lavc: split snow and dirac DWTs

There is only about 4 lines of common code, so it alot cleaner
when seperated.

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

15 files changed, 744 insertions, 660 deletions

diff --git a/libavcodec/Makefile b/libavcodec/Makefile
index 52e6fce6fe..0925e9606b 100644
--- a/libavcodec/Makefile
+++ b/libavcodec/Makefile
@@ -41,7 +41,6 @@ OBJS-$(CONFIG_AANDCTTABLES)            += aandcttab.o
 OBJS-$(CONFIG_AC3DSP)                  += ac3dsp.o
 OBJS-$(CONFIG_CRYSTALHD)               += crystalhd.o
 OBJS-$(CONFIG_DCT)                     += dct.o dct32_fixed.o dct32_float.o
-OBJS-$(CONFIG_DWT)                     += dwt.o snow.o
 OBJS-$(CONFIG_DXVA2)                   += dxva2.o
 OBJS-$(CONFIG_ENCODERS)                += faandct.o jfdctfst.o jfdctint.o
 OBJS-$(CONFIG_ERROR_RESILIENCE)        += error_resilience.o
@@ -158,7 +157,7 @@ OBJS-$(CONFIG_DCA_DECODER)             += dcadec.o dca.o dcadsp.o      \
                                           dca_parser.o synth_filter.o
 OBJS-$(CONFIG_DCA_ENCODER)             += dcaenc.o dca.o
 OBJS-$(CONFIG_DIRAC_DECODER)           += diracdec.o dirac.o diracdsp.o \
-                                          dirac_arith.o mpeg12data.o dwt.o
+                                          dirac_arith.o mpeg12data.o dirac_dwt.o
 OBJS-$(CONFIG_DFA_DECODER)             += dfa.o
 OBJS-$(CONFIG_DNXHD_DECODER)           += dnxhddec.o dnxhddata.o
 OBJS-$(CONFIG_DNXHD_ENCODER)           += dnxhdenc.o dnxhddata.o
@@ -396,8 +395,8 @@ OBJS-$(CONFIG_SIPR_DECODER)            += sipr.o acelp_pitch_delay.o \
 OBJS-$(CONFIG_SMACKAUD_DECODER)        += smacker.o
 OBJS-$(CONFIG_SMACKER_DECODER)         += smacker.o
 OBJS-$(CONFIG_SMC_DECODER)             += smc.o
-OBJS-$(CONFIG_SNOW_DECODER)            += snowdec.o snow.o
-OBJS-$(CONFIG_SNOW_ENCODER)            += snowenc.o snow.o              \
+OBJS-$(CONFIG_SNOW_DECODER)            += snowdec.o snow.o snow_dwt.o
+OBJS-$(CONFIG_SNOW_ENCODER)            += snowenc.o snow.o snow_dwt.o             \
                                           h263.o ituh263enc.o
 OBJS-$(CONFIG_SOL_DPCM_DECODER)        += dpcm.o
 OBJS-$(CONFIG_SONIC_DECODER)           += sonic.o
diff --git a/libavcodec/dirac_dwt.c b/libavcodec/dirac_dwt.c
new file mode 100644
index 0000000000..a2848f01ac
--- /dev/null
+++ b/libavcodec/dirac_dwt.c
@@ -0,0 +1,585 @@
+/*
+ * Copyright (C) 2004-2010 Michael Niedermayer <michaelni@gmx.at>
+ * Copyright (C) 2008 David Conrad
+ *
+ * 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 "libavutil/attributes.h"
+#include "libavutil/avassert.h"
+#include "libavutil/common.h"
+#include "dsputil.h"
+#include "dirac_dwt.h"
+#include "libavcodec/x86/dirac_dwt.h"
+
+
+static inline int mirror(int v, int m)
+{
+    while ((unsigned)v > (unsigned)m) {
+        v = -v;
+        if (v < 0)
+            v += 2 * m;
+    }
+    return v;
+}
+
+static void vertical_compose53iL0(IDWTELEM *b0, IDWTELEM *b1, IDWTELEM *b2,
+                                  int width)
+{
+    int i;
+
+    for (i = 0; i < width; i++)
+        b1[i] -= (b0[i] + b2[i] + 2) >> 2;
+}
+
+
+static av_always_inline
+void interleave(IDWTELEM *dst, IDWTELEM *src0, IDWTELEM *src1, int w2, int add, int shift)
+{
+    int i;
+    for (i = 0; i < w2; i++) {
+        dst[2*i  ] = (src0[i] + add) >> shift;
+        dst[2*i+1] = (src1[i] + add) >> shift;
+    }
+}
+
+static void horizontal_compose_dirac53i(IDWTELEM *b, IDWTELEM *temp, int w)
+{
+    const int w2 = w >> 1;
+    int x;
+
+    temp[0] = COMPOSE_53iL0(b[w2], b[0], b[w2]);
+    for (x = 1; x < w2; x++) {
+        temp[x     ] = COMPOSE_53iL0     (b[x+w2-1], b[x     ], b[x+w2]);
+        temp[x+w2-1] = COMPOSE_DIRAC53iH0(temp[x-1], b[x+w2-1], temp[x]);
+    }
+    temp[w-1] = COMPOSE_DIRAC53iH0(temp[w2-1], b[w-1], temp[w2-1]);
+
+    interleave(b, temp, temp+w2, w2, 1, 1);
+}
+
+static void horizontal_compose_dd97i(IDWTELEM *b, IDWTELEM *tmp, int w)
+{
+    const int w2 = w >> 1;
+    int x;
+
+    tmp[0] = COMPOSE_53iL0(b[w2], b[0], b[w2]);
+    for (x = 1; x < w2; x++)
+        tmp[x] = COMPOSE_53iL0(b[x+w2-1], b[x], b[x+w2]);
+
+    // extend the edges
+    tmp[-1]   = tmp[0];
+    tmp[w2+1] = tmp[w2] = tmp[w2-1];
+
+    for (x = 0; x < w2; x++) {
+        b[2*x  ] = (tmp[x] + 1)>>1;
+        b[2*x+1] = (COMPOSE_DD97iH0(tmp[x-1], tmp[x], b[x+w2], tmp[x+1], tmp[x+2]) + 1)>>1;
+    }
+}
+
+static void horizontal_compose_dd137i(IDWTELEM *b, IDWTELEM *tmp, int w)
+{
+    const int w2 = w >> 1;
+    int x;
+
+    tmp[0] = COMPOSE_DD137iL0(b[w2], b[w2], b[0], b[w2  ], b[w2+1]);
+    tmp[1] = COMPOSE_DD137iL0(b[w2], b[w2], b[1], b[w2+1], b[w2+2]);
+    for (x = 2; x < w2-1; x++)
+        tmp[x] = COMPOSE_DD137iL0(b[x+w2-2], b[x+w2-1], b[x], b[x+w2], b[x+w2+1]);
+    tmp[w2-1] = COMPOSE_DD137iL0(b[w-3], b[w-2], b[w2-1], b[w-1], b[w-1]);
+
+    // extend the edges
+    tmp[-1]   = tmp[0];
+    tmp[w2+1] = tmp[w2] = tmp[w2-1];
+
+    for (x = 0; x < w2; x++) {
+        b[2*x  ] = (tmp[x] + 1)>>1;
+        b[2*x+1] = (COMPOSE_DD97iH0(tmp[x-1], tmp[x], b[x+w2], tmp[x+1], tmp[x+2]) + 1)>>1;
+    }
+}
+
+static av_always_inline
+void horizontal_compose_haari(IDWTELEM *b, IDWTELEM *temp, int w, int shift)
+{
+    const int w2 = w >> 1;
+    int x;
+
+    for (x = 0; x < w2; x++) {
+        temp[x   ] = COMPOSE_HAARiL0(b[x   ], b[x+w2]);
+        temp[x+w2] = COMPOSE_HAARiH0(b[x+w2], temp[x]);
+    }
+
+    interleave(b, temp, temp+w2, w2, shift, shift);
+}
+
+static void horizontal_compose_haar0i(IDWTELEM *b, IDWTELEM *temp, int w)
+{
+    horizontal_compose_haari(b, temp, w, 0);
+}
+
+static void horizontal_compose_haar1i(IDWTELEM *b, IDWTELEM *temp, int w)
+{
+    horizontal_compose_haari(b, temp, w, 1);
+}
+
+static void horizontal_compose_fidelityi(IDWTELEM *b, IDWTELEM *tmp, int w)
+{
+    const int w2 = w >> 1;
+    int i, x;
+    IDWTELEM v[8];
+
+    for (x = 0; x < w2; x++) {
+        for (i = 0; i < 8; i++)
+            v[i] = b[av_clip(x-3+i, 0, w2-1)];
+        tmp[x] = COMPOSE_FIDELITYiH0(v[0], v[1], v[2], v[3], b[x+w2], v[4], v[5], v[6], v[7]);
+    }
+
+    for (x = 0; x < w2; x++) {
+        for (i = 0; i < 8; i++)
+            v[i] = tmp[av_clip(x-4+i, 0, w2-1)];
+        tmp[x+w2] = COMPOSE_FIDELITYiL0(v[0], v[1], v[2], v[3], b[x], v[4], v[5], v[6], v[7]);
+    }
+
+    interleave(b, tmp+w2, tmp, w2, 0, 0);
+}
+
+static void horizontal_compose_daub97i(IDWTELEM *b, IDWTELEM *temp, int w)
+{
+    const int w2 = w >> 1;
+    int x, b0, b1, b2;
+
+    temp[0] = COMPOSE_DAUB97iL1(b[w2], b[0], b[w2]);
+    for (x = 1; x < w2; x++) {
+        temp[x     ] = COMPOSE_DAUB97iL1(b[x+w2-1], b[x     ], b[x+w2]);
+        temp[x+w2-1] = COMPOSE_DAUB97iH1(temp[x-1], b[x+w2-1], temp[x]);
+    }
+    temp[w-1] = COMPOSE_DAUB97iH1(temp[w2-1], b[w-1], temp[w2-1]);
+
+    // second stage combined with interleave and shift
+    b0 = b2 = COMPOSE_DAUB97iL0(temp[w2], temp[0], temp[w2]);
+    b[0] = (b0 + 1) >> 1;
+    for (x = 1; x < w2; x++) {
+        b2 = COMPOSE_DAUB97iL0(temp[x+w2-1], temp[x     ], temp[x+w2]);
+        b1 = COMPOSE_DAUB97iH0(          b0, temp[x+w2-1], b2        );
+        b[2*x-1] = (b1 + 1) >> 1;
+        b[2*x  ] = (b2 + 1) >> 1;
+        b0 = b2;
+    }
+    b[w-1] = (COMPOSE_DAUB97iH0(b2, temp[w-1], b2) + 1) >> 1;
+}
+
+static void vertical_compose_dirac53iH0(IDWTELEM *b0, IDWTELEM *b1, IDWTELEM *b2, int width)
+{
+    int i;
+
+    for(i=0; i<width; i++){
+        b1[i] = COMPOSE_DIRAC53iH0(b0[i], b1[i], b2[i]);
+    }
+}
+
+static void vertical_compose_dd97iH0(IDWTELEM *b0, IDWTELEM *b1, IDWTELEM *b2,
+                                  IDWTELEM *b3, IDWTELEM *b4, int width)
+{
+    int i;
+
+    for(i=0; i<width; i++){
+        b2[i] = COMPOSE_DD97iH0(b0[i], b1[i], b2[i], b3[i], b4[i]);
+    }
+}
+
+static void vertical_compose_dd137iL0(IDWTELEM *b0, IDWTELEM *b1, IDWTELEM *b2,
+                                      IDWTELEM *b3, IDWTELEM *b4, int width)
+{
+    int i;
+
+    for(i=0; i<width; i++){
+        b2[i] = COMPOSE_DD137iL0(b0[i], b1[i], b2[i], b3[i], b4[i]);
+    }
+}
+
+static void vertical_compose_haar(IDWTELEM *b0, IDWTELEM *b1, int width)
+{
+    int i;
+
+    for (i = 0; i < width; i++) {
+        b0[i] = COMPOSE_HAARiL0(b0[i], b1[i]);
+        b1[i] = COMPOSE_HAARiH0(b1[i], b0[i]);
+    }
+}
+
+static void vertical_compose_fidelityiH0(IDWTELEM *dst, IDWTELEM *b[8], int width)
+{
+    int i;
+
+    for(i=0; i<width; i++){
