_H */

libavcodec/dca.c

@@ -42,6 +42,10 @@
 #include "dcadsp.h"
 #include "fmtconvert.h"
 
+#if ARCH_ARM
+#   include "arm/dca.h"
+#endif
+
 //#define TRACE
 
 #define DCA_PRIM_CHANNELS_MAX (7)
@@ -320,7 +324,7 @@ typedef struct {
     int lfe_scale_factor;
 
     /* Subband samples history (for ADPCM) */
-    float subband_samples_hist[DCA_PRIM_CHANNELS_MAX][DCA_SUBBANDS][4];
+    DECLARE_ALIGNED(16, float, subband_samples_hist)[DCA_PRIM_CHANNELS_MAX][DCA_SUBBANDS][4];
     DECLARE_ALIGNED(32, float, subband_fir_hist)[DCA_PRIM_CHANNELS_MAX][512];
     DECLARE_ALIGNED(32, float, subband_fir_noidea)[DCA_PRIM_CHANNELS_MAX][32];
     int hist_index[DCA_PRIM_CHANNELS_MAX];
@@ -1057,6 +1061,16 @@ static int decode_blockcode(int code, int levels, int *values)
 static const uint8_t abits_sizes[7] = { 7, 10, 12, 13, 15, 17, 19 };
 static const uint8_t abits_levels[7] = { 3, 5, 7, 9, 13, 17, 25 };
 
+#ifndef int8x8_fmul_int32
+static inline void int8x8_fmul_int32(float *dst, const int8_t *src, int scale)
+{
+    float fscale = scale / 16.0;
+    int i;
+    for (i = 0; i < 8; i++)
+        dst[i] = src[i] * fscale;
+}
+#endif
+
 static int dca_subsubframe(DCAContext * s, int base_channel, int block_index)
 {
     int k, l;
@@ -1161,19 +1175,16 @@ static int dca_subsubframe(DCAContext * s, int base_channel, int block_index)
         for (l = s->vq_start_subband[k]; l < s->subband_activity[k]; l++) {
             /* 1 vector -> 32 samples but we only need the 8 samples
              * for this subsubframe. */
-            int m;
+            int hfvq = s->high_freq_vq[k][l];
 
             if (!s->debug_flag & 0x01) {
                 av_log(s->avctx, AV_LOG_DEBUG, "Stream with high frequencies VQ coding\n");
                 s->debug_flag |= 0x01;
             }
 
-            for (m = 0; m < 8; m++) {
-                subband_samples[k][l][m] =
-                    high_freq_vq[s->high_freq_vq[k][l]][subsubframe * 8 +
-                                                        m]
-                    * (float) s->scale_factor[k][l][0] / 16.0;
-            }
+            int8x8_fmul_int32(subband_samples[k][l],
+                              &high_freq_vq[hfvq][subsubframe * 8],
+                              s->scale_factor[k][l][0]);
         }
     }
 

libavcodec/dcadata.h

@@ -4224,7 +4224,7 @@ static const float lossless_quant_d[32] = {
 
 /* Vector quantization tables */
 
-static const int8_t high_freq_vq[1024][32] =
+DECLARE_ALIGNED(8, static const int8_t, high_freq_vq)[1024][32] =
 {
   {  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,
      0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0 },