Merge commit '88bd7fdc821aaa0cbcf44cf075c62aaa42121e3f'

* commit '88bd7fdc821aaa0cbcf44cf075c62aaa42121e3f':
  Drop DCTELEM typedef

Conflicts:
	libavcodec/alpha/dsputil_alpha.h
	libavcodec/alpha/motion_est_alpha.c
	libavcodec/arm/dsputil_init_armv6.c
	libavcodec/bfin/dsputil_bfin.h
	libavcodec/bfin/pixels_bfin.S
	libavcodec/cavs.c
	libavcodec/cavsdec.c
	libavcodec/dct-test.c
	libavcodec/dnxhdenc.c
	libavcodec/dsputil.c
	libavcodec/dsputil.h
	libavcodec/dsputil_template.c
	libavcodec/eamad.c
	libavcodec/h264_cavlc.c
	libavcodec/h264idct_template.c
	libavcodec/mpeg12.c
	libavcodec/mpegvideo.c
	libavcodec/mpegvideo.h
	libavcodec/mpegvideo_enc.c
	libavcodec/ppc/dsputil_altivec.c
	libavcodec/proresdsp.c

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

169 files changed, 929 insertions, 907 deletions

diff --git a/doc/optimization.txt b/doc/optimization.txt
index e3fd32401c..5a66d6bb3f 100644
--- a/doc/optimization.txt
+++ b/doc/optimization.txt
@@ -148,7 +148,7 @@ Alignment:
 Some instructions on some architectures have strict alignment restrictions,
 for example most SSE/SSE2 instructions on x86.
 The minimum guaranteed alignment is written in the .h files, for example:
-    void (*put_pixels_clamped)(const DCTELEM *block/*align 16*/, UINT8 *pixels/*align 8*/, int line_size);
+    void (*put_pixels_clamped)(const int16_t *block/*align 16*/, UINT8 *pixels/*align 8*/, int line_size);
 
 
 General Tips:
diff --git a/libavcodec/4xm.c b/libavcodec/4xm.c
index 9d2b4f54c6..2db3026756 100644
--- a/libavcodec/4xm.c
+++ b/libavcodec/4xm.c
@@ -138,7 +138,7 @@ typedef struct FourXContext {
     int mv[256];
     VLC pre_vlc;
     int last_dc;
-    DECLARE_ALIGNED(16, DCTELEM, block)[6][64];
+    DECLARE_ALIGNED(16, int16_t, block)[6][64];
     void *bitstream_buffer;
     unsigned int bitstream_buffer_size;
     int version;
@@ -153,7 +153,7 @@ typedef struct FourXContext {
 
 #define MULTIPLY(var, const) (((var) * (const)) >> 16)
 
-static void idct(DCTELEM block[64])
+static void idct(int16_t block[64])
 {
     int tmp0, tmp1, tmp2, tmp3, tmp4, tmp5, tmp6, tmp7;
     int tmp10, tmp11, tmp12, tmp13;
@@ -471,7 +471,7 @@ static int decode_p_frame(FourXContext *f, const uint8_t *buf, int length)
  * decode block and dequantize.
  * Note this is almost identical to MJPEG.
  */
-static int decode_i_block(FourXContext *f, DCTELEM *block)
+static int decode_i_block(FourXContext *f, int16_t *block)
 {
     int code, i, j, level, val;
 
@@ -521,7 +521,7 @@ static int decode_i_block(FourXContext *f, DCTELEM *block)
 
 static inline void idct_put(FourXContext *f, int x, int y)
 {
-    DCTELEM (*block)[64] = f->block;
+    int16_t (*block)[64] = f->block;
     int stride           = f->current_picture->linesize[0] >> 1;
     int i;
     uint16_t *dst = ((uint16_t*)f->current_picture->data[0]) + y * stride + x;
@@ -542,7 +542,7 @@ static inline void idct_put(FourXContext *f, int x, int y)
      * cr = (-1b - 4g + 5r) / 14 */
     for (y = 0; y < 8; y++) {
         for (x = 0; x < 8; x++) {
-            DCTELEM *temp = block[(x >> 2) + 2 * (y >> 2)] +
+            int16_t *temp = block[(x >> 2) + 2 * (y >> 2)] +
                             2 * (x & 3) + 2 * 8 * (y & 3); // FIXME optimize
             int cb = block[4][x + 8 * y];
             int cr = block[5][x + 8 * y];
diff --git a/libavcodec/alpha/dsputil_alpha.c b/libavcodec/alpha/dsputil_alpha.c
index 38ac0209a8..7cfc1cb62b 100644
--- a/libavcodec/alpha/dsputil_alpha.c
+++ b/libavcodec/alpha/dsputil_alpha.c
@@ -23,15 +23,15 @@
 #include "dsputil_alpha.h"
 #include "asm.h"
 
-void (*put_pixels_clamped_axp_p)(const DCTELEM *block, uint8_t *pixels,
+void (*put_pixels_clamped_axp_p)(const int16_t *block, uint8_t *pixels,
                                  int line_size);
-void (*add_pixels_clamped_axp_p)(const DCTELEM *block, uint8_t *pixels,
+void (*add_pixels_clamped_axp_p)(const int16_t *block, uint8_t *pixels,
                                  int line_size);
 
 #if 0
 /* These functions were the base for the optimized assembler routines,
    and remain here for documentation purposes.  */
-static void put_pixels_clamped_mvi(const DCTELEM *block, uint8_t *pixels,
+static void put_pixels_clamped_mvi(const int16_t *block, uint8_t *pixels,
                                    int line_size)
 {
     int i = 8;
@@ -55,7 +55,7 @@ static void put_pixels_clamped_mvi(const DCTELEM *block, uint8_t *pixels,
     } while (--i);
 }
 
-void add_pixels_clamped_mvi(const DCTELEM *block, uint8_t *pixels,
+void add_pixels_clamped_mvi(const int16_t *block, uint8_t *pixels,
                             int line_size)
 {
     int h = 8;
@@ -100,9 +100,9 @@ void add_pixels_clamped_mvi(const DCTELEM *block, uint8_t *pixels,
 }
 #endif
 
-static void clear_blocks_axp(DCTELEM *blocks) {
+static void clear_blocks_axp(int16_t *blocks) {
     uint64_t *p = (uint64_t *) blocks;
-    int n = sizeof(DCTELEM) * 6 * 64;
+    int n = sizeof(int16_t) * 6 * 64;
 
     do {
         p[0] = 0;
diff --git a/libavcodec/alpha/dsputil_alpha.h b/libavcodec/alpha/dsputil_alpha.h
index 4bc09a497d..cdb51358de 100644
--- a/libavcodec/alpha/dsputil_alpha.h
+++ b/libavcodec/alpha/dsputil_alpha.h
@@ -19,26 +19,26 @@
 #ifndef AVCODEC_ALPHA_DSPUTIL_ALPHA_H
 #define AVCODEC_ALPHA_DSPUTIL_ALPHA_H
 
-#include "libavcodec/dsputil.h"
+#include <stdint.h>
 
-void ff_simple_idct_axp(DCTELEM *block);
-void ff_simple_idct_put_axp(uint8_t *dest, int line_size, DCTELEM *block);
-void ff_simple_idct_add_axp(uint8_t *dest, int line_size, DCTELEM *block);
+void ff_simple_idct_axp(int16_t *block);
+void ff_simple_idct_put_axp(uint8_t *dest, int line_size, int16_t *block);
+void ff_simple_idct_add_axp(uint8_t *dest, int line_size, int16_t *block);
 
 void put_pixels_axp_asm(uint8_t *block, const uint8_t *pixels,
                         int line_size, int h);
-void put_pixels_clamped_mvi_asm(const DCTELEM *block, uint8_t *pixels,
+void put_pixels_clamped_mvi_asm(const int16_t *block, uint8_t *pixels,
                                 int line_size);
-void add_pixels_clamped_mvi_asm(const DCTELEM *block, uint8_t *pixels,
+void add_pixels_clamped_mvi_asm(const int16_t *block, uint8_t *pixels,
                                 int line_size);
-extern void (*put_pixels_clamped_axp_p)(const DCTELEM *block, uint8_t *pixels,
+extern void (*put_pixels_clamped_axp_p)(const int16_t *block, uint8_t *pixels,
                                         int line_size);
-extern void (*add_pixels_clamped_axp_p)(const DCTELEM *block, uint8_t *pixels,
+extern void (*add_pixels_clamped_axp_p)(const int16_t *block, uint8_t *pixels,
                                         int line_size);
 
-void get_pixels_mvi(DCTELEM *av_restrict block,
-                    const uint8_t *av_restrict pixels, int line_size);
-void diff_pixels_mvi(DCTELEM *block, const uint8_t *s1, const uint8_t *s2,
+void get_pixels_mvi(int16_t *restrict block,
+                    const uint8_t *restrict pixels, int line_size);
+void diff_pixels_mvi(int16_t *block, const uint8_t *s1, const uint8_t *s2,
                      int stride);
 int pix_abs8x8_mvi(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h);
 int pix_abs16x16_mvi_asm(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h);
diff --git a/libavcodec/alpha/dsputil_alpha_asm.S b/libavcodec/alpha/dsputil_alpha_asm.S
index 32a8bc9562..557ba57a95 100644
--- a/libavcodec/alpha/dsputil_alpha_asm.S
+++ b/libavcodec/alpha/dsputil_alpha_asm.S
@@ -135,7 +135,7 @@ $aligned:
         .end put_pixels_axp_asm
 