+        dst[i] = COMPOSE_FIDELITYiH0(b[0][i], b[1][i], b[2][i], b[3][i], dst[i], b[4][i], b[5][i], b[6][i], b[7][i]);
+    }
+}
+
+static void vertical_compose_fidelityiL0(IDWTELEM *dst, IDWTELEM *b[8], int width)
+{
+    int i;
+
+    for(i=0; i<width; i++){
+        dst[i] = COMPOSE_FIDELITYiL0(b[0][i], b[1][i], b[2][i], b[3][i], dst[i], b[4][i], b[5][i], b[6][i], b[7][i]);
+    }
+}
+
+static void vertical_compose_daub97iH0(IDWTELEM *b0, IDWTELEM *b1, IDWTELEM *b2, int width)
+{
+    int i;
+
+    for(i=0; i<width; i++){
+        b1[i] = COMPOSE_DAUB97iH0(b0[i], b1[i], b2[i]);
+    }
+}
+
+static void vertical_compose_daub97iH1(IDWTELEM *b0, IDWTELEM *b1, IDWTELEM *b2, int width)
+{
+    int i;
+
+    for(i=0; i<width; i++){
+        b1[i] = COMPOSE_DAUB97iH1(b0[i], b1[i], b2[i]);
+    }
+}
+
+static void vertical_compose_daub97iL0(IDWTELEM *b0, IDWTELEM *b1, IDWTELEM *b2, int width)
+{
+    int i;
+
+    for(i=0; i<width; i++){
+        b1[i] = COMPOSE_DAUB97iL0(b0[i], b1[i], b2[i]);
+    }
+}
+
+static void vertical_compose_daub97iL1(IDWTELEM *b0, IDWTELEM *b1, IDWTELEM *b2, int width)
+{
+    int i;
+
+    for(i=0; i<width; i++){
+        b1[i] = COMPOSE_DAUB97iL1(b0[i], b1[i], b2[i]);
+    }
+}
+
+
+static void spatial_compose_dd97i_dy(DWTContext *d, int level, int width, int height, int stride)
+{
+    vertical_compose_3tap vertical_compose_l0 = (void*)d->vertical_compose_l0;
+    vertical_compose_5tap vertical_compose_h0 = (void*)d->vertical_compose_h0;
+    DWTCompose *cs = d->cs + level;
+
+    int i, y = cs->y;
+    IDWTELEM *b[8];
+    for (i = 0; i < 6; i++)
+        b[i] = cs->b[i];
+    b[6] = d->buffer + av_clip(y+5, 0, height-2)*stride;
+    b[7] = d->buffer + av_clip(y+6, 1, height-1)*stride;
+
+        if(y+5<(unsigned)height) vertical_compose_l0(      b[5], b[6], b[7],       width);
+        if(y+1<(unsigned)height) vertical_compose_h0(b[0], b[2], b[3], b[4], b[6], width);
+
+        if(y-1<(unsigned)height) d->horizontal_compose(b[0], d->temp, width);
+        if(y+0<(unsigned)height) d->horizontal_compose(b[1], d->temp, width);
+
+    for (i = 0; i < 6; i++)
+        cs->b[i] = b[i+2];
+    cs->y += 2;
+}
+
+static void spatial_compose_dirac53i_dy(DWTContext *d, int level, int width, int height, int stride)
+{
+    vertical_compose_3tap vertical_compose_l0 = (void*)d->vertical_compose_l0;
+    vertical_compose_3tap vertical_compose_h0 = (void*)d->vertical_compose_h0;
+    DWTCompose *cs = d->cs + level;
+
+    int y= cs->y;
+    IDWTELEM *b[4] = { cs->b[0], cs->b[1] };
+    b[2] = d->buffer + mirror(y+1, height-1)*stride;
+    b[3] = d->buffer + mirror(y+2, height-1)*stride;
+
+        if(y+1<(unsigned)height) vertical_compose_l0(b[1], b[2], b[3], width);
+        if(y+0<(unsigned)height) vertical_compose_h0(b[0], b[1], b[2], width);
+
+        if(y-1<(unsigned)height) d->horizontal_compose(b[0], d->temp, width);
+        if(y+0<(unsigned)height) d->horizontal_compose(b[1], d->temp, width);
+
+    cs->b[0] = b[2];
+    cs->b[1] = b[3];
+    cs->y += 2;
+}
+
+
+static void spatial_compose_dd137i_dy(DWTContext *d, int level, int width, int height, int stride)
+{
+    vertical_compose_5tap vertical_compose_l0 = (void*)d->vertical_compose_l0;
+    vertical_compose_5tap vertical_compose_h0 = (void*)d->vertical_compose_h0;
+    DWTCompose *cs = d->cs + level;
+
+    int i, y = cs->y;
+    IDWTELEM *b[10];
+    for (i = 0; i < 8; i++)
+        b[i] = cs->b[i];
+    b[8] = d->buffer + av_clip(y+7, 0, height-2)*stride;
+    b[9] = d->buffer + av_clip(y+8, 1, height-1)*stride;
+
+        if(y+5<(unsigned)height) vertical_compose_l0(b[3], b[5], b[6], b[7], b[9], width);
+        if(y+1<(unsigned)height) vertical_compose_h0(b[0], b[2], b[3], b[4], b[6], width);
+
+        if(y-1<(unsigned)height) d->horizontal_compose(b[0], d->temp, width);
+        if(y+0<(unsigned)height) d->horizontal_compose(b[1], d->temp, width);
+
+    for (i = 0; i < 8; i++)
+        cs->b[i] = b[i+2];
+    cs->y += 2;
+}
+
+// haar makes the assumption that height is even (always true for dirac)
+static void spatial_compose_haari_dy(DWTContext *d, int level, int width, int height, int stride)
+{
+    vertical_compose_2tap vertical_compose = (void*)d->vertical_compose;
+    int y = d->cs[level].y;
+    IDWTELEM *b0 = d->buffer + (y-1)*stride;
+    IDWTELEM *b1 = d->buffer + (y  )*stride;
+
+    vertical_compose(b0, b1, width);
+    d->horizontal_compose(b0, d->temp, width);
+    d->horizontal_compose(b1, d->temp, width);
+
+    d->cs[level].y += 2;
+}
+
+// Don't do sliced idwt for fidelity; the 9 tap filter makes it a bit annoying
+// Fortunately, this filter isn't used in practice.
+static void spatial_compose_fidelity(DWTContext *d, int level, int width, int height, int stride)
+{
+    vertical_compose_9tap vertical_compose_l0 = (void*)d->vertical_compose_l0;
+    vertical_compose_9tap vertical_compose_h0 = (void*)d->vertical_compose_h0;
+    int i, y;
+    IDWTELEM *b[8];
+
+    for (y = 1; y < height; y += 2) {
+        for (i = 0; i < 8; i++)
+            b[i] = d->buffer + av_clip((y-7 + 2*i), 0, height-2)*stride;
+        vertical_compose_h0(d->buffer + y*stride, b, width);
+    }
+
+    for (y = 0; y < height; y += 2) {
+        for (i = 0; i < 8; i++)
+            b[i] = d->buffer + av_clip((y-7 + 2*i), 1, height-1)*stride;
+        vertical_compose_l0(d->buffer + y*stride, b, width);
+    }
+
+    for (y = 0; y < height; y++)
+        d->horizontal_compose(d->buffer + y*stride, d->temp, width);
+
+    d->cs[level].y = height+1;
+}
+
+static void spatial_compose_daub97i_dy(DWTContext *d, int level, int width, int height, int stride)
+{
+    vertical_compose_3tap vertical_compose_l0 = (void*)d->vertical_compose_l0;
+    vertical_compose_3tap vertical_compose_h0 = (void*)d->vertical_compose_h0;
+    vertical_compose_3tap vertical_compose_l1 = (void*)d->vertical_compose_l1;
+    vertical_compose_3tap vertical_compose_h1 = (void*)d->vertical_compose_h1;
+    DWTCompose *cs = d->cs + level;
+
+    int i, y = cs->y;
+    IDWTELEM *b[6];
+    for (i = 0; i < 4; i++)
+        b[i] = cs->b[i];
+    b[4] = d->buffer + mirror(y+3, height-1)*stride;
+    b[5] = d->buffer + mirror(y+4, height-1)*stride;
+
+        if(y+3<(unsigned)height) vertical_compose_l1(b[3], b[4], b[5], width);
+        if(y+2<(unsigned)height) vertical_compose_h1(b[2], b[3], b[4], width);
+        if(y+1<(unsigned)height) vertical_compose_l0(b[1], b[2], b[3], width);
+        if(y+0<(unsigned)height) vertical_compose_h0(b[0], b[1], b[2], width);
+
+        if(y-1<(unsigned)height) d->horizontal_compose(b[0], d->temp, width);
+        if(y+0<(unsigned)height) d->horizontal_compose(b[1], d->temp, width);
+
+    for (i = 0; i < 4; i++)
+        cs->b[i] = b[i+2];
+    cs->y += 2;
+}
+
+
+static void spatial_compose97i_init2(DWTCompose *cs, IDWTELEM *buffer, int height, int stride)
+{
+    cs->b[0] = buffer + mirror(-3-1, height-1)*stride;
+    cs->b[1] = buffer + mirror(-3  , height-1)*stride;
+    cs->b[2] = buffer + mirror(-3+1, height-1)*stride;
+    cs->b[3] = buffer + mirror(-3+2, height-1)*stride;
+    cs->y = -3;
+}
+
+static void spatial_compose53i_init2(DWTCompose *cs, IDWTELEM *buffer, int height, int stride)
+{
+    cs->b[0] = buffer + mirror(-1-1, height-1)*stride;
+    cs->b[1] = buffer + mirror(-1  , height-1)*stride;
+    cs->y = -1;
+}
+
+static void spatial_compose_dd97i_init(DWTCompose *cs, IDWTELEM *buffer, int height, int stride)
+{
+    cs->b[0] = buffer + av_clip(-5-1, 0, height-2)*stride;
+    cs->b[1] = buffer + av_clip(-5  , 1, height-1)*stride;
+    cs->b[2] = buffer + av_clip(-5+1, 0, height-2)*stride;
+    cs->b[3] = buffer + av_clip(-5+2, 1, height-1)*stride;
+    cs->b[4] = buffer + av_clip(-5+3, 0, height-2)*stride;
+    cs->b[5] = buffer + av_clip(-5+4, 1, height-1)*stride;
+    cs->y = -5;
+}
+
+static void spatial_compose_dd137i_init(DWTCompose *cs, IDWTELEM *buffer, int height, int stride)
+{
+    cs->b[0] = buffer + av_clip(-5-1, 0, height-2)*stride;
+    cs->b[1] = buffer + av_clip(-5  , 1, height-1)*stride;
+    cs->b[2] = buffer + av_clip(-5+1, 0, height-2)*stride;
+    cs->b[3] = buffer + av_clip(-5+2, 1, height-1)*stride;
+    cs->b[4] = buffer + av_clip(-5+3, 0, height-2)*stride;
+    cs->b[5] = buffer + av_clip(-5+4, 1, height-1)*stride;
+    cs->b[6] = buffer + av_clip(-5+5, 0, height-2)*stride;
+    cs->b[7] = buffer + av_clip(-5+6, 1, height-1)*stride;
