# HG changeset patch
# User vitor
# Date 1278018252 0
# Node ID b4888704c11e0f0d1599b3c775e0e3a68a065ec6
# Parent  d5705b52b76ebcc02c36d3a6744211f068097f25
Altivec-optimized mp{1,2,3} windowing

diff -r d5705b52b76e -r b4888704c11e mpegaudio.h
--- a/mpegaudio.h	Thu Jul 01 18:49:44 2010 +0000
+++ b/mpegaudio.h	Thu Jul 01 21:04:12 2010 +0000
@@ -188,6 +188,7 @@
                          INTFLOAT sb_samples[SBLIMIT]);
 
 void ff_mpegaudiodec_init_mmx(MPADecodeContext *s);
+void ff_mpegaudiodec_init_altivec(MPADecodeContext *s);
 
 /* fast header check for resync */
 static inline int ff_mpa_check_header(uint32_t header){
diff -r d5705b52b76e -r b4888704c11e mpegaudiodec.c
--- a/mpegaudiodec.c	Thu Jul 01 18:49:44 2010 +0000
+++ b/mpegaudiodec.c	Thu Jul 01 21:04:12 2010 +0000
@@ -325,6 +325,8 @@
 #if HAVE_MMX
     ff_mpegaudiodec_init_mmx(s);
 #endif
+    if (HAVE_ALTIVEC && CONFIG_FLOAT) ff_mpegaudiodec_init_altivec(s);
+
     avctx->sample_fmt= OUT_FMT;
     s->error_recognition= avctx->error_recognition;
 
diff -r d5705b52b76e -r b4888704c11e ppc/Makefile
--- a/ppc/Makefile	Thu Jul 01 18:49:44 2010 +0000
+++ b/ppc/Makefile	Thu Jul 01 21:04:12 2010 +0000
@@ -7,6 +7,12 @@
 ALTIVEC-OBJS-$(CONFIG_VP6_DECODER)     += ppc/vp3dsp_altivec.o
 ALTIVEC-OBJS-$(CONFIG_VP8_DECODER)     += ppc/vp8dsp_altivec.o
 
+ALTIVEC-OBJS-$(CONFIG_MP1FLOAT_DECODER)    += ppc/mpegaudiodec_altivec.o
+ALTIVEC-OBJS-$(CONFIG_MP2FLOAT_DECODER)    += ppc/mpegaudiodec_altivec.o
+ALTIVEC-OBJS-$(CONFIG_MP3FLOAT_DECODER)    += ppc/mpegaudiodec_altivec.o
+ALTIVEC-OBJS-$(CONFIG_MP3ON4FLOAT_DECODER) += ppc/mpegaudiodec_altivec.o
+ALTIVEC-OBJS-$(CONFIG_MP3ADUFLOAT_DECODER) += ppc/mpegaudiodec_altivec.o
+
 OBJS-$(HAVE_ALTIVEC)                   += ppc/check_altivec.o           \
                                           ppc/dsputil_altivec.o         \
                                           ppc/fdct_altivec.o            \
diff -r d5705b52b76e -r b4888704c11e ppc/mpegaudiodec_altivec.c
--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/ppc/mpegaudiodec_altivec.c	Thu Jul 01 21:04:12 2010 +0000
@@ -0,0 +1,130 @@
+/*
+ * Altivec optimized MP3 decoding functions
+ * Copyright (c) 2010 Vitor Sessak
+ *
+ * This file is part of FFmpeg.
+ *
+ * FFmpeg 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.
+ *
+ * FFmpeg 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 FFmpeg; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+
+#include "dsputil_altivec.h"
+#include "util_altivec.h"
+
+#define CONFIG_FLOAT 1
+#include "libavcodec/mpegaudio.h"
+
+#define MACS(rt, ra, rb) rt+=(ra)*(rb)
+#define MLSS(rt, ra, rb) rt-=(ra)*(rb)
+
+#define SUM8(op, sum, w, p)               \
+{                                         \
+    op(sum, (w)[0 * 64], (p)[0 * 64]);    \
+    op(sum, (w)[1 * 64], (p)[1 * 64]);    \
+    op(sum, (w)[2 * 64], (p)[2 * 64]);    \
+    op(sum, (w)[3 * 64], (p)[3 * 64]);    \
+    op(sum, (w)[4 * 64], (p)[4 * 64]);    \
+    op(sum, (w)[5 * 64], (p)[5 * 64]);    \
+    op(sum, (w)[6 * 64], (p)[6 * 64]);    \
+    op(sum, (w)[7 * 64], (p)[7 * 64]);    \
+}
+
+static void apply_window(const float *buf, const float *win1,
+                         const float *win2, float *sum1, float *sum2, int len)
+{
+    const vector float *win1a = (const vector float *) win1;
+    const vector float *win2a = (const vector float *) win2;
+    const vector float *bufa  = (const vector float *) buf;
+    vector float *sum1a = (vector float *) sum1;
+    vector float *sum2a = (vector float *) sum2;
+    vector float av_uninit(v0), av_uninit(v4);
+    vector float v1, v2, v3;
+
+    len = len >> 2;
+
+#define MULT(a, b)                         \
+    {                                      \
+        v1 = vec_ld(a, win1a);             \
+        v2 = vec_ld(b, win2a);             \
+        v3 = vec_ld(a, bufa);              \
+        v0 = vec_madd(v3, v1, v0);         \
+        v4 = vec_madd(v2, v3, v4);         \
+    }
+
+    while (len--) {
+        v0 = vec_xor(v0, v0);
+        v4 = vec_xor(v4, v4);
+
+        MULT(   0,   0);
+        MULT( 256,  64);
+        MULT( 512, 128);
+        MULT( 768, 192);
+        MULT(1024, 256);
+        MULT(1280, 320);
+        MULT(1536, 384);
+        MULT(1792, 448);
+
+        vec_st(v0, 0, sum1a);
+        vec_st(v4, 0, sum2a);
+        sum1a++;
+        sum2a++;
+        win1a++;
+        win2a++;
+        bufa++;
+    }
+}
+
+static void apply_window_mp3(float *in, float *win, int *unused, float *out,
+                             int incr)
+{
+    LOCAL_ALIGNED_16(float, suma, [17]);
+    LOCAL_ALIGNED_16(float, sumb, [17]);
+    LOCAL_ALIGNED_16(float, sumc, [17]);
+    LOCAL_ALIGNED_16(float, sumd, [17]);
+
+    float sum;
+    int j;
+    float *out2 = out + 32 * incr;
+
+    /* copy to avoid wrap */
+    memcpy(in + 512, in, 32 * sizeof(*in));
+
+    apply_window(in + 16, win     , win + 512, suma, sumc, 16);
+    apply_window(in + 32, win + 48, win + 640, sumb, sumd, 16);
+
+    SUM8(MLSS, suma[0], win + 32, in + 48);
+
+    sumc[ 0] = 0;
+    sumb[16] = 0;
+    sumd[16] = 0;
+
+    out[0  ]  = suma[   0];
+    out += incr;
+    out2 -= incr;
+    for(j=1;j<16;j++) {
+        *out  =  suma[   j] - sumd[16-j];
+        *out2 = -sumb[16-j] - sumc[   j];
+        out  += incr;
+        out2 -= incr;
+    }
+
+    sum = 0;
+    SUM8(MLSS, sum, win + 16 + 32, in + 32);
+    *out = sum;
+}
+
+void ff_mpegaudiodec_init_altivec(MPADecodeContext *s)
+{
+    s->apply_window_mp3 = apply_window_mp3;
+}