 /************************************************************************
- * void put_pixels_clamped_mvi_asm(const DCTELEM *block, uint8_t *pixels,
+ * void put_pixels_clamped_mvi_asm(const int16_t *block, uint8_t *pixels,
  *                                 int line_size)
  */
         .align 6
@@ -185,7 +185,7 @@ put_pixels_clamped_mvi_asm:
         .end put_pixels_clamped_mvi_asm
 
 /************************************************************************
- * void add_pixels_clamped_mvi_asm(const DCTELEM *block, uint8_t *pixels,
+ * void add_pixels_clamped_mvi_asm(const int16_t *block, uint8_t *pixels,
  *                                 int line_size)
  */
         .align 6
diff --git a/libavcodec/alpha/motion_est_alpha.c b/libavcodec/alpha/motion_est_alpha.c
index 927b25d992..9883feea85 100644
--- a/libavcodec/alpha/motion_est_alpha.c
+++ b/libavcodec/alpha/motion_est_alpha.c
@@ -23,8 +23,8 @@
 #include "dsputil_alpha.h"
 #include "asm.h"
 
-void get_pixels_mvi(DCTELEM *av_restrict block,
-                    const uint8_t *av_restrict pixels, int line_size)
+void get_pixels_mvi(int16_t *restrict block,
+                    const uint8_t *restrict pixels, int line_size)
 {
     int h = 8;
 
@@ -40,7 +40,7 @@ void get_pixels_mvi(DCTELEM *av_restrict block,
     } while (--h);
 }
 
-void diff_pixels_mvi(DCTELEM *block, const uint8_t *s1, const uint8_t *s2,
+void diff_pixels_mvi(int16_t *block, const uint8_t *s1, const uint8_t *s2,
                      int stride) {
     int h = 8;
     uint64_t mask = 0x4040;
diff --git a/libavcodec/alpha/mpegvideo_alpha.c b/libavcodec/alpha/mpegvideo_alpha.c
index 28eca072c0..2e8cd79a84 100644
--- a/libavcodec/alpha/mpegvideo_alpha.c
+++ b/libavcodec/alpha/mpegvideo_alpha.c
@@ -23,7 +23,7 @@
 #include "libavcodec/mpegvideo.h"
 #include "asm.h"
 
-static void dct_unquantize_h263_axp(DCTELEM *block, int n_coeffs,
+static void dct_unquantize_h263_axp(int16_t *block, int n_coeffs,
                                     uint64_t qscale, uint64_t qadd)
 {
     uint64_t qmul = qscale << 1;
@@ -69,12 +69,12 @@ static void dct_unquantize_h263_axp(DCTELEM *block, int n_coeffs,
     }
 }
 
-static void dct_unquantize_h263_intra_axp(MpegEncContext *s, DCTELEM *block,
+static void dct_unquantize_h263_intra_axp(MpegEncContext *s, int16_t *block,
                                     int n, int qscale)
 {
     int n_coeffs;
     uint64_t qadd;
-    DCTELEM block0 = block[0];
+    int16_t block0 = block[0];
 
     if (!s->h263_aic) {
         if (n < 4)
@@ -96,7 +96,7 @@ static void dct_unquantize_h263_intra_axp(MpegEncContext *s, DCTELEM *block,
     block[0] = block0;
 }
 
-static void dct_unquantize_h263_inter_axp(MpegEncContext *s, DCTELEM *block,
+static void dct_unquantize_h263_inter_axp(MpegEncContext *s, int16_t *block,
                                     int n, int qscale)
 {
     int n_coeffs = s->inter_scantable.raster_end[s->block_last_index[n]];
diff --git a/libavcodec/alpha/simple_idct_alpha.c b/libavcodec/alpha/simple_idct_alpha.c
index 522efd2b4d..1a272bddaf 100644
--- a/libavcodec/alpha/simple_idct_alpha.c
+++ b/libavcodec/alpha/simple_idct_alpha.c
@@ -44,7 +44,7 @@
 #define COL_SHIFT 20
 
 /* 0: all entries 0, 1: only first entry nonzero, 2: otherwise  */
-static inline int idct_row(DCTELEM *row)
+static inline int idct_row(int16_t *row)
 {
     int a0, a1, a2, a3, b0, b1, b2, b3, t;
     uint64_t l, r, t2;
@@ -152,7 +152,7 @@ static inline int idct_row(DCTELEM *row)
     return 2;
 }
 
-static inline void idct_col(DCTELEM *col)
+static inline void idct_col(int16_t *col)
 {
     int a0, a1, a2, a3, b0, b1, b2, b3;
 
@@ -229,7 +229,7 @@ static inline void idct_col(DCTELEM *col)
 
 /* If all rows but the first one are zero after row transformation,
    all rows will be identical after column transformation.  */
-static inline void idct_col2(DCTELEM *col)
+static inline void idct_col2(int16_t *col)
 {
     int i;
     uint64_t l, r;
@@ -251,7 +251,7 @@ static inline void idct_col2(DCTELEM *col)
     stq(l, col + 14 * 4); stq(r, col + 15 * 4);
 }
 
-void ff_simple_idct_axp(DCTELEM *block)
+void ff_simple_idct_axp(int16_t *block)
 {
 
     int i;
@@ -291,13 +291,13 @@ void ff_simple_idct_axp(DCTELEM *block)
     }
 }
 
-void ff_simple_idct_put_axp(uint8_t *dest, int line_size, DCTELEM *block)
+void ff_simple_idct_put_axp(uint8_t *dest, int line_size, int16_t *block)
 {
     ff_simple_idct_axp(block);
     put_pixels_clamped_axp_p(block, dest, line_size);
 }
 
-void ff_simple_idct_add_axp(uint8_t *dest, int line_size, DCTELEM *block)
+void ff_simple_idct_add_axp(uint8_t *dest, int line_size, int16_t *block)
 {
     ff_simple_idct_axp(block);
     add_pixels_clamped_axp_p(block, dest, line_size);
diff --git a/libavcodec/arm/dsputil_init_arm.c b/libavcodec/arm/dsputil_init_arm.c
index 45192d1b5c..ea75c3cc12 100644
--- a/libavcodec/arm/dsputil_init_arm.c
+++ b/libavcodec/arm/dsputil_init_arm.c
@@ -23,12 +23,12 @@
 #include "libavcodec/dsputil.h"
 #include "dsputil_arm.h"
 
-void ff_j_rev_dct_arm(DCTELEM *data);
-void ff_simple_idct_arm(DCTELEM *data);
+void ff_j_rev_dct_arm(int16_t *data);
+void ff_simple_idct_arm(int16_t *data);
 
 /* XXX: local hack */
-static void (*ff_put_pixels_clamped)(const DCTELEM *block, uint8_t *pixels, int line_size);
-static void (*ff_add_pixels_clamped)(const DCTELEM *block, uint8_t *pixels, int line_size);
+static void (*ff_put_pixels_clamped)(const int16_t *block, uint8_t *pixels, int line_size);
+static void (*ff_add_pixels_clamped)(const int16_t *block, uint8_t *pixels, int line_size);
 
 void ff_put_pixels8_arm(uint8_t *block, const uint8_t *pixels, int line_size, int h);
 void ff_put_pixels8_x2_arm(uint8_t *block, const uint8_t *pixels, int line_size, int h);
@@ -48,27 +48,27 @@ CALL_2X_PIXELS(ff_put_no_rnd_pixels16_x2_arm,  ff_put_no_rnd_pixels8_x2_arm, 8)
 CALL_2X_PIXELS(ff_put_no_rnd_pixels16_y2_arm,  ff_put_no_rnd_pixels8_y2_arm, 8)
 CALL_2X_PIXELS(ff_put_no_rnd_pixels16_xy2_arm, ff_put_no_rnd_pixels8_xy2_arm,8)
 
-void ff_add_pixels_clamped_arm(const DCTELEM *block, uint8_t *dest,
+void ff_add_pixels_clamped_arm(const int16_t *block, uint8_t *dest,
                                int line_size);
 
 /* XXX: those functions should be suppressed ASAP when all IDCTs are
    converted */
-static void j_rev_dct_arm_put(uint8_t *dest, int line_size, DCTELEM *block)
+static void j_rev_dct_arm_put(uint8_t *dest, int line_size, int16_t *block)
 {
     ff_j_rev_dct_arm (block);
     ff_put_pixels_clamped(block, dest, line_size);
 }
-static void j_rev_dct_arm_add(uint8_t *dest, int line_size, DCTELEM *block)
+static void j_rev_dct_arm_add(uint8_t *dest, int line_size, int16_t *block)
 {
     ff_j_rev_dct_arm (block);
     ff_add_pixels_clamped(block, dest, line_size);
 }
-static void simple_idct_arm_put(uint8_t *dest, int line_size, DCTELEM *block)
+static void simple_idct_arm_put(uint8_t *dest, int line_size, int16_t *block)
 {
     ff_simple_idct_arm (block);
     ff_put_pixels_clamped(block, dest, line_size);
 }
-static void simple_idct_arm_add(uint8_t *dest, int line_size, DCTELEM *block)
+static void simple_idct_arm_add(uint8_t *dest, int line_size, int16_t *block)
 {
     ff_simple_idct_arm (block);
     ff_add_pixels_clamped(block, dest, line_size);
diff --git a/libavcodec/arm/dsputil_init_armv5te.c b/libavcodec/arm/dsputil_init_armv5te.c
index b4f3f0616e..33d7970e68 100644
--- a/libavcodec/arm/dsputil_init_armv5te.c
+++ b/libavcodec/arm/dsputil_init_armv5te.c
@@ -21,9 +21,9 @@
 #include "libavcodec/dsputil.h"
 #include "dsputil_arm.h"
 