+    cs->y = -5;
+}
+
+int ff_spatial_idwt_init2(DWTContext *d, IDWTELEM *buffer, int width, int height,
+                          int stride, enum dwt_type type, int decomposition_count,
+                          IDWTELEM *temp)
+{
+    int level;
+
+    d->buffer = buffer;
+    d->width = width;
+    d->height = height;
+    d->stride = stride;
+    d->decomposition_count = decomposition_count;
+    d->temp = temp + 8;
+
+    for(level=decomposition_count-1; level>=0; level--){
+        int hl = height >> level;
+        int stride_l = stride << level;
+
+        switch(type){
+        case DWT_DIRAC_DD9_7:
+            spatial_compose_dd97i_init(d->cs+level, buffer, hl, stride_l);
+            break;
+        case DWT_DIRAC_LEGALL5_3:
+            spatial_compose53i_init2(d->cs+level, buffer, hl, stride_l);
+            break;
+        case DWT_DIRAC_DD13_7:
+            spatial_compose_dd137i_init(d->cs+level, buffer, hl, stride_l);
+            break;
+        case DWT_DIRAC_HAAR0:
+        case DWT_DIRAC_HAAR1:
+            d->cs[level].y = 1;
+            break;
+        case DWT_DIRAC_DAUB9_7:
+            spatial_compose97i_init2(d->cs+level, buffer, hl, stride_l);
+            break;
+        default:
+            d->cs[level].y = 0;
+            break;
+        }
+    }
+
+    switch (type) {
+    case DWT_DIRAC_DD9_7:
+        d->spatial_compose = spatial_compose_dd97i_dy;
+        d->vertical_compose_l0 = (void*)vertical_compose53iL0;
+        d->vertical_compose_h0 = (void*)vertical_compose_dd97iH0;
+        d->horizontal_compose = horizontal_compose_dd97i;
+        d->support = 7;
+        break;
+    case DWT_DIRAC_LEGALL5_3:
+        d->spatial_compose = spatial_compose_dirac53i_dy;
+        d->vertical_compose_l0 = (void*)vertical_compose53iL0;
+        d->vertical_compose_h0 = (void*)vertical_compose_dirac53iH0;
+        d->horizontal_compose = horizontal_compose_dirac53i;
+        d->support = 3;
+        break;
+    case DWT_DIRAC_DD13_7:
+        d->spatial_compose = spatial_compose_dd137i_dy;
+        d->vertical_compose_l0 = (void*)vertical_compose_dd137iL0;
+        d->vertical_compose_h0 = (void*)vertical_compose_dd97iH0;
+        d->horizontal_compose = horizontal_compose_dd137i;
+        d->support = 7;
+        break;
+    case DWT_DIRAC_HAAR0:
+    case DWT_DIRAC_HAAR1:
+        d->spatial_compose = spatial_compose_haari_dy;
+        d->vertical_compose = (void*)vertical_compose_haar;
+        if (type == DWT_DIRAC_HAAR0)
+            d->horizontal_compose = horizontal_compose_haar0i;
+        else
+            d->horizontal_compose = horizontal_compose_haar1i;
+        d->support = 1;
+        break;
+    case DWT_DIRAC_FIDELITY:
+        d->spatial_compose = spatial_compose_fidelity;
+        d->vertical_compose_l0 = (void*)vertical_compose_fidelityiL0;
+        d->vertical_compose_h0 = (void*)vertical_compose_fidelityiH0;
+        d->horizontal_compose = horizontal_compose_fidelityi;
+        break;
+    case DWT_DIRAC_DAUB9_7:
+        d->spatial_compose = spatial_compose_daub97i_dy;
+        d->vertical_compose_l0 = (void*)vertical_compose_daub97iL0;
+        d->vertical_compose_h0 = (void*)vertical_compose_daub97iH0;
+        d->vertical_compose_l1 = (void*)vertical_compose_daub97iL1;
+        d->vertical_compose_h1 = (void*)vertical_compose_daub97iH1;
+        d->horizontal_compose = horizontal_compose_daub97i;
+        d->support = 5;
+        break;
+    default:
+        av_log(NULL, AV_LOG_ERROR, "Unknown wavelet type %d\n", type);
+        return -1;
+    }
+
+    if (HAVE_MMX) ff_spatial_idwt_init_mmx(d, type);
+
+    return 0;
+}
+
+void ff_spatial_idwt_slice2(DWTContext *d, int y)
+{
+    int level, support = d->support;
+
+    for (level = d->decomposition_count-1; level >= 0; level--) {
+        int wl = d->width  >> level;
+        int hl = d->height >> level;
+        int stride_l = d->stride << level;
+
+        while (d->cs[level].y <= FFMIN((y>>level)+support, hl))
+            d->spatial_compose(d, level, wl, hl, stride_l);
+    }
+}
+
+int ff_spatial_idwt2(IDWTELEM *buffer, int width, int height, int stride,
+                     enum dwt_type type, int decomposition_count, IDWTELEM *temp)
+{
+    DWTContext d;
+    int y;
+
+    if (ff_spatial_idwt_init2(&d, buffer, width, height, stride, type, decomposition_count, temp))
+        return -1;
+
+    for (y = 0; y < d.height; y += 4)
+        ff_spatial_idwt_slice2(&d, y);
+
+    return 0;
+}
diff --git a/libavcodec/dirac_dwt.h b/libavcodec/dirac_dwt.h
new file mode 100644
index 0000000000..9514e9575b
--- /dev/null
+++ b/libavcodec/dirac_dwt.h
@@ -0,0 +1,126 @@
+/*
+ * Copyright (C) 2004-2010 Michael Niedermayer <michaelni@gmx.at>
+ *
+ * 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
+ */
+
+#ifndef AVCODEC_DIRAC_DWT_H
+#define AVCODEC_DIRAC_DWT_H
+
+#include <stdint.h>
+
+typedef int DWTELEM;
+typedef short IDWTELEM;
+
+#define MAX_DWT_SUPPORT 8
+#define MAX_DECOMPOSITIONS 8
+
+typedef struct DWTCompose {
+    IDWTELEM *b[MAX_DWT_SUPPORT];
+    int y;
+} DWTCompose;
+
+struct DWTContext;
+
+// Possible prototypes for vertical_compose functions
+typedef void (*vertical_compose_2tap)(IDWTELEM *b0, IDWTELEM *b1, int width);
+typedef void (*vertical_compose_3tap)(IDWTELEM *b0, IDWTELEM *b1, IDWTELEM *b2, int width);
+typedef void (*vertical_compose_5tap)(IDWTELEM *b0, IDWTELEM *b1, IDWTELEM *b2, IDWTELEM *b3, IDWTELEM *b4, int width);
+typedef void (*vertical_compose_9tap)(IDWTELEM *dst, IDWTELEM *b[8], int width);
+
+typedef struct DWTContext {
+    IDWTELEM *buffer;
+    IDWTELEM *temp;
+    int width;
+    int height;
+    int stride;
+    int decomposition_count;
+    int support;
+
+    void (*spatial_compose)(struct DWTContext *cs, int level, int width, int height, int stride);
+    void (*vertical_compose_l0)(void);
+    void (*vertical_compose_h0)(void);
+    void (*vertical_compose_l1)(void);
+    void (*vertical_compose_h1)(void);
+    void (*vertical_compose)(void);     ///< one set of lowpass and highpass combined
+    void (*horizontal_compose)(IDWTELEM *b, IDWTELEM *tmp, int width);
+
+    DWTCompose cs[MAX_DECOMPOSITIONS];
+} DWTContext;
+
+enum dwt_type {
+    DWT_SNOW_DAUB9_7,
+    DWT_SNOW_LEGALL5_3,
+    DWT_DIRAC_DD9_7,
+    DWT_DIRAC_LEGALL5_3,
+    DWT_DIRAC_DD13_7,
+    DWT_DIRAC_HAAR0,
+    DWT_DIRAC_HAAR1,
+    DWT_DIRAC_FIDELITY,
+    DWT_DIRAC_DAUB9_7,
+    DWT_NUM_TYPES
+};
+
+// -1 if an error occurred, e.g. the dwt_type isn't recognized
+int ff_spatial_idwt_init2(DWTContext *d, IDWTELEM *buffer, int width, int height,
+                          int stride, enum dwt_type type, int decomposition_count,
+                          IDWTELEM *temp);
+
+int ff_spatial_idwt2(IDWTELEM *buffer, int width, int height, int stride,
+                     enum dwt_type type, int decomposition_count, IDWTELEM *temp);
+
+void ff_spatial_idwt_slice2(DWTContext *d, int y);
+
+// shared stuff for simd optimiztions
+#define COMPOSE_53iL0(b0, b1, b2)\
+    (b1 - ((b0 + b2 + 2) >> 2))
+
+#define COMPOSE_DIRAC53iH0(b0, b1, b2)\
+    (b1 + ((b0 + b2 + 1) >> 1))
+
+#define COMPOSE_DD97iH0(b0, b1, b2, b3, b4)\
+    (b2 + ((-b0 + 9*b1 + 9*b3 - b4 + 8) >> 4))
+
+#define COMPOSE_DD137iL0(b0, b1, b2, b3, b4)\
+    (b2 - ((-b0 + 9*b1 + 9*b3 - b4 + 16) >> 5))
+
+#define COMPOSE_HAARiL0(b0, b1)\
+    (b0 - ((b1 + 1) >> 1))
+
+#define COMPOSE_HAARiH0(b0, b1)\
+    (b0 + b1)
+
+#define COMPOSE_FIDELITYiL0(b0, b1, b2, b3, b4, b5, b6, b7, b8)\
+    (b4 - ((-8*(b0+b8) + 21*(b1+b7) - 46*(b2+b6) + 161*(b3+b5) + 128) >> 8))
+
+#define COMPOSE_FIDELITYiH0(b0, b1, b2, b3, b4, b5, b6, b7, b8)\
+    (b4 + ((-2*(b0+b8) + 10*(b1+b7) - 25*(b2+b6) + 81*(b3+b5) + 128) >> 8))
+
+#define COMPOSE_DAUB97iL1(b0, b1, b2)\
+    (b1 - ((1817*(b0 + b2) + 2048) >> 12))
+
+#define COMPOSE_DAUB97iH1(b0, b1, b2)\
+    (b1 - (( 113*(b0 + b2) + 64) >> 7))
+
+#define COMPOSE_DAUB97iL0(b0, b1, b2)\
+    (b1 + (( 217*(b0 + b2) + 2048) >> 12))
+
+#define COMPOSE_DAUB97iH0(b0, b1, b2)\
+    (b1 + ((6497*(b0 + b2) + 2048) >> 12))
+
+
+#endif /* AVCODEC_DWT_H */
diff --git a/libavcodec/diracdec.c b/libavcodec/diracdec.c
index c431b97019..a9191cd9cd 100644
--- a/libavcodec/diracdec.c
+++ b/libavcodec/diracdec.c
@@ -34,7 +34,7 @@
 #include "golomb.h"
 #include "dirac_arith.h"
 #include "mpeg12data.h"
-#include "dwt.h"
+#include "dirac_dwt.h"
 #include "dirac.h"
 #include "diracdsp.h"
 
diff --git a/libavcodec/ivi_dsp.c b/libavcodec/ivi_dsp.c
index 52cd402cb1..4411c090c4 100644
--- a/libavcodec/ivi_dsp.c
+++ b/libavcodec/ivi_dsp.c
@@ -28,7 +28,6 @@
 
 #include "avcodec.h"
 #include "dsputil.h"
-#include "dwt.h"
 #include "ivi_common.h"
 #include "ivi_dsp.h"
 
@@ -40,7 +39,7 @@ void ff_ivi_recompose53(const IVIPlaneDesc *plane, uint8_t *dst,
     int32_t         b0_1, b0_2, b1_1, b1_2, b1_3, b2_1, b2_2, b2_3, b2_4, b2_5, b2_6;
     int32_t         b3_1, b3_2, b3_3, b3_4, b3_5, b3_6, b3_7, b3_8, b3_9;
     int32_t         pitch, back_pitch;
-    const IDWTELEM *b0_ptr, *b1_ptr, *b2_ptr, *b3_ptr;
+    const short     *b0_ptr, *b1_ptr, *b2_ptr, *b3_ptr;
     const int       num_bands = 4;
 
     /* all bands should have the same pitch */
@@ -193,7 +192,7 @@ void ff_ivi_recompose_haar(const IVIPlaneDesc *plane, uint8_t *dst,
                            const int dst_pitch)
 {
     int             x, y, indx, b0, b1, b2, b3, p0, p1, p2, p3;
-    const IDWTELEM *b0_ptr, *b1_ptr, *b2_ptr, *b3_ptr;
+    const short     *b0_ptr, *b1_ptr, *b2_ptr, *b3_ptr;
     int32_t         pitch;
 
     /* all bands should have the same pitch */
diff --git a/libavcodec/snow.c b/libavcodec/snow.c
index 3adf723f70..1efe560f2d 100644
--- a/libavcodec/snow.c
+++ b/libavcodec/snow.c
@@ -23,7 +23,7 @@
 #include "libavutil/opt.h"
 #include "avcodec.h"
 #include "dsputil.h"
-#include "dwt.h"
+#include "snow_dwt.h"
 #include "internal.h"
 #include "snow.h"
 #include "snowdata.h"
diff --git a/libavcodec/snow.h b/libavcodec/snow.h
index 2565988a57..83c64c0e1b 100644
--- a/libavcodec/snow.h
+++ b/libavcodec/snow.h
@@ -23,7 +23,7 @@
 #define AVCODEC_SNOW_H
 
 #include "dsputil.h"
-#include "dwt.h"
+#include "snow_dwt.h"
 
 #include "rangecoder.h"
 #include "mathops.h"
@@ -109,7 +109,7 @@ typedef struct SnowContext{
     RangeCoder c;
     DSPContext dsp;
     VideoDSPContext vdsp;
-    DWTContext dwt;
+    SnowDWTContext dwt;
     AVFrame new_picture;
     AVFrame input_picture;              ///< new_picture with the internal linesizes
     AVFrame current_picture;
diff --git a/libavcodec/dwt.c b/libavcodec/snow_dwt.c
index d36d25d1a6..28edf6aafb 100644
--- a/libavcodec/dwt.c
+++ b/libavcodec/snow_dwt.c
@@ -23,8 +23,7 @@
 #include "libavutil/avassert.h"
 #include "libavutil/common.h"
 #include "dsputil.h"
-#include "dwt.h"
-#include "libavcodec/x86/dwt.h"
+#include "snow_dwt.h"
 
 int ff_slice_buffer_init(slice_buffer *buf, int line_count,
                          int max_allocated_lines, int line_width,
@@ -566,7 +565,7 @@ static void spatial_compose97i_init(DWTCompose *cs, IDWTELEM *buffer, int height
     cs->y  = -3;
 }
 