-void ff_simple_idct_armv5te(DCTELEM *data);
-void ff_simple_idct_put_armv5te(uint8_t *dest, int line_size, DCTELEM *data);
-void ff_simple_idct_add_armv5te(uint8_t *dest, int line_size, DCTELEM *data);
+void ff_simple_idct_armv5te(int16_t *data);
+void ff_simple_idct_put_armv5te(uint8_t *dest, int line_size, int16_t *data);
+void ff_simple_idct_add_armv5te(uint8_t *dest, int line_size, int16_t *data);
 
 av_cold void ff_dsputil_init_armv5te(DSPContext *c, AVCodecContext *avctx)
 {
diff --git a/libavcodec/arm/dsputil_init_armv6.c b/libavcodec/arm/dsputil_init_armv6.c
index 1081135874..dc126fef8a 100644
--- a/libavcodec/arm/dsputil_init_armv6.c
+++ b/libavcodec/arm/dsputil_init_armv6.c
@@ -24,9 +24,9 @@
 #include "libavcodec/dsputil.h"
 #include "dsputil_arm.h"
 
-void ff_simple_idct_armv6(DCTELEM *data);
-void ff_simple_idct_put_armv6(uint8_t *dest, int line_size, DCTELEM *data);
-void ff_simple_idct_add_armv6(uint8_t *dest, int line_size, DCTELEM *data);
+void ff_simple_idct_armv6(int16_t *data);
+void ff_simple_idct_put_armv6(uint8_t *dest, int line_size, int16_t *data);
+void ff_simple_idct_add_armv6(uint8_t *dest, int line_size, int16_t *data);
 
 void ff_put_pixels16_armv6(uint8_t *, const uint8_t *, int, int);
 void ff_put_pixels16_x2_armv6(uint8_t *, const uint8_t *, int, int);
@@ -46,12 +46,12 @@ void ff_put_pixels8_y2_no_rnd_armv6(uint8_t *, const uint8_t *, int, int);
 
 void ff_avg_pixels8_armv6(uint8_t *, const uint8_t *, int, int);
 
-void ff_add_pixels_clamped_armv6(const DCTELEM *block,
-                                 uint8_t *av_restrict pixels,
+void ff_add_pixels_clamped_armv6(const int16_t *block,
+                                 uint8_t *restrict pixels,
                                  int line_size);
 
-void ff_get_pixels_armv6(DCTELEM *block, const uint8_t *pixels, int stride);
-void ff_diff_pixels_armv6(DCTELEM *block, const uint8_t *s1,
+void ff_get_pixels_armv6(int16_t *block, const uint8_t *pixels, int stride);
+void ff_diff_pixels_armv6(int16_t *block, const uint8_t *s1,
                           const uint8_t *s2, int stride);
 
 int ff_pix_abs16_armv6(void *s, uint8_t *blk1, uint8_t *blk2,
diff --git a/libavcodec/arm/dsputil_init_neon.c b/libavcodec/arm/dsputil_init_neon.c
index 09248cc5fe..6aa430daba 100644
--- a/libavcodec/arm/dsputil_init_neon.c
+++ b/libavcodec/arm/dsputil_init_neon.c
@@ -25,12 +25,12 @@
 #include "libavcodec/dsputil.h"
 #include "dsputil_arm.h"
 
-void ff_simple_idct_neon(DCTELEM *data);
-void ff_simple_idct_put_neon(uint8_t *dest, int line_size, DCTELEM *data);
-void ff_simple_idct_add_neon(uint8_t *dest, int line_size, DCTELEM *data);
+void ff_simple_idct_neon(int16_t *data);
+void ff_simple_idct_put_neon(uint8_t *dest, int line_size, int16_t *data);
+void ff_simple_idct_add_neon(uint8_t *dest, int line_size, int16_t *data);
 
-void ff_clear_block_neon(DCTELEM *block);
-void ff_clear_blocks_neon(DCTELEM *blocks);
+void ff_clear_block_neon(int16_t *block);
+void ff_clear_blocks_neon(int16_t *blocks);
 
 void ff_put_pixels16_neon(uint8_t *, const uint8_t *, int, int);
 void ff_put_pixels16_x2_neon(uint8_t *, const uint8_t *, int, int);
@@ -62,9 +62,9 @@ void ff_avg_pixels8_x2_no_rnd_neon(uint8_t *, const uint8_t *, int, int);
 void ff_avg_pixels8_y2_no_rnd_neon(uint8_t *, const uint8_t *, int, int);
 void ff_avg_pixels8_xy2_no_rnd_neon(uint8_t *, const uint8_t *, int, int);
 
-void ff_add_pixels_clamped_neon(const DCTELEM *, uint8_t *, int);
-void ff_put_pixels_clamped_neon(const DCTELEM *, uint8_t *, int);
-void ff_put_signed_pixels_clamped_neon(const DCTELEM *, uint8_t *, int);
+void ff_add_pixels_clamped_neon(const int16_t *, uint8_t *, int);
+void ff_put_pixels_clamped_neon(const int16_t *, uint8_t *, int);
+void ff_put_signed_pixels_clamped_neon(const int16_t *, uint8_t *, int);
 
 void ff_put_h264_qpel16_mc00_neon(uint8_t *, uint8_t *, int);
 void ff_put_h264_qpel16_mc10_neon(uint8_t *, uint8_t *, int);
diff --git a/libavcodec/arm/h264dsp_init_arm.c b/libavcodec/arm/h264dsp_init_arm.c
index f150a4508d..4695707cf7 100644
--- a/libavcodec/arm/h264dsp_init_arm.c
+++ b/libavcodec/arm/h264dsp_init_arm.c
@@ -50,22 +50,22 @@ void ff_biweight_h264_pixels_4_neon(uint8_t *dst, uint8_t *src, int stride,
                                     int height, int log2_den, int weightd,
                                     int weights, int offset);
 
-void ff_h264_idct_add_neon(uint8_t *dst, DCTELEM *block, int stride);
-void ff_h264_idct_dc_add_neon(uint8_t *dst, DCTELEM *block, int stride);
+void ff_h264_idct_add_neon(uint8_t *dst, int16_t *block, int stride);
+void ff_h264_idct_dc_add_neon(uint8_t *dst, int16_t *block, int stride);
 void ff_h264_idct_add16_neon(uint8_t *dst, const int *block_offset,
-                             DCTELEM *block, int stride,
+                             int16_t *block, int stride,
                              const uint8_t nnzc[6*8]);
 void ff_h264_idct_add16intra_neon(uint8_t *dst, const int *block_offset,
-                                  DCTELEM *block, int stride,
+                                  int16_t *block, int stride,
                                   const uint8_t nnzc[6*8]);
 void ff_h264_idct_add8_neon(uint8_t **dest, const int *block_offset,
-                            DCTELEM *block, int stride,
+                            int16_t *block, int stride,
                             const uint8_t nnzc[6*8]);
 
-void ff_h264_idct8_add_neon(uint8_t *dst, DCTELEM *block, int stride);
-void ff_h264_idct8_dc_add_neon(uint8_t *dst, DCTELEM *block, int stride);
+void ff_h264_idct8_add_neon(uint8_t *dst, int16_t *block, int stride);
+void ff_h264_idct8_dc_add_neon(uint8_t *dst, int16_t *block, int stride);
 void ff_h264_idct8_add4_neon(uint8_t *dst, const int *block_offset,
-                             DCTELEM *block, int stride,
+                             int16_t *block, int stride,
                              const uint8_t nnzc[6*8]);
 
 static void ff_h264dsp_init_neon(H264DSPContext *c, const int bit_depth, const int chroma_format_idc)
diff --git a/libavcodec/arm/jrevdct_arm.S b/libavcodec/arm/jrevdct_arm.S
index 331ad85328..f951e2af34 100644
--- a/libavcodec/arm/jrevdct_arm.S
+++ b/libavcodec/arm/jrevdct_arm.S
@@ -66,7 +66,7 @@ row_loop:
         ldrsh r2, [lr, # 2]             @ r2 = 'd2'
 