-static void spatial_compose97i_dy_buffered(DWTContext *dsp, DWTCompose *cs,
+static void spatial_compose97i_dy_buffered(SnowDWTContext *dsp, DWTCompose *cs,
                                            slice_buffer * sb, IDWTELEM *temp,
                                            int width, int height,
                                            int stride_line)
@@ -661,7 +660,7 @@ void ff_spatial_idwt_buffered_init(DWTCompose *cs, slice_buffer *sb, int width,
     }
 }
 
-void ff_spatial_idwt_buffered_slice(DWTContext *dsp, DWTCompose *cs,
+void ff_spatial_idwt_buffered_slice(SnowDWTContext *dsp, DWTCompose *cs,
                                     slice_buffer *slice_buf, IDWTELEM *temp,
                                     int width, int height, int stride_line,
                                     int type, int decomposition_count, int y)
@@ -853,7 +852,7 @@ void ff_dsputil_init_dwt(DSPContext *c)
     c->w97[1] = w97_8_c;
 }
 
-void ff_dwt_init(DWTContext *c)
+void ff_dwt_init(SnowDWTContext *c)
 {
     c->vertical_compose97i   = ff_snow_vertical_compose97i;
     c->horizontal_compose97i = ff_snow_horizontal_compose97i;
@@ -864,539 +863,3 @@ void ff_dwt_init(DWTContext *c)
 }
 