         @ Optimization for row that have all items except the first set to 0
-        @ (this works as the DCTELEMS are always 4-byte aligned)
+        @ (this works as the int16_t are always 4-byte aligned)
         ldr r5, [lr, # 0]
         ldr r6, [lr, # 4]
         ldr r3, [lr, # 8]
diff --git a/libavcodec/arm/mpegvideo_arm.c b/libavcodec/arm/mpegvideo_arm.c
index 28838bb039..2e4f156d62 100644
--- a/libavcodec/arm/mpegvideo_arm.c
+++ b/libavcodec/arm/mpegvideo_arm.c
@@ -34,9 +34,9 @@ CHK_OFFS(MpegEncContext, inter_scantable.raster_end, INTER_SCANTAB_RASTER_END);
 CHK_OFFS(MpegEncContext, h263_aic,         H263_AIC);
 #endif
 
-void ff_dct_unquantize_h263_inter_neon(MpegEncContext *s, DCTELEM *block,
+void ff_dct_unquantize_h263_inter_neon(MpegEncContext *s, int16_t *block,
                                        int n, int qscale);
-void ff_dct_unquantize_h263_intra_neon(MpegEncContext *s, DCTELEM *block,
+void ff_dct_unquantize_h263_intra_neon(MpegEncContext *s, int16_t *block,
                                        int n, int qscale);
 
 void ff_MPV_common_init_arm(MpegEncContext *s)
diff --git a/libavcodec/arm/mpegvideo_armv5te.c b/libavcodec/arm/mpegvideo_armv5te.c
index 842620bad7..832c541084 100644
--- a/libavcodec/arm/mpegvideo_armv5te.c
+++ b/libavcodec/arm/mpegvideo_armv5te.c
@@ -25,7 +25,7 @@
 #include "libavcodec/mpegvideo.h"
 #include "mpegvideo_arm.h"
 
-void ff_dct_unquantize_h263_armv5te(DCTELEM *block, int qmul, int qadd, int count);
+void ff_dct_unquantize_h263_armv5te(int16_t *block, int qmul, int qadd, int count);
 
 #ifdef ENABLE_ARM_TESTS
 /**
@@ -33,7 +33,7 @@ void ff_dct_unquantize_h263_armv5te(DCTELEM *block, int qmul, int qadd, int coun
  * have optimized implementations for each architecture. Is also used as a reference
  * implementation in regression tests
  */
-static inline void dct_unquantize_h263_helper_c(DCTELEM *block, int qmul, int qadd, int count)
+static inline void dct_unquantize_h263_helper_c(int16_t *block, int qmul, int qadd, int count)
 {
     int i, level;
     for (i = 0; i < count; i++) {
@@ -51,7 +51,7 @@ static inline void dct_unquantize_h263_helper_c(DCTELEM *block, int qmul, int qa
 #endif
 
 static void dct_unquantize_h263_intra_armv5te(MpegEncContext *s,
-                                  DCTELEM *block, int n, int qscale)
+                                  int16_t *block, int n, int qscale)
 {
     int level, qmul, qadd;
     int nCoeffs;
@@ -80,7 +80,7 @@ static void dct_unquantize_h263_intra_armv5te(MpegEncContext *s,
 }
 
 static void dct_unquantize_h263_inter_armv5te(MpegEncContext *s,
-                                  DCTELEM *block, int n, int qscale)
+                                  int16_t *block, int n, int qscale)
 {
     int qmul, qadd;
     int nCoeffs;
diff --git a/libavcodec/arm/rv34dsp_init_arm.c b/libavcodec/arm/rv34dsp_init_arm.c
index 60ab82eed4..d3ce7a0516 100644
--- a/libavcodec/arm/rv34dsp_init_arm.c
+++ b/libavcodec/arm/rv34dsp_init_arm.c
@@ -24,11 +24,11 @@
 #include "libavcodec/rv34dsp.h"
 #include "libavutil/arm/cpu.h"
 
-void ff_rv34_inv_transform_noround_neon(DCTELEM *block);
+void ff_rv34_inv_transform_noround_neon(int16_t *block);
 
-void ff_rv34_inv_transform_noround_dc_neon(DCTELEM *block);
+void ff_rv34_inv_transform_noround_dc_neon(int16_t *block);
 
-void ff_rv34_idct_add_neon(uint8_t *dst, ptrdiff_t stride, DCTELEM *block);
+void ff_rv34_idct_add_neon(uint8_t *dst, ptrdiff_t stride, int16_t *block);
 void ff_rv34_idct_dc_add_neon(uint8_t *dst, ptrdiff_t stride, int dc);
 
 void ff_rv34dsp_init_arm(RV34DSPContext *c, DSPContext* dsp)
diff --git a/libavcodec/arm/rv34dsp_neon.S b/libavcodec/arm/rv34dsp_neon.S
index 522e387582..a29123f772 100644
--- a/libavcodec/arm/rv34dsp_neon.S
+++ b/libavcodec/arm/rv34dsp_neon.S
@@ -67,7 +67,7 @@
         vsub.s32        q15, q14, q9    @ z0 - z3
 .endm
 
-/* void rv34_idct_add_c(uint8_t *dst, int stride, DCTELEM *block) */
+/* void rv34_idct_add_c(uint8_t *dst, int stride, int16_t *block) */
 function ff_rv34_idct_add_neon, export=1
         mov             r3,  r0
         rv34_inv_transform   r2
@@ -97,7 +97,7 @@ function ff_rv34_idct_add_neon, export=1
         bx              lr
 endfunc
 
-/* void rv34_inv_transform_noround_neon(DCTELEM *block); */
+/* void rv34_inv_transform_noround_neon(int16_t *block); */
 function ff_rv34_inv_transform_noround_neon, export=1
         rv34_inv_transform   r0
         vshl.s32        q11, q2,  #1
@@ -142,7 +142,7 @@ function ff_rv34_idct_dc_add_neon, export=1
         bx              lr
 endfunc
 
-/* void rv34_inv_transform_dc_noround_c(DCTELEM *block) */
+/* void rv34_inv_transform_dc_noround_c(int16_t *block) */
 function ff_rv34_inv_transform_noround_dc_neon, export=1
         vld1.16         {d28[]}, [r0,:16]       @ block[0]
         vmov.i16        d4,  #251
diff --git a/libavcodec/arm/simple_idct_armv6.S b/libavcodec/arm/simple_idct_armv6.S
index 0c19d267a8..79cf5d41fb 100644
--- a/libavcodec/arm/simple_idct_armv6.S
+++ b/libavcodec/arm/simple_idct_armv6.S
@@ -375,7 +375,7 @@ endfunc
         sub    r0, r0, #(16*7)
         .endm
 
-/* void ff_simple_idct_armv6(DCTELEM *data); */
+/* void ff_simple_idct_armv6(int16_t *data); */
 function ff_simple_idct_armv6, export=1
         push   {r4-r11, lr}
         sub    sp, sp, #128
@@ -390,7 +390,7 @@ function ff_simple_idct_armv6, export=1
         pop    {r4-r11, pc}
 endfunc
 
-/* ff_simple_idct_add_armv6(uint8_t *dest, int line_size, DCTELEM *data); */
+/* ff_simple_idct_add_armv6(uint8_t *dest, int line_size, int16_t *data); */
 function ff_simple_idct_add_armv6, export=1
         push   {r0, r1, r4-r11, lr}
         sub    sp, sp, #128
@@ -407,7 +407,7 @@ function ff_simple_idct_add_armv6, export=1
         pop    {r4-r11, pc}
 endfunc
 
-/* ff_simple_idct_put_armv6(uint8_t *dest, int line_size, DCTELEM *data); */
+/* ff_simple_idct_put_armv6(uint8_t *dest, int line_size, int16_t *data); */
 function ff_simple_idct_put_armv6, export=1
         push   {r0, r1, r4-r11, lr}
         sub    sp, sp, #128
diff --git a/libavcodec/arm/simple_idct_neon.S b/libavcodec/arm/simple_idct_neon.S
index a8fc13768e..c3e573c00a 100644
--- a/libavcodec/arm/simple_idct_neon.S
+++ b/libavcodec/arm/simple_idct_neon.S
@@ -261,7 +261,7 @@ endconst
         pop             {r4-r7, pc}
         .endm
 
-/* void ff_simple_idct_put_neon(uint8_t *dst, int line_size, DCTELEM *data); */
+/* void ff_simple_idct_put_neon(uint8_t *dst, int line_size, int16_t *data); */
 function ff_simple_idct_put_neon, export=1
         idct_start      r2
 
@@ -316,7 +316,7 @@ function idct_col4_add8_neon
         bx              lr
 endfunc
 
-/* void ff_simple_idct_add_neon(uint8_t *dst, int line_size, DCTELEM *data); */
+/* void ff_simple_idct_add_neon(uint8_t *dst, int line_size, int16_t *data); */
 function ff_simple_idct_add_neon, export=1
         idct_start      r2
 
@@ -355,7 +355,7 @@ function idct_col4_st16_neon
         bx              lr
 endfunc
 
-/* void ff_simple_idct_neon(DCTELEM *data); */
+/* void ff_simple_idct_neon(int16_t *data); */
 function ff_simple_idct_neon, export=1
         idct_start      r0
 
diff --git a/libavcodec/arm/vp3dsp_init_arm.c b/libavcodec/arm/vp3dsp_init_arm.c
index 4ce75fe651..5ffc8fd440 100644
--- a/libavcodec/arm/vp3dsp_init_arm.c
+++ b/libavcodec/arm/vp3dsp_init_arm.c
@@ -24,9 +24,9 @@
 #include "libavcodec/dsputil.h"
 #include "libavcodec/vp3dsp.h"
 