 
-static av_always_inline
-void interleave(IDWTELEM *dst, IDWTELEM *src0, IDWTELEM *src1, int w2, int add, int shift)
-{
-    int i;
-    for (i = 0; i < w2; i++) {
-        dst[2*i  ] = (src0[i] + add) >> shift;
-        dst[2*i+1] = (src1[i] + add) >> shift;
-    }
-}
-
-static void horizontal_compose_dirac53i(IDWTELEM *b, IDWTELEM *temp, int w)
-{
-    const int w2 = w >> 1;
-    int x;
-
-    temp[0] = COMPOSE_53iL0(b[w2], b[0], b[w2]);
-    for (x = 1; x < w2; x++) {
-        temp[x     ] = COMPOSE_53iL0     (b[x+w2-1], b[x     ], b[x+w2]);
-        temp[x+w2-1] = COMPOSE_DIRAC53iH0(temp[x-1], b[x+w2-1], temp[x]);
-    }
-    temp[w-1] = COMPOSE_DIRAC53iH0(temp[w2-1], b[w-1], temp[w2-1]);
-
-    interleave(b, temp, temp+w2, w2, 1, 1);
-}
-
-static void horizontal_compose_dd97i(IDWTELEM *b, IDWTELEM *tmp, int w)
-{
-    const int w2 = w >> 1;
-    int x;
-
-    tmp[0] = COMPOSE_53iL0(b[w2], b[0], b[w2]);
-    for (x = 1; x < w2; x++)
-        tmp[x] = COMPOSE_53iL0(b[x+w2-1], b[x], b[x+w2]);
-
-    // extend the edges
-    tmp[-1]   = tmp[0];
-    tmp[w2+1] = tmp[w2] = tmp[w2-1];
-
-    for (x = 0; x < w2; x++) {
-        b[2*x  ] = (tmp[x] + 1)>>1;
-        b[2*x+1] = (COMPOSE_DD97iH0(tmp[x-1], tmp[x], b[x+w2], tmp[x+1], tmp[x+2]) + 1)>>1;
-    }
-}
-
-static void horizontal_compose_dd137i(IDWTELEM *b, IDWTELEM *tmp, int w)
-{
-    const int w2 = w >> 1;
-    int x;
-
-    tmp[0] = COMPOSE_DD137iL0(b[w2], b[w2], b[0], b[w2  ], b[w2+1]);
-    tmp[1] = COMPOSE_DD137iL0(b[w2], b[w2], b[1], b[w2+1], b[w2+2]);
-    for (x = 2; x < w2-1; x++)
-        tmp[x] = COMPOSE_DD137iL0(b[x+w2-2], b[x+w2-1], b[x], b[x+w2], b[x+w2+1]);
-    tmp[w2-1] = COMPOSE_DD137iL0(b[w-3], b[w-2], b[w2-1], b[w-1], b[w-1]);
-
-    // extend the edges
-    tmp[-1]   = tmp[0];
-    tmp[w2+1] = tmp[w2] = tmp[w2-1];
-
-    for (x = 0; x < w2; x++) {
-        b[2*x  ] = (tmp[x] + 1)>>1;
-        b[2*x+1] = (COMPOSE_DD97iH0(tmp[x-1], tmp[x], b[x+w2], tmp[x+1], tmp[x+2]) + 1)>>1;
-    }
-}
-
-static av_always_inline
-void horizontal_compose_haari(IDWTELEM *b, IDWTELEM *temp, int w, int shift)
-{
-    const int w2 = w >> 1;
-    int x;
-
-    for (x = 0; x < w2; x++) {
-        temp[x   ] = COMPOSE_HAARiL0(b[x   ], b[x+w2]);
-        temp[x+w2] = COMPOSE_HAARiH0(b[x+w2], temp[x]);
-    }
-
-    interleave(b, temp, temp+w2, w2, shift, shift);
-}
-
-static void horizontal_compose_haar0i(IDWTELEM *b, IDWTELEM *temp, int w)
-{
-    horizontal_compose_haari(b, temp, w, 0);
-}
-
-static void horizontal_compose_haar1i(IDWTELEM *b, IDWTELEM *temp, int w)
-{
-    horizontal_compose_haari(b, temp, w, 1);
-}
-
-static void horizontal_compose_fidelityi(IDWTELEM *b, IDWTELEM *tmp, int w)
-{
-    const int w2 = w >> 1;
-    int i, x;
-    IDWTELEM v[8];
-
-    for (x = 0; x < w2; x++) {
-        for (i = 0; i < 8; i++)
-            v[i] = b[av_clip(x-3+i, 0, w2-1)];
-        tmp[x] = COMPOSE_FIDELITYiH0(v[0], v[1], v[2], v[3], b[x+w2], v[4], v[5], v[6], v[7]);
-    }
-
-    for (x = 0; x < w2; x++) {
-        for (i = 0; i < 8; i++)
-            v[i] = tmp[av_clip(x-4+i, 0, w2-1)];
-        tmp[x+w2] = COMPOSE_FIDELITYiL0(v[0], v[1], v[2], v[3], b[x], v[4], v[5], v[6], v[7]);
-    }
-
-    interleave(b, tmp+w2, tmp, w2, 0, 0);
-}
-
-static void horizontal_compose_daub97i(IDWTELEM *b, IDWTELEM *temp, int w)
-{
-    const int w2 = w >> 1;
-    int x, b0, b1, b2;
-
-    temp[0] = COMPOSE_DAUB97iL1(b[w2], b[0], b[w2]);
-    for (x = 1; x < w2; x++) {
-        temp[x     ] = COMPOSE_DAUB97iL1(b[x+w2-1], b[x     ], b[x+w2]);
-        temp[x+w2-1] = COMPOSE_DAUB97iH1(temp[x-1], b[x+w2-1], temp[x]);
-    }
-    temp[w-1] = COMPOSE_DAUB97iH1(temp[w2-1], b[w-1], temp[w2-1]);
-
-    // second stage combined with interleave and shift
-    b0 = b2 = COMPOSE_DAUB97iL0(temp[w2], temp[0], temp[w2]);
-    b[0] = (b0 + 1) >> 1;
-    for (x = 1; x < w2; x++) {
-        b2 = COMPOSE_DAUB97iL0(temp[x+w2-1], temp[x     ], temp[x+w2]);
-        b1 = COMPOSE_DAUB97iH0(          b0, temp[x+w2-1], b2        );
-        b[2*x-1] = (b1 + 1) >> 1;
-        b[2*x  ] = (b2 + 1) >> 1;
-        b0 = b2;
-    }
-    b[w-1] = (COMPOSE_DAUB97iH0(b2, temp[w-1], b2) + 1) >> 1;
-}
-
-static void vertical_compose_dirac53iH0(IDWTELEM *b0, IDWTELEM *b1, IDWTELEM *b2, int width)
-{
-    int i;
-
-    for(i=0; i<width; i++){
-        b1[i] = COMPOSE_DIRAC53iH0(b0[i], b1[i], b2[i]);
-    }
-}
-
-static void vertical_compose_dd97iH0(IDWTELEM *b0, IDWTELEM *b1, IDWTELEM *b2,
-                                  IDWTELEM *b3, IDWTELEM *b4, int width)
-{
-    int i;
-
-    for(i=0; i<width; i++){
-        b2[i] = COMPOSE_DD97iH0(b0[i], b1[i], b2[i], b3[i], b4[i]);
-    }
-}
-
-static void vertical_compose_dd137iL0(IDWTELEM *b0, IDWTELEM *b1, IDWTELEM *b2,
-                                      IDWTELEM *b3, IDWTELEM *b4, int width)
-{
-    int i;
-
-    for(i=0; i<width; i++){
-        b2[i] = COMPOSE_DD137iL0(b0[i], b1[i], b2[i], b3[i], b4[i]);
-    }
-}
-
-static void vertical_compose_haar(IDWTELEM *b0, IDWTELEM *b1, int width)
-{
-    int i;
-
-    for (i = 0; i < width; i++) {
-        b0[i] = COMPOSE_HAARiL0(b0[i], b1[i]);
-        b1[i] = COMPOSE_HAARiH0(b1[i], b0[i]);
-    }
-}
-
-static void vertical_compose_fidelityiH0(IDWTELEM *dst, IDWTELEM *b[8], int width)
-{
-    int i;
-
-    for(i=0; i<width; i++){
-        dst[i] = COMPOSE_FIDELITYiH0(b[0][i], b[1][i], b[2][i], b[3][i], dst[i], b[4][i], b[5][i], b[6][i], b[7][i]);
-    }
-}
-
-static void vertical_compose_fidelityiL0(IDWTELEM *dst, IDWTELEM *b[8], int width)
-{
-    int i;
-
-    for(i=0; i<width; i++){
-        dst[i] = COMPOSE_FIDELITYiL0(b[0][i], b[1][i], b[2][i], b[3][i], dst[i], b[4][i], b[5][i], b[6][i], b[7][i]);
-    }
-}
-
-static void vertical_compose_daub97iH0(IDWTELEM *b0, IDWTELEM *b1, IDWTELEM *b2, int width)
-{
-    int i;
-
-    for(i=0; i<width; i++){
-        b1[i] = COMPOSE_DAUB97iH0(b0[i], b1[i], b2[i]);
-    }
-}
-
-static void vertical_compose_daub97iH1(IDWTELEM *b0, IDWTELEM *b1, IDWTELEM *b2, int width)
-{
-    int i;
-
-    for(i=0; i<width; i++){
-        b1[i] = COMPOSE_DAUB97iH1(b0[i], b1[i], b2[i]);
-    }
-}
-
-static void vertical_compose_daub97iL0(IDWTELEM *b0, IDWTELEM *b1, IDWTELEM *b2, int width)
-{
-    int i;
-
-    for(i=0; i<width; i++){
-        b1[i] = COMPOSE_DAUB97iL0(b0[i], b1[i], b2[i]);
-    }
-}
-
-static void vertical_compose_daub97iL1(IDWTELEM *b0, IDWTELEM *b1, IDWTELEM *b2, int width)
-{
-    int i;
-
-    for(i=0; i<width; i++){
-        b1[i] = COMPOSE_DAUB97iL1(b0[i], b1[i], b2[i]);
-    }
-}
-
-
-static void spatial_compose_dd97i_dy(DWTContext *d, int level, int width, int height, int stride)
-{
-    vertical_compose_3tap vertical_compose_l0 = (void*)d->vertical_compose_l0;
-    vertical_compose_5tap vertical_compose_h0 = (void*)d->vertical_compose_h0;
-    DWTCompose *cs = d->cs + level;
-
-    int i, y = cs->y;
-    IDWTELEM *b[8];
-    for (i = 0; i < 6; i++)
-        b[i] = cs->b[i];
-    b[6] = d->buffer + av_clip(y+5, 0, height-2)*stride;
-    b[7] = d->buffer + av_clip(y+6, 1, height-1)*stride;
-
-        if(y+5<(unsigned)height) vertical_compose_l0(      b[5], b[6], b[7],       width);
-        if(y+1<(unsigned)height) vertical_compose_h0(b[0], b[2], b[3], b[4], b[6], width);
-
-        if(y-1<(unsigned)height) d->horizontal_compose(b[0], d->temp, width);
-        if(y+0<(unsigned)height) d->horizontal_compose(b[1], d->temp, width);
-
-    for (i = 0; i < 6; i++)
-        cs->b[i] = b[i+2];
-    cs->y += 2;
-}
-
-static void spatial_compose_dirac53i_dy(DWTContext *d, int level, int width, int height, int stride)
-{
-    vertical_compose_3tap vertical_compose_l0 = (void*)d->vertical_compose_l0;
-    vertical_compose_3tap vertical_compose_h0 = (void*)d->vertical_compose_h0;
-    DWTCompose *cs = d->cs + level;
-
-    int y= cs->y;
-    IDWTELEM *b[4] = { cs->b[0], cs->b[1] };
-    b[2] = d->buffer + mirror(y+1, height-1)*stride;
-    b[3] = d->buffer + mirror(y+2, height-1)*stride;
-
-        if(y+1<(unsigned)height) vertical_compose_l0(b[1], b[2], b[3], width);
-        if(y+0<(unsigned)height) vertical_compose_h0(b[0], b[1], b[2], width);
-
-        if(y-1<(unsigned)height) d->horizontal_compose(b[0], d->temp, width);
-        if(y+0<(unsigned)height) d->horizontal_compose(b[1], d->temp, width);
-
-    cs->b[0] = b[2];
-    cs->b[1] = b[3];
-    cs->y += 2;
-}
-
-
-static void spatial_compose_dd137i_dy(DWTContext *d, int level, int width, int height, int stride)
-{
-    vertical_compose_5tap vertical_compose_l0 = (void*)d->vertical_compose_l0;
-    vertical_compose_5tap vertical_compose_h0 = (void*)d->vertical_compose_h0;
-    DWTCompose *cs = d->cs + level;
-
-    int i, y = cs->y;
-    IDWTELEM *b[10];
-    for (i = 0; i < 8; i++)
-        b[i] = cs->b[i];
-    b[8] = d->buffer + av_clip(y+7, 0, height-2)*stride;
-    b[9] = d->buffer + av_clip(y+8, 1, height-1)*stride;
-
-        if(y+5<(unsigned)height) vertical_compose_l0(b[3], b[5], b[6], b[7], b[9], width);
-        if(y+1<(unsigned)height) vertical_compose_h0(b[0], b[2], b[3], b[4], b[6], width);
-
-        if(y-1<(unsigned)height) d->horizontal_compose(b[0], d->temp, width);
-        if(y+0<(unsigned)height) d->horizontal_compose(b[1], d->temp, width);
-
-    for (i = 0; i < 8; i++)
-        cs->b[i] = b[i+2];
-    cs->y += 2;
-}
-
-// haar makes the assumption that height is even (always true for dirac)
-static void spatial_compose_haari_dy(DWTContext *d, int level, int width, int height, int stride)
-{
-    vertical_compose_2tap vertical_compose = (void*)d->vertical_compose;
-    int y = d->cs[level].y;
-    IDWTELEM *b0 = d->buffer + (y-1)*stride;
-    IDWTELEM *b1 = d->buffer + (y  )*stride;
-
-    vertical_compose(b0, b1, width);
-    d->horizontal_compose(b0, d->temp, width);
-    d->horizontal_compose(b1, d->temp, width);
-
-    d->cs[level].y += 2;
-}
-
-// Don't do sliced idwt for fidelity; the 9 tap filter makes it a bit annoying
-// Fortunately, this filter isn't used in practice.
-static void spatial_compose_fidelity(DWTContext *d, int level, int width, int height, int stride)
-{
-    vertical_compose_9tap vertical_compose_l0 = (void*)d->vertical_compose_l0;
-    vertical_compose_9tap vertical_compose_h0 = (void*)d->vertical_compose_h0;
-    int i, y;
-    IDWTELEM *b[8];
-
-    for (y = 1; y < height; y += 2) {
-        for (i = 0; i < 8; i++)
-            b[i] = d->buffer + av_clip((y-7 + 2*i), 0, height-2)*stride;
-        vertical_compose_h0(d->buffer + y*stride, b, width);
-    }
-
-    for (y = 0; y < height; y += 2) {
-        for (i = 0; i < 8; i++)
-            b[i] = d->buffer + av_clip((y-7 + 2*i), 1, height-1)*stride;
-        vertical_compose_l0(d->buffer + y*stride, b, width);
-    }
-
-    for (y = 0; y < height; y++)
-        d->horizontal_compose(d->buffer + y*stride, d->temp, width);
-
-    d->cs[level].y = height+1;
-}
-
-static void spatial_compose_daub97i_dy(DWTContext *d, int level, int width, int height, int stride)
-{
-    vertical_compose_3tap vertical_compose_l0 = (void*)d->vertical_compose_l0;
-    vertical_compose_3tap vertical_compose_h0 = (void*)d->vertical_compose_h0;
-    vertical_compose_3tap vertical_compose_l1 = (void*)d->vertical_compose_l1;
-    vertical_compose_3tap vertical_compose_h1 = (void*)d->vertical_compose_h1;
-    DWTCompose *cs = d->cs + level;
-
-    int i, y = cs->y;
-    IDWTELEM *b[6];
-    for (i = 0; i < 4; i++)
-        b[i] = cs->b[i];
-    b[4] = d->buffer + mirror(y+3, height-1)*stride;
-    b[5] = d->buffer + mirror(y+4, height-1)*stride;
-
-        if(y+3<(unsigned)height) vertical_compose_l1(b[3], b[4], b[5], width);
-        if(y+2<(unsigned)height) vertical_compose_h1(b[2], b[3], b[4], width);
-        if(y+1<(unsigned)height) vertical_compose_l0(b[1], b[2], b[3], width);
-        if(y+0<(unsigned)height) vertical_compose_h0(b[0], b[1], b[2], width);
-
-        if(y-1<(unsigned)height) d->horizontal_compose(b[0], d->temp, width);
-        if(y+0<(unsigned)height) d->horizontal_compose(b[1], d->temp, width);
-
-    for (i = 0; i < 4; i++)
-        cs->b[i] = b[i+2];
-    cs->y += 2;
-}
-
-
-static void spatial_compose97i_init2(DWTCompose *cs, IDWTELEM *buffer, int height, int stride)
-{
-    cs->b[0] = buffer + mirror(-3-1, height-1)*stride;
-    cs->b[1] = buffer + mirror(-3  , height-1)*stride;
-    cs->b[2] = buffer + mirror(-3+1, height-1)*stride;
-    cs->b[3] = buffer + mirror(-3+2, height-1)*stride;
-    cs->y = -3;
-}
-
-static void spatial_compose53i_init2(DWTCompose *cs, IDWTELEM *buffer, int height, int stride)
-{
-    cs->b[0] = buffer + mirror(-1-1, height-1)*stride;
-    cs->b[1] = buffer + mirror(-1  , height-1)*stride;
-    cs->y = -1;
-}
-
-static void spatial_compose_dd97i_init(DWTCompose *cs, IDWTELEM *buffer, int height, int stride)
-{
-    cs->b[0] = buffer + av_clip(-5-1, 0, height-2)*stride;
-    cs->b[1] = buffer + av_clip(-5  , 1, height-1)*stride;
-    cs->b[2] = buffer + av_clip(-5+1, 0, height-2)*stride;
-    cs->b[3] = buffer + av_clip(-5+2, 1, height-1)*stride;
-    cs->b[4] = buffer + av_clip(-5+3, 0, height-2)*stride;
-    cs->b[5] = buffer + av_clip(-5+4, 1, height-1)*stride;
-    cs->y = -5;
-}
-
-static void spatial_compose_dd137i_init(DWTCompose *cs, IDWTELEM *buffer, int height, int stride)
-{
-    cs->b[0] = buffer + av_clip(-5-1, 0, height-2)*stride;
-    cs->b[1] = buffer + av_clip(-5  , 1, height-1)*stride;
-    cs->b[2] = buffer + av_clip(-5+1, 0, height-2)*stride;
-    cs->b[3] = buffer + av_clip(-5+2, 1, height-1)*stride;
-    cs->b[4] = buffer + av_clip(-5+3, 0, height-2)*stride;
-    cs->b[5] = buffer + av_clip(-5+4, 1, height-1)*stride;
-    cs->b[6] = buffer + av_clip(-5+5, 0, height-2)*stride;
-    cs->b[7] = buffer + av_clip(-5+6, 1, height-1)*stride;
-    cs->y = -5;
-}
-
-int ff_spatial_idwt_init2(DWTContext *d, IDWTELEM *buffer, int width, int height,
-                          int stride, enum dwt_type type, int decomposition_count,
-                          IDWTELEM *temp)
-{
-    int level;
-
-    d->buffer = buffer;
-    d->width = width;
-    d->height = height;
-    d->stride = stride;
-    d->decomposition_count = decomposition_count;
-    d->temp = temp + 8;
-
-    for(level=decomposition_count-1; level>=0; level--){
-        int hl = height >> level;
-        int stride_l = stride << level;
-
-        switch(type){
-        case DWT_DIRAC_DD9_7:
-            spatial_compose_dd97i_init(d->cs+level, buffer, hl, stride_l);
-            break;
-        case DWT_DIRAC_LEGALL5_3:
-            spatial_compose53i_init2(d->cs+level, buffer, hl, stride_l);
-            break;
-        case DWT_DIRAC_DD13_7:
-            spatial_compose_dd137i_init(d->cs+level, buffer, hl, stride_l);
-            break;
-        case DWT_DIRAC_HAAR0:
-        case DWT_DIRAC_HAAR1:
-            d->cs[level].y = 1;
-            break;
-        case DWT_DIRAC_DAUB9_7:
-            spatial_compose97i_init2(d->cs+level, buffer, hl, stride_l);
-            break;
-        default:
-            d->cs[level].y = 0;
-            break;
-        }
-    }
-
-    switch (type) {
-    case DWT_DIRAC_DD9_7:
-        d->spatial_compose = spatial_compose_dd97i_dy;
-        d->vertical_compose_l0 = (void*)vertical_compose53iL0;
-        d->vertical_compose_h0 = (void*)vertical_compose_dd97iH0;
-        d->horizontal_compose = horizontal_compose_dd97i;
-        d->support = 7;
-        break;
-    case DWT_DIRAC_LEGALL5_3:
-        d->spatial_compose = spatial_compose_dirac53i_dy;
-        d->vertical_compose_l0 = (void*)vertical_compose53iL0;
-        d->vertical_compose_h0 = (void*)vertical_compose_dirac53iH0;
-        d->horizontal_compose = horizontal_compose_dirac53i;
-        d->support = 3;
-        break;
-    case DWT_DIRAC_DD13_7:
-        d->spatial_compose = spatial_compose_dd137i_dy;
-        d->vertical_compose_l0 = (void*)vertical_compose_dd137iL0;
-        d->vertical_compose_h0 = (void*)vertical_compose_dd97iH0;
-        d->horizontal_compose = horizontal_compose_dd137i;
-        d->support = 7;
-        break;
-    case DWT_DIRAC_HAAR0:
-    case DWT_DIRAC_HAAR1:
-        d->spatial_compose = spatial_compose_haari_dy;
-        d->vertical_compose = (void*)vertical_compose_haar;
-        if (type == DWT_DIRAC_HAAR0)
-            d->horizontal_compose = horizontal_compose_haar0i;
-        else
-            d->horizontal_compose = horizontal_compose_haar1i;
-        d->support = 1;
-        break;
-    case DWT_DIRAC_FIDELITY:
-        d->spatial_compose = spatial_compose_fidelity;
-        d->vertical_compose_l0 = (void*)vertical_compose_fidelityiL0;
-        d->vertical_compose_h0 = (void*)vertical_compose_fidelityiH0;
-        d->horizontal_compose = horizontal_compose_fidelityi;
-        break;
-    case DWT_DIRAC_DAUB9_7:
-        d->spatial_compose = spatial_compose_daub97i_dy;
-        d->vertical_compose_l0 = (void*)vertical_compose_daub97iL0;
-        d->vertical_compose_h0 = (void*)vertical_compose_daub97iH0;
-        d->vertical_compose_l1 = (void*)vertical_compose_daub97iL1;
-        d->vertical_compose_h1 = (void*)vertical_compose_daub97iH1;
-        d->horizontal_compose = horizontal_compose_daub97i;
-        d->support = 5;
-        break;
-    default:
-        av_log(NULL, AV_LOG_ERROR, "Unknown wavelet type %d\n", type);
-        return -1;
-    }
-
-    if (HAVE_MMX) ff_spatial_idwt_init_mmx(d, type);
-
-    return 0;
-}
-
-void ff_spatial_idwt_slice2(DWTContext *d, int y)
-{
-    int level, support = d->support;
-
-    for (level = d->decomposition_count-1; level >= 0; level--) {
-        int wl = d->width  >> level;
-        int hl = d->height >> level;
-        int stride_l = d->stride << level;
-
-        while (d->cs[level].y <= FFMIN((y>>level)+support, hl))
-            d->spatial_compose(d, level, wl, hl, stride_l);
-    }
-}
-
-int ff_spatial_idwt2(IDWTELEM *buffer, int width, int height, int stride,
-                     enum dwt_type type, int decomposition_count, IDWTELEM *temp)
-{
-    DWTContext d;
-    int y;
-
-    if (ff_spatial_idwt_init2(&d, buffer, width, height, stride, type, decomposition_count, temp))
-        return -1;
-
-    for (y = 0; y < d.height; y += 4)
-        ff_spatial_idwt_slice2(&d, y);
-
-    return 0;
-}
diff --git a/libavcodec/dwt.h b/libavcodec/snow_dwt.h
index d82e510006..0806b389e2 100644
--- a/libavcodec/dwt.h
+++ b/libavcodec/snow_dwt.h
@@ -18,20 +18,17 @@
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  */
 