-void ff_vp3_idct_put_neon(uint8_t *dest, int line_size, DCTELEM *data);
-void ff_vp3_idct_add_neon(uint8_t *dest, int line_size, DCTELEM *data);
-void ff_vp3_idct_dc_add_neon(uint8_t *dest, int line_size, const DCTELEM *data);
+void ff_vp3_idct_put_neon(uint8_t *dest, int line_size, int16_t *data);
+void ff_vp3_idct_add_neon(uint8_t *dest, int line_size, int16_t *data);
+void ff_vp3_idct_dc_add_neon(uint8_t *dest, int line_size, const int16_t *data);
 
 void ff_vp3_v_loop_filter_neon(uint8_t *, int, int *);
 void ff_vp3_h_loop_filter_neon(uint8_t *, int, int *);
diff --git a/libavcodec/arm/vp8.h b/libavcodec/arm/vp8.h
index 7574a51f50..709f83d9e5 100644
--- a/libavcodec/arm/vp8.h
+++ b/libavcodec/arm/vp8.h
@@ -23,7 +23,7 @@
 
 #if HAVE_ARMV6_EXTERNAL
 #define decode_block_coeffs_internal ff_decode_block_coeffs_armv6
-int ff_decode_block_coeffs_armv6(VP56RangeCoder *rc, DCTELEM block[16],
+int ff_decode_block_coeffs_armv6(VP56RangeCoder *rc, int16_t block[16],
                                  uint8_t probs[8][3][NUM_DCT_TOKENS-1],
                                  int i, uint8_t *token_prob, int16_t qmul[2]);
 #endif
diff --git a/libavcodec/arm/vp8dsp_armv6.S b/libavcodec/arm/vp8dsp_armv6.S
index fd254ccc30..dfaba315b1 100644
--- a/libavcodec/arm/vp8dsp_armv6.S
+++ b/libavcodec/arm/vp8dsp_armv6.S
@@ -56,7 +56,7 @@
 
 @ idct
 
-@ void vp8_luma_dc_wht(DCTELEM block[4][4][16], DCTELEM dc[16])
+@ void vp8_luma_dc_wht(int16_t block[4][4][16], int16_t dc[16])
 function ff_vp8_luma_dc_wht_armv6, export=1
         push            {r4-r10, lr}
 
@@ -179,7 +179,7 @@ function ff_vp8_luma_dc_wht_armv6, export=1
         pop             {r4-r10, pc}
 endfunc
 
-@ void vp8_luma_dc_wht_dc(DCTELEM block[4][4][16], DCTELEM dc[16])
+@ void vp8_luma_dc_wht_dc(int16_t block[4][4][16], int16_t dc[16])
 function ff_vp8_luma_dc_wht_dc_armv6, export=1
         ldrsh           r2,  [r1]
         mov             r3,  #0
@@ -192,7 +192,7 @@ function ff_vp8_luma_dc_wht_dc_armv6, export=1
         bx              lr
 endfunc
 
-@ void vp8_idct_add(uint8_t *dst, DCTELEM block[16], int stride)
+@ void vp8_idct_add(uint8_t *dst, int16_t block[16], int stride)
 function ff_vp8_idct_add_armv6, export=1
         push            {r4-r12, lr}
         sub             sp,  sp,  #32
@@ -314,7 +314,7 @@ function ff_vp8_idct_add_armv6, export=1
         pop             {r4-r12, pc}
 endfunc
 
-@ void vp8_idct_dc_add(uint8_t *dst, DCTELEM block[16], int stride)
+@ void vp8_idct_dc_add(uint8_t *dst, int16_t block[16], int stride)
 function ff_vp8_idct_dc_add_armv6, export=1
         push            {r4-r6, lr}
         add             r6,  r0,  r2,  lsl #1
@@ -355,7 +355,7 @@ function ff_vp8_idct_dc_add_armv6, export=1
         pop             {r4-r6, pc}
 endfunc
 
-@ void vp8_idct_dc_add4uv(uint8_t *dst, DCTELEM block[4][16], int stride)
+@ void vp8_idct_dc_add4uv(uint8_t *dst, int16_t block[4][16], int stride)
 function ff_vp8_idct_dc_add4uv_armv6, export=1
         push            {r4, lr}
 
@@ -371,7 +371,7 @@ function ff_vp8_idct_dc_add4uv_armv6, export=1
         pop             {r4, pc}
 endfunc
 
-@ void vp8_idct_dc_add4y(uint8_t *dst, DCTELEM block[4][16], int stride)
+@ void vp8_idct_dc_add4y(uint8_t *dst, int16_t block[4][16], int stride)
 function ff_vp8_idct_dc_add4y_armv6, export=1
         push            {r4, lr}
 
diff --git a/libavcodec/arm/vp8dsp_init_armv6.c b/libavcodec/arm/vp8dsp_init_armv6.c
index c3d024f5cb..e15e191e30 100644
--- a/libavcodec/arm/vp8dsp_init_armv6.c
+++ b/libavcodec/arm/vp8dsp_init_armv6.c
@@ -22,13 +22,13 @@
 #include "libavcodec/vp8dsp.h"
 #include "vp8dsp.h"
 
-void ff_vp8_luma_dc_wht_armv6(DCTELEM block[4][4][16], DCTELEM dc[16]);
-void ff_vp8_luma_dc_wht_dc_armv6(DCTELEM block[4][4][16], DCTELEM dc[16]);
+void ff_vp8_luma_dc_wht_armv6(int16_t block[4][4][16], int16_t dc[16]);
+void ff_vp8_luma_dc_wht_dc_armv6(int16_t block[4][4][16], int16_t dc[16]);
 
-void ff_vp8_idct_add_armv6(uint8_t *dst, DCTELEM block[16], ptrdiff_t stride);
-void ff_vp8_idct_dc_add_armv6(uint8_t *dst, DCTELEM block[16], ptrdiff_t stride);
-void ff_vp8_idct_dc_add4y_armv6(uint8_t *dst, DCTELEM block[4][16], ptrdiff_t stride);
-void ff_vp8_idct_dc_add4uv_armv6(uint8_t *dst, DCTELEM block[4][16], ptrdiff_t stride);
+void ff_vp8_idct_add_armv6(uint8_t *dst, int16_t block[16], ptrdiff_t stride);
+void ff_vp8_idct_dc_add_armv6(uint8_t *dst, int16_t block[16], ptrdiff_t stride);
+void ff_vp8_idct_dc_add4y_armv6(uint8_t *dst, int16_t block[4][16], ptrdiff_t stride);
+void ff_vp8_idct_dc_add4uv_armv6(uint8_t *dst, int16_t block[4][16], ptrdiff_t stride);
 
 VP8_LF(armv6);
 
diff --git a/libavcodec/arm/vp8dsp_init_neon.c b/libavcodec/arm/vp8dsp_init_neon.c
index 965243c3e4..0468181f70 100644
--- a/libavcodec/arm/vp8dsp_init_neon.c
+++ b/libavcodec/arm/vp8dsp_init_neon.c
@@ -22,12 +22,12 @@
 #include "libavcodec/vp8dsp.h"
 #include "vp8dsp.h"
 
-void ff_vp8_luma_dc_wht_neon(DCTELEM block[4][4][16], DCTELEM dc[16]);
+void ff_vp8_luma_dc_wht_neon(int16_t block[4][4][16], int16_t dc[16]);
 
-void ff_vp8_idct_add_neon(uint8_t *dst, DCTELEM block[16], ptrdiff_t stride);
-void ff_vp8_idct_dc_add_neon(uint8_t *dst, DCTELEM block[16], ptrdiff_t stride);
-void ff_vp8_idct_dc_add4y_neon(uint8_t *dst, DCTELEM block[4][16], ptrdiff_t stride);
-void ff_vp8_idct_dc_add4uv_neon(uint8_t *dst, DCTELEM block[4][16], ptrdiff_t stride);
+void ff_vp8_idct_add_neon(uint8_t *dst, int16_t block[16], ptrdiff_t stride);
+void ff_vp8_idct_dc_add_neon(uint8_t *dst, int16_t block[16], ptrdiff_t stride);
+void ff_vp8_idct_dc_add4y_neon(uint8_t *dst, int16_t block[4][16], ptrdiff_t stride);
+void ff_vp8_idct_dc_add4uv_neon(uint8_t *dst, int16_t block[4][16], ptrdiff_t stride);
 
 VP8_LF(neon);
 
diff --git a/libavcodec/asv.h b/libavcodec/asv.h
index 4a6c799738..72d98c0b33 100644
--- a/libavcodec/asv.h
+++ b/libavcodec/asv.h
@@ -48,7 +48,7 @@ typedef struct ASV1Context{
     int mb_height;
     int mb_width2;
     int mb_height2;
-    DECLARE_ALIGNED(16, DCTELEM, block)[6][64];
+    DECLARE_ALIGNED(16, int16_t, block)[6][64];
     uint16_t intra_matrix[64];
     int q_intra_matrix[64];
     uint8_t *bitstream_buffer;
diff --git a/libavcodec/asvdec.c b/libavcodec/asvdec.c
index e4d950400f..b8a1f21026 100644
--- a/libavcodec/asvdec.c
+++ b/libavcodec/asvdec.c
@@ -94,7 +94,7 @@ static inline int asv2_get_level(GetBitContext *gb)
         return code - 31;
 }
 