-#ifndef AVCODEC_DWT_H
-#define AVCODEC_DWT_H
+#ifndef AVCODEC_SNOW_DWT_H
+#define AVCODEC_SNOW_DWT_H
 
 #include <stdint.h>
 
 typedef int DWTELEM;
 typedef short IDWTELEM;
 
-#define MAX_DWT_SUPPORT 8
 #define MAX_DECOMPOSITIONS 8
 
 typedef struct DWTCompose {
-    IDWTELEM *b[MAX_DWT_SUPPORT];
-
     IDWTELEM *b0;
     IDWTELEM *b1;
     IDWTELEM *b2;
@@ -51,31 +48,9 @@ typedef struct slice_buffer_s {
     IDWTELEM *base_buffer;  ///< Buffer that this structure is caching.
 } slice_buffer;
 
-struct DWTContext;
-
-// Possible prototypes for vertical_compose functions
-typedef void (*vertical_compose_2tap)(IDWTELEM *b0, IDWTELEM *b1, int width);
-typedef void (*vertical_compose_3tap)(IDWTELEM *b0, IDWTELEM *b1, IDWTELEM *b2, int width);
-typedef void (*vertical_compose_5tap)(IDWTELEM *b0, IDWTELEM *b1, IDWTELEM *b2, IDWTELEM *b3, IDWTELEM *b4, int width);
-typedef void (*vertical_compose_9tap)(IDWTELEM *dst, IDWTELEM *b[8], int width);
-
-typedef struct DWTContext {
-    IDWTELEM *buffer;
-    IDWTELEM *temp;
-    int width;
-    int height;
-    int stride;
-    int decomposition_count;
-    int support;
-
-    void (*spatial_compose)(struct DWTContext *cs, int level, int width, int height, int stride);
-    void (*vertical_compose_l0)(void);
-    void (*vertical_compose_h0)(void);
-    void (*vertical_compose_l1)(void);
-    void (*vertical_compose_h1)(void);
-    void (*vertical_compose)(void);     ///< one set of lowpass and highpass combined
-    void (*horizontal_compose)(IDWTELEM *b, IDWTELEM *tmp, int width);
+struct SnowDWTContext;
 