-static inline int asv1_decode_block(ASV1Context *a, DCTELEM block[64])
+static inline int asv1_decode_block(ASV1Context *a, int16_t block[64])
 {
     int i;
 
@@ -125,7 +125,7 @@ static inline int asv1_decode_block(ASV1Context *a, DCTELEM block[64])
     return 0;
 }
 
-static inline int asv2_decode_block(ASV1Context *a, DCTELEM block[64])
+static inline int asv2_decode_block(ASV1Context *a, int16_t block[64])
 {
     int i, count, ccp;
 
@@ -161,7 +161,7 @@ static inline int asv2_decode_block(ASV1Context *a, DCTELEM block[64])
     return 0;
 }
 
-static inline int decode_mb(ASV1Context *a, DCTELEM block[6][64])
+static inline int decode_mb(ASV1Context *a, int16_t block[6][64])
 {
     int i;
 
@@ -183,7 +183,7 @@ static inline int decode_mb(ASV1Context *a, DCTELEM block[6][64])
 
 static inline void idct_put(ASV1Context *a, int mb_x, int mb_y)
 {
-    DCTELEM (*block)[64] = a->block;
+    int16_t (*block)[64] = a->block;
     int linesize         = a->picture.linesize[0];
 
     uint8_t *dest_y  = a->picture.data[0] + (mb_y * 16* linesize              ) + mb_x * 16;
diff --git a/libavcodec/asvenc.c b/libavcodec/asvenc.c
index e5f3b48ea4..a09aa73ce0 100644
--- a/libavcodec/asvenc.c
+++ b/libavcodec/asvenc.c
@@ -56,7 +56,7 @@ static inline void asv2_put_level(PutBitContext *pb, int level){
     }
 }
 
-static inline void asv1_encode_block(ASV1Context *a, DCTELEM block[64]){
+static inline void asv1_encode_block(ASV1Context *a, int16_t block[64]){
     int i;
     int nc_count=0;
 
@@ -89,7 +89,7 @@ static inline void asv1_encode_block(ASV1Context *a, DCTELEM block[64]){
     put_bits(&a->pb, ff_asv_ccp_tab[16][1], ff_asv_ccp_tab[16][0]);
 }
 
-static inline void asv2_encode_block(ASV1Context *a, DCTELEM block[64]){
+static inline void asv2_encode_block(ASV1Context *a, int16_t block[64]){
     int i;
     int count=0;
 
@@ -130,7 +130,7 @@ static inline void asv2_encode_block(ASV1Context *a, DCTELEM block[64]){
 
 #define MAX_MB_SIZE (30*16*16*3/2/8)
 
-static inline int encode_mb(ASV1Context *a, DCTELEM block[6][64]){
+static inline int encode_mb(ASV1Context *a, int16_t block[6][64]){
     int i;
 
     if (a->pb.buf_end - a->pb.buf - (put_bits_count(&a->pb)>>3) < MAX_MB_SIZE) {
@@ -149,7 +149,7 @@ static inline int encode_mb(ASV1Context *a, DCTELEM block[6][64]){
 }
 
 static inline void dct_get(ASV1Context *a, int mb_x, int mb_y){
-    DCTELEM (*block)[64]= a->block;
+    int16_t (*block)[64]= a->block;
     int linesize= a->picture.linesize[0];
     int i;
 
diff --git a/libavcodec/bfin/config_bfin.h b/libavcodec/bfin/config_bfin.h
index f3a2c6ebc6..36439530e9 100644
--- a/libavcodec/bfin/config_bfin.h
+++ b/libavcodec/bfin/config_bfin.h
@@ -21,7 +21,7 @@
    low level assembler interface wrapper
 
 DEFUN(put_pixels_clamped,mL1,
-        (DCTELEM *block, uint8_t *dest, int line_size)):
+        (int16_t *block, uint8_t *dest, int line_size)):
 
       body
 
diff --git a/libavcodec/bfin/dsputil_bfin.c b/libavcodec/bfin/dsputil_bfin.c
index a1dcb14224..955aea01f6 100644
--- a/libavcodec/bfin/dsputil_bfin.c
+++ b/libavcodec/bfin/dsputil_bfin.c
@@ -27,20 +27,20 @@
 
 int off;
 
-static void bfin_idct_add (uint8_t *dest, int line_size, DCTELEM *block)
+static void bfin_idct_add (uint8_t *dest, int line_size, int16_t *block)
 {
     ff_bfin_idct (block);
     ff_bfin_add_pixels_clamped (block, dest, line_size);
 }
 
-static void bfin_idct_put (uint8_t *dest, int line_size, DCTELEM *block)
+static void bfin_idct_put (uint8_t *dest, int line_size, int16_t *block)
 {
     ff_bfin_idct (block);
     ff_bfin_put_pixels_clamped (block, dest, line_size);
 }
 
 
-static void bfin_clear_blocks (DCTELEM *blocks)
+static void bfin_clear_blocks (int16_t *blocks)
 {
     // This is just a simple memset.
     //
diff --git a/libavcodec/bfin/dsputil_bfin.h b/libavcodec/bfin/dsputil_bfin.h
index 43ac39c9e0..04f87cc2b8 100644
--- a/libavcodec/bfin/dsputil_bfin.h
+++ b/libavcodec/bfin/dsputil_bfin.h
@@ -24,8 +24,9 @@
 #ifndef AVCODEC_BFIN_DSPUTIL_BFIN_H
 #define AVCODEC_BFIN_DSPUTIL_BFIN_H
 
+#include <stdint.h>
+
 #include "config.h"
-#include "libavcodec/dsputil.h"
 
 #if defined(__FDPIC__) && CONFIG_SRAM
 #define attribute_l1_text  __attribute__ ((l1_text))
@@ -35,15 +36,15 @@
 #define attribute_l1_data_b
 #endif
 
-void ff_bfin_idct (DCTELEM *block) attribute_l1_text;
-void ff_bfin_fdct (DCTELEM *block) attribute_l1_text;
-void ff_bfin_vp3_idct (DCTELEM *block);
-void ff_bfin_vp3_idct_put (uint8_t *dest, int line_size, DCTELEM *block);
-void ff_bfin_vp3_idct_add (uint8_t *dest, int line_size, DCTELEM *block);
-void ff_bfin_add_pixels_clamped (const DCTELEM *block, uint8_t *dest, int line_size) attribute_l1_text;
-void ff_bfin_put_pixels_clamped (const DCTELEM *block, uint8_t *dest, int line_size) attribute_l1_text;
-void ff_bfin_diff_pixels (DCTELEM *block, const uint8_t *s1, const uint8_t *s2, int stride)  attribute_l1_text;
-void ff_bfin_get_pixels  (DCTELEM *av_restrict block, const uint8_t *pixels, int line_size) attribute_l1_text;
+void ff_bfin_idct (int16_t *block) attribute_l1_text;
+void ff_bfin_fdct (int16_t *block) attribute_l1_text;
+void ff_bfin_vp3_idct (int16_t *block);
+void ff_bfin_vp3_idct_put (uint8_t *dest, int line_size, int16_t *block);
+void ff_bfin_vp3_idct_add (uint8_t *dest, int line_size, int16_t *block);
+void ff_bfin_add_pixels_clamped (const int16_t *block, uint8_t *dest, int line_size) attribute_l1_text;
+void ff_bfin_put_pixels_clamped (const int16_t *block, uint8_t *dest, int line_size) attribute_l1_text;
+void ff_bfin_diff_pixels (int16_t *block, const uint8_t *s1, const uint8_t *s2, int stride)  attribute_l1_text;
+void ff_bfin_get_pixels  (int16_t *restrict block, const uint8_t *pixels, int line_size) attribute_l1_text;
 int  ff_bfin_pix_norm1  (uint8_t * pix, int line_size) attribute_l1_text;
 int  ff_bfin_z_sad8x8   (uint8_t *blk1, uint8_t *blk2, int dsz, int line_size, int h) attribute_l1_text;
 int  ff_bfin_z_sad16x16 (uint8_t *blk1, uint8_t *blk2, int dsz, int line_size, int h) attribute_l1_text;
diff --git a/libavcodec/bfin/fdct_bfin.S b/libavcodec/bfin/fdct_bfin.S
index e15acb654f..916957467c 100644
--- a/libavcodec/bfin/fdct_bfin.S
+++ b/libavcodec/bfin/fdct_bfin.S
@@ -20,7 +20,7 @@
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  */
 /*
-  void ff_bfin_fdct (DCTELEM *buf);
+  void ff_bfin_fdct (int16_t *buf);
 
   This implementation works only for 8x8 input. The range of input
   must be -256 to 255 i.e. 8bit input represented in a 16bit data
@@ -61,9 +61,9 @@ Notation
   Other registers used:
         I0, I1, I2, I3, B0, B2, B3, M0, M1, L3 registers and LC0.
 
-  Input - r0 - pointer to start of DCTELEM *block
+  Input - r0 - pointer to start of int16_t *block
 
-  Output - The DCT output coefficients in the DCTELEM *block
+  Output - The DCT output coefficients in the int16_t *block
 
   Register constraint:
                This code is called from jpeg_encode.
@@ -147,7 +147,7 @@ vtmp:   .space 128
 
 .text
 DEFUN(fdct,mL1,
-        (DCTELEM *block)):
+        (int16_t *block)):
     [--SP] = (R7:4, P5:3);          // Push the registers onto the stack.
 
     b0 = r0;
diff --git a/libavcodec/bfin/idct_bfin.S b/libavcodec/bfin/idct_bfin.S
index 66430096f0..bd80447bde 100644
--- a/libavcodec/bfin/idct_bfin.S
+++ b/libavcodec/bfin/idct_bfin.S
@@ -22,7 +22,7 @@
 /*
    This blackfin DSP code implements an 8x8 inverse type II DCT.
 
-Prototype       : void ff_bfin_idct(DCTELEM *in)
+Prototype       : void ff_bfin_idct(int16_t *in)
 
 Registers Used  : A0, A1, R0-R7, I0-I3, B0, B2, B3, M0-M2, L0-L3, P0-P5, LC0.
 
@@ -90,7 +90,7 @@ vtmp: .space 256
 
 .text
 DEFUN(idct,mL1,
-        (DCTELEM *block)):
+        (int16_t *block)):
 
 /********************** Function Prologue *********************************/
     link 16;
diff --git a/libavcodec/bfin/mpegvideo_bfin.c b/libavcodec/bfin/mpegvideo_bfin.c
index f7f3d729de..41c47f4d27 100644
--- a/libavcodec/bfin/mpegvideo_bfin.c
+++ b/libavcodec/bfin/mpegvideo_bfin.c
@@ -26,7 +26,7 @@
 #include "dsputil_bfin.h"
 
 static int dct_quantize_bfin (MpegEncContext *s,
-                              DCTELEM *block, int n,
+                              int16_t *block, int n,
                               int qscale, int *overflow)
 {
     int last_non_zero, q, start_i;
diff --git a/libavcodec/bfin/pixels_bfin.S b/libavcodec/bfin/pixels_bfin.S
index 995d7b8cc8..2e3cd0f9bc 100644
--- a/libavcodec/bfin/pixels_bfin.S
+++ b/libavcodec/bfin/pixels_bfin.S
@@ -21,7 +21,7 @@
 #include "config_bfin.h"
 
 DEFUN(put_pixels_clamped,mL1,
-        (DCTELEM *block, uint8_t *dest, int line_size)):
+        (int16_t *block, uint8_t *dest, int line_size)):
     [--SP] = (R7:4);
     R4 = 0;
     R5.l = 0x00ff;
@@ -51,7 +51,7 @@ ppc$1: R2 = Max(R0, R4) (V)      || [I1++M1] = R6;
 DEFUN_END(put_pixels_clamped)
 
 DEFUN(add_pixels_clamped,mL1,
-        (DCTELEM *block, uint8_t *dest, int line_size)):
+        (int16_t *block, uint8_t *dest, int line_size)):
     [-- SP] = (R7:4);
     R4 = 0;
     I0 = 0;
@@ -442,7 +442,7 @@ LE$8OT: DISALGNEXCPT                       || R2  =[I1++]   || [I3++M2] = R5;
         rts;
 
 DEFUN(diff_pixels,mL1,
-       (DCTELEM *block, uint8_t *s1, uint8_t *s2, int stride)):
+       (int16_t *block, uint8_t *s1, uint8_t *s2, int stride)):
         link 0;
         [--sp] = (r7:4);
         p0=8;
@@ -518,7 +518,7 @@ DEFUN_END(pix_sum)
 
 
 DEFUN(get_pixels,mL1,
-        (DCTELEM *av_restrict block, const uint8_t *pixels, int line_size)):
+        (int16_t *av_restrict block, const uint8_t *pixels, int line_size)):
         [--sp] = (r7:4);
         i3=r0;        // dest
         i0=r1;        // src0
diff --git a/libavcodec/bfin/vp3_bfin.c b/libavcodec/bfin/vp3_bfin.c
index 6d6ac82636..06d3443a2f 100644
--- a/libavcodec/bfin/vp3_bfin.c
+++ b/libavcodec/bfin/vp3_bfin.c
@@ -23,7 +23,7 @@
 #include "dsputil_bfin.h"
 
 /* Intra iDCT offset 128 */
-void ff_bfin_vp3_idct_put (uint8_t *dest, int line_size, DCTELEM *block)
+void ff_bfin_vp3_idct_put (uint8_t *dest, int line_size, int16_t *block)
 {
     uint8_t *cm = ff_cropTbl + MAX_NEG_CROP + 128;
     int i,j;
@@ -36,7 +36,7 @@ void ff_bfin_vp3_idct_put (uint8_t *dest, int line_size, DCTELEM *block)
 }
 
 /* Inter iDCT */
-void ff_bfin_vp3_idct_add (uint8_t *dest, int line_size, DCTELEM *block)
+void ff_bfin_vp3_idct_add (uint8_t *dest, int line_size, int16_t *block)
 {
     ff_bfin_vp3_idct (block);
     ff_bfin_add_pixels_clamped (block, dest, line_size);
diff --git a/libavcodec/bfin/vp3_idct_bfin.S b/libavcodec/bfin/vp3_idct_bfin.S
index 83747f9fcd..4e200459f4 100644
--- a/libavcodec/bfin/vp3_idct_bfin.S
+++ b/libavcodec/bfin/vp3_idct_bfin.S
@@ -22,7 +22,7 @@
 /*
    This blackfin DSP code implements an 8x8 inverse type II DCT.
 
-Prototype       : void ff_bfin_vp3_idct(DCTELEM *in)
+Prototype       : void ff_bfin_vp3_idct(int16_t *in)
 
 Registers Used  : A0, A1, R0-R7, I0-I3, B0, B2, B3, M0-M2, L0-L3, P0-P5, LC0.
 
@@ -63,7 +63,7 @@ vtmp: .space 256
 
 .text
 DEFUN(vp3_idct,mL1,
-        (DCTELEM *block)):
+        (int16_t *block)):
 
 /********************** Function Prologue *********************************/
     link 16;
diff --git a/libavcodec/bink.c b/libavcodec/bink.c
index 892e92184b..35fe962e8d 100644
--- a/libavcodec/bink.c
+++ b/libavcodec/bink.c
@@ -700,7 +700,7 @@ static int read_dct_coeffs(GetBitContext *gb, int32_t block[64], const uint8_t *
  * @param masks_count number of masks to decode
  * @return 0 on success, negative value in other cases
  */
-static int read_residue(GetBitContext *gb, DCTELEM block[64], int masks_count)
+static int read_residue(GetBitContext *gb, int16_t block[64], int masks_count)
 {
     int coef_list[128];
     int mode_list[128];
@@ -804,7 +804,7 @@ static int binkb_decode_plane(BinkContext *c, GetBitContext *gb, int plane_idx,
     int v, col[2];
     const uint8_t *scan;
     int xoff, yoff;
-    LOCAL_ALIGNED_16(DCTELEM, block, [64]);
+    LOCAL_ALIGNED_16(int16_t, block, [64]);
     LOCAL_ALIGNED_16(int32_t, dctblock, [64]);
     int coordmap[64];
     int ybias = is_key ? -15 : 0;
@@ -950,7 +950,7 @@ static int bink_decode_plane(BinkContext *c, GetBitContext *gb, int plane_idx,
     int v, col[2];
     const uint8_t *scan;
     int xoff, yoff;
-    LOCAL_ALIGNED_16(DCTELEM, block, [64]);
+    LOCAL_ALIGNED_16(int16_t, block, [64]);
     LOCAL_ALIGNED_16(uint8_t, ublock, [64]);
     LOCAL_ALIGNED_16(int32_t, dctblock, [64]);
     int coordmap[64];
diff --git a/libavcodec/cavs.c b/libavcodec/cavs.c
index e1ac0befa3..bcdeb7c228 100644
--- a/libavcodec/cavs.c
+++ b/libavcodec/cavs.c
@@ -715,7 +715,7 @@ void ff_cavs_init_top_lines(AVSContext *h) {
     /* alloc space for co-located MVs and types */
     h->col_mv       = av_mallocz( h->mb_width*h->mb_height*4*sizeof(cavs_vector));
     h->col_type_base = av_mallocz(h->mb_width*h->mb_height);
-    h->block        = av_mallocz(64*sizeof(DCTELEM));
+    h->block        = av_mallocz(64*sizeof(int16_t));
 }
 
 av_cold int ff_cavs_init(AVCodecContext *avctx) {
diff --git a/libavcodec/cavs.h b/libavcodec/cavs.h
index 649a567b41..73e6bfe8c6 100644
--- a/libavcodec/cavs.h
+++ b/libavcodec/cavs.h
@@ -234,7 +234,7 @@ typedef struct AVSContext {
     uint8_t *edge_emu_buffer;
 
     int got_keyframe;
-    DCTELEM *block;
+    int16_t *block;
 } AVSContext;
 
 extern const uint8_t     ff_cavs_partition_flags[30];
diff --git a/libavcodec/cavsdec.c b/libavcodec/cavsdec.c
index 0074c4cf6d..9450ed137f 100644
--- a/libavcodec/cavsdec.c
+++ b/libavcodec/cavsdec.c
@@ -517,8 +517,8 @@ static inline int get_ue_code(GetBitContext *gb, int order)
     return get_ue_golomb(gb);
 }
 