+typedef struct SnowDWTContext {
     void (*vertical_compose97i)(IDWTELEM *b0, IDWTELEM *b1, IDWTELEM *b2,
                                 IDWTELEM *b3, IDWTELEM *b4, IDWTELEM *b5,
                                 int width);
@@ -84,70 +59,7 @@ typedef struct DWTContext {
                              uint8_t **block, int b_w, int b_h, int src_x,
                              int src_y, int src_stride, slice_buffer *sb,
                              int add, uint8_t *dst8);
-
-    DWTCompose cs[MAX_DECOMPOSITIONS];
-} DWTContext;
-
-enum dwt_type {
-    DWT_SNOW_DAUB9_7,
-    DWT_SNOW_LEGALL5_3,
-    DWT_DIRAC_DD9_7,
-    DWT_DIRAC_LEGALL5_3,
-    DWT_DIRAC_DD13_7,
-    DWT_DIRAC_HAAR0,
-    DWT_DIRAC_HAAR1,
-    DWT_DIRAC_FIDELITY,
-    DWT_DIRAC_DAUB9_7,
-    DWT_NUM_TYPES
-};
-
-// -1 if an error occurred, e.g. the dwt_type isn't recognized
-int ff_spatial_idwt_init2(DWTContext *d, IDWTELEM *buffer, int width, int height,
-                          int stride, enum dwt_type type, int decomposition_count,
-                          IDWTELEM *temp);
-
-int ff_spatial_idwt2(IDWTELEM *buffer, int width, int height, int stride,
-                     enum dwt_type type, int decomposition_count, IDWTELEM *temp);
-
-void ff_spatial_idwt_slice2(DWTContext *d, int y);
-
-// shared stuff for simd optimiztions
-#define COMPOSE_53iL0(b0, b1, b2)\
-    (b1 - ((b0 + b2 + 2) >> 2))
-
-#define COMPOSE_DIRAC53iH0(b0, b1, b2)\
-    (b1 + ((b0 + b2 + 1) >> 1))
-
-#define COMPOSE_DD97iH0(b0, b1, b2, b3, b4)\
-    (b2 + ((-b0 + 9*b1 + 9*b3 - b4 + 8) >> 4))
-
-#define COMPOSE_DD137iL0(b0, b1, b2, b3, b4)\
-    (b2 - ((-b0 + 9*b1 + 9*b3 - b4 + 16) >> 5))
-
-#define COMPOSE_HAARiL0(b0, b1)\
-    (b0 - ((b1 + 1) >> 1))
-
-#define COMPOSE_HAARiH0(b0, b1)\
-    (b0 + b1)
-
-#define COMPOSE_FIDELITYiL0(b0, b1, b2, b3, b4, b5, b6, b7, b8)\
-    (b4 - ((-8*(b0+b8) + 21*(b1+b7) - 46*(b2+b6) + 161*(b3+b5) + 128) >> 8))
-
-#define COMPOSE_FIDELITYiH0(b0, b1, b2, b3, b4, b5, b6, b7, b8)\
-    (b4 + ((-2*(b0+b8) + 10*(b1+b7) - 25*(b2+b6) + 81*(b3+b5) + 128) >> 8))
-
-#define COMPOSE_DAUB97iL1(b0, b1, b2)\
-    (b1 - ((1817*(b0 + b2) + 2048) >> 12))
-
-#define COMPOSE_DAUB97iH1(b0, b1, b2)\
-    (b1 - (( 113*(b0 + b2) + 64) >> 7))
-
-#define COMPOSE_DAUB97iL0(b0, b1, b2)\
-    (b1 + (( 217*(b0 + b2) + 2048) >> 12))
-
-#define COMPOSE_DAUB97iH0(b0, b1, b2)\
-    (b1 + ((6497*(b0 + b2) + 2048) >> 12))
-
+} SnowDWTContext;
 
 
 #define DWT_97 0
@@ -202,14 +114,14 @@ void ff_spatial_dwt(int *buffer, int *temp, int width, int height, int stride,
 void ff_spatial_idwt_buffered_init(DWTCompose *cs, slice_buffer *sb, int width,
                                    int height, int stride_line, int type,
                                    int decomposition_count);
-void ff_spatial_idwt_buffered_slice(DWTContext *dsp, DWTCompose *cs,
+void ff_spatial_idwt_buffered_slice(SnowDWTContext *dsp, DWTCompose *cs,
                                     slice_buffer *slice_buf, IDWTELEM *temp,
                                     int width, int height, int stride_line,
                                     int type, int decomposition_count, int y);
 void ff_spatial_idwt(IDWTELEM *buffer, IDWTELEM *temp, int width, int height,
                      int stride, int type, int decomposition_count);
 
-void ff_dwt_init(DWTContext *c);
-void ff_dwt_init_x86(DWTContext *c);
+void ff_dwt_init(SnowDWTContext *c);
+void ff_dwt_init_x86(SnowDWTContext *c);
 
 #endif /* AVCODEC_DWT_H */
diff --git a/libavcodec/snowdec.c b/libavcodec/snowdec.c
index c4baf0a7d1..e2115068c8 100644
--- a/libavcodec/snowdec.c
+++ b/libavcodec/snowdec.c
@@ -23,7 +23,7 @@
 #include "libavutil/opt.h"
 #include "avcodec.h"
 #include "dsputil.h"
-#include "dwt.h"
+#include "snow_dwt.h"
 #include "internal.h"
 #include "snow.h"
 
diff --git a/libavcodec/snowenc.c b/libavcodec/snowenc.c
index de2c859302..a062ac508d 100644
--- a/libavcodec/snowenc.c
+++ b/libavcodec/snowenc.c
@@ -25,7 +25,7 @@
 #include "internal.h"
 #include "dsputil.h"
 #include "internal.h"
-#include "dwt.h"
+#include "snow_dwt.h"
 #include "snow.h"
 
 #include "rangecoder.h"
diff --git a/libavcodec/x86/Makefile b/libavcodec/x86/Makefile
index 34254e4fde..6f10d060f9 100644
--- a/libavcodec/x86/Makefile
+++ b/libavcodec/x86/Makefile
@@ -35,17 +35,17 @@ MMX-OBJS                               += x86/dsputil_mmx.o             \
                                           x86/idct_sse2_xvid.o          \
                                           x86/simple_idct.o             \
 
-MMX-OBJS-$(CONFIG_DWT)                 += x86/snowdsp.o \
-                                          x86/dwt.o
 MMX-OBJS-$(CONFIG_ENCODERS)            += x86/dsputilenc_mmx.o          \
                                           x86/motion_est.o
+MMX-OBJS-$(CONFIG_SNOW_DECODER)        += x86/snowdsp.o
+MMX-OBJS-$(CONFIG_SNOW_ENCODER)        += x86/snowdsp.o
 MMX-OBJS-$(CONFIG_VC1_DECODER)         += x86/vc1dsp_mmx.o
 
 YASM-OBJS-$(CONFIG_AAC_DECODER)        += x86/sbrdsp.o
 YASM-OBJS-$(CONFIG_AC3DSP)             += x86/ac3dsp.o
 YASM-OBJS-$(CONFIG_DCT)                += x86/dct32.o
-YASM-OBJS-$(CONFIG_DIRAC_DECODER)      += x86/diracdsp_mmx.o x86/diracdsp_yasm.o
-YASM-OBJS-$(CONFIG_DWT)                += x86/dwt_yasm.o
+YASM-OBJS-$(CONFIG_DIRAC_DECODER)      += x86/diracdsp_mmx.o x86/diracdsp_yasm.o\
+                                          x86/dwt_yasm.o x86/dirac_dwt.o
 YASM-OBJS-$(CONFIG_ENCODERS)           += x86/dsputilenc.o
 YASM-OBJS-$(CONFIG_FFT)                += x86/fft.o
 YASM-OBJS-$(CONFIG_H264CHROMA)         += x86/h264_chromamc.o           \
diff --git a/libavcodec/x86/dwt.c b/libavcodec/x86/dirac_dwt.c
index e718c80fef..fbb25a417c 100644
--- a/libavcodec/x86/dwt.c
+++ b/libavcodec/x86/dirac_dwt.c
@@ -22,7 +22,7 @@
 
 #include "libavutil/x86/asm.h"
 #include "dsputil_mmx.h"
-#include "dwt.h"
+#include "dirac_dwt.h"
 
 #define COMPOSE_VERTICAL(ext, align) \
 void ff_vertical_compose53iL0##ext(IDWTELEM *b0, IDWTELEM *b1, IDWTELEM *b2, int width); \
diff --git a/libavcodec/x86/dwt.h b/libavcodec/x86/dirac_dwt.h
index 199f61175b..126b29029f 100644
--- a/libavcodec/x86/dwt.h
+++ b/libavcodec/x86/dirac_dwt.h
@@ -16,10 +16,10 @@
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  */
 
-#ifndef AVCODEC_X86_DWT_H
-#define AVCODEC_X86_DWT_H
+#ifndef AVCODEC_X86_DIRAC_DWT_H
+#define AVCODEC_X86_DIRAC_DWT_H
 
-#include "libavcodec/dwt.h"
+#include "libavcodec/dirac_dwt.h"
 
 void ff_horizontal_compose_dd97i_end_c(IDWTELEM *b, IDWTELEM *tmp, int w2, int x);
 void ff_horizontal_compose_haar1i_end_c(IDWTELEM *b, IDWTELEM *tmp, int w2, int x);
diff --git a/libavcodec/x86/snowdsp.c b/libavcodec/x86/snowdsp.c
index 631291aafe..5505ee8bfd 100644
--- a/libavcodec/x86/snowdsp.c
+++ b/libavcodec/x86/snowdsp.c
@@ -23,7 +23,7 @@
 #include "libavutil/x86/asm.h"
 #include "libavcodec/avcodec.h"
 #include "libavcodec/snow.h"
-#include "libavcodec/dwt.h"
+#include "libavcodec/snow_dwt.h"
 #include "dsputil_mmx.h"
 
 #if HAVE_INLINE_ASM
@@ -875,7 +875,7 @@ static void ff_snow_inner_add_yblock_mmx(const uint8_t *obmc, const int obmc_str
 
 #endif /* HAVE_INLINE_ASM */
 
-void ff_dwt_init_x86(DWTContext *c)
+void ff_dwt_init_x86(SnowDWTContext *c)
 {
 #if HAVE_INLINE_ASM
     int mm_flags = av_get_cpu_flags();