-static inline int dequant(AVSContext *h, DCTELEM *level_buf, uint8_t *run_buf,
-                          DCTELEM *dst, int mul, int shift, int coeff_num)
+static inline int dequant(AVSContext *h, int16_t *level_buf, uint8_t *run_buf,
+                          int16_t *dst, int mul, int shift, int coeff_num)
 {
     int round = 1 << (shift - 1);
     int pos = -1;
@@ -553,9 +553,9 @@ static int decode_residual_block(AVSContext *h, GetBitContext *gb,
 {
     int i, esc_code, level, mask;
     unsigned int level_code, run;
-    DCTELEM level_buf[65];
+    int16_t level_buf[65];
     uint8_t run_buf[65];
-    DCTELEM *block = h->block;
+    int16_t *block = h->block;
 
     for (i = 0; i < 65; i++) {
         level_code = get_ue_code(gb, r->golomb_order);
diff --git a/libavcodec/cavsdsp.c b/libavcodec/cavsdsp.c
index a9b136e2a3..f66cabca28 100644
--- a/libavcodec/cavsdsp.c
+++ b/libavcodec/cavsdsp.c
@@ -183,9 +183,9 @@ static void cavs_filter_ch_c(uint8_t *d, int stride, int alpha, int beta, int tc
  *
  ****************************************************************************/
 
-static void cavs_idct8_add_c(uint8_t *dst, DCTELEM *block, int stride) {
+static void cavs_idct8_add_c(uint8_t *dst, int16_t *block, int stride) {
     int i;
-    DCTELEM (*src)[8] = (DCTELEM(*)[8])block;
+    int16_t (*src)[8] = (int16_t(*)[8])block;
     uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;
 
     src[0][0] += 8;
diff --git a/libavcodec/cavsdsp.h b/libavcodec/cavsdsp.h
index f6e3e18e83..4598c7d745 100644
--- a/libavcodec/cavsdsp.h
+++ b/libavcodec/cavsdsp.h
@@ -32,7 +32,7 @@ typedef struct CAVSDSPContext {
     void (*cavs_filter_lh)(uint8_t *pix, int stride, int alpha, int beta, int tc, int bs1, int bs2);
     void (*cavs_filter_cv)(uint8_t *pix, int stride, int alpha, int beta, int tc, int bs1, int bs2);
     void (*cavs_filter_ch)(uint8_t *pix, int stride, int alpha, int beta, int tc, int bs1, int bs2);
-    void (*cavs_idct8_add)(uint8_t *dst, DCTELEM *block, int stride);
+    void (*cavs_idct8_add)(uint8_t *dst, int16_t *block, int stride);
     int idct_perm;
 } CAVSDSPContext;
 
diff --git a/libavcodec/dct-test.c b/libavcodec/dct-test.c
index 2c43b36aa4..e44ab125c2 100644
--- a/libavcodec/dct-test.c
+++ b/libavcodec/dct-test.c
@@ -39,6 +39,7 @@
 #include "libavutil/lfg.h"
 #include "libavutil/time.h"
 
+#include "dsputil.h"
 #include "simple_idct.h"
 #include "aandcttab.h"
 #include "faandct.h"
@@ -48,28 +49,28 @@
 
 #undef printf
 
-void ff_mmx_idct(DCTELEM *data);
-void ff_mmxext_idct(DCTELEM *data);
+void ff_mmx_idct(int16_t *data);
+void ff_mmxext_idct(int16_t *data);
 
 // BFIN
-void ff_bfin_idct(DCTELEM *block);
-void ff_bfin_fdct(DCTELEM *block);
+void ff_bfin_idct(int16_t *block);
+void ff_bfin_fdct(int16_t *block);
 
 // ALTIVEC
-void ff_fdct_altivec(DCTELEM *block);
+void ff_fdct_altivec(int16_t *block);
 
 // ARM
-void ff_j_rev_dct_arm(DCTELEM *data);
-void ff_simple_idct_arm(DCTELEM *data);
-void ff_simple_idct_armv5te(DCTELEM *data);
-void ff_simple_idct_armv6(DCTELEM *data);
-void ff_simple_idct_neon(DCTELEM *data);
+void ff_j_rev_dct_arm(int16_t *data);
+void ff_simple_idct_arm(int16_t *data);
+void ff_simple_idct_armv5te(int16_t *data);
+void ff_simple_idct_armv6(int16_t *data);
+void ff_simple_idct_neon(int16_t *data);
 
-void ff_simple_idct_axp(DCTELEM *data);
+void ff_simple_idct_axp(int16_t *data);
 
 struct algo {
     const char *name;
-    void (*func)(DCTELEM *block);
+    void (*func)(int16_t *block);
     enum formattag { NO_PERM, MMX_PERM, MMX_SIMPLE_PERM, SCALE_PERM,
                      SSE2_PERM, PARTTRANS_PERM, TRANSPOSE_PERM } format;
     int mm_support;
@@ -103,9 +104,9 @@ static const struct algo fdct_tab[] = {
 
 #if ARCH_X86_64 && HAVE_MMX && HAVE_YASM
 void ff_prores_idct_put_10_sse2(uint16_t *dst, int linesize,
-                                DCTELEM *block, int16_t *qmat);
+                                int16_t *block, int16_t *qmat);
 
-static void ff_prores_idct_put_10_sse2_wrap(DCTELEM *dst){
+static void ff_prores_idct_put_10_sse2_wrap(int16_t *dst){
     DECLARE_ALIGNED(16, static int16_t, qmat)[64];
     DECLARE_ALIGNED(16, static int16_t, tmp)[64];
     int i;
@@ -193,10 +194,10 @@ static void idct_mmx_init(void)
     }
 }
 
-DECLARE_ALIGNED(16, static DCTELEM, block)[64];
-DECLARE_ALIGNED(8,  static DCTELEM, block1)[64];
+DECLARE_ALIGNED(16, static int16_t, block)[64];
+DECLARE_ALIGNED(8,  static int16_t, block1)[64];
 
-static void init_block(DCTELEM block[64], int test, int is_idct, AVLFG *prng, int vals)
+static void init_block(int16_t block[64], int test, int is_idct, AVLFG *prng, int vals)
 {
     int i, j;
 
@@ -226,7 +227,7 @@ static void init_block(DCTELEM block[64], int test, int is_idct, AVLFG *prng, in
     }
 }
 
-static void permute(DCTELEM dst[64], const DCTELEM src[64], int perm)
+static void permute(int16_t dst[64], const int16_t src[64], int perm)
 {
     int i;
 
@@ -253,7 +254,7 @@ static void permute(DCTELEM dst[64], const DCTELEM src[64], int perm)
 
 static int dct_error(const struct algo *dct, int test, int is_idct, int speed, const int bits)
 {
-    void (*ref)(DCTELEM *block) = is_idct ? ff_ref_idct : ff_ref_fdct;
+    void (*ref)(int16_t *block) = is_idct ? ff_ref_idct : ff_ref_fdct;
     int it, i, scale;
     int err_inf, v;
     int64_t err2, ti, ti1, it1, err_sum = 0;
diff --git a/libavcodec/dnxhddec.c b/libavcodec/dnxhddec.c
index 0fe03f9978..6e94d32ecc 100644
--- a/libavcodec/dnxhddec.c
+++ b/libavcodec/dnxhddec.c
@@ -45,11 +45,11 @@ typedef struct DNXHDContext {
     VLC ac_vlc, dc_vlc, run_vlc;
     int last_dc[3];
     DSPContext dsp;
-    DECLARE_ALIGNED(16, DCTELEM, blocks)[8][64];
+    DECLARE_ALIGNED(16, int16_t, blocks)[8][64];
     ScanTable scantable;
     const CIDEntry *cid_table;
     int bit_depth; // 8, 10 or 0 if not initialized at all.
-    void (*decode_dct_block)(struct DNXHDContext *ctx, DCTELEM *block,
+    void (*decode_dct_block)(struct DNXHDContext *ctx, int16_t *block,
                              int n, int qscale);
     int last_qscale;
     int luma_scale[64];
@@ -59,8 +59,8 @@ typedef struct DNXHDContext {
 #define DNXHD_VLC_BITS 9
 #define DNXHD_DC_VLC_BITS 7
 
-static void dnxhd_decode_dct_block_8(DNXHDContext *ctx, DCTELEM *block, int n, int qscale);
-static void dnxhd_decode_dct_block_10(DNXHDContext *ctx, DCTELEM *block, int n, int qscale);
+static void dnxhd_decode_dct_block_8(DNXHDContext *ctx, int16_t *block, int n, int qscale);
+static void dnxhd_decode_dct_block_10(DNXHDContext *ctx, int16_t *block, int n, int qscale);
 
 static av_cold int dnxhd_decode_init(AVCodecContext *avctx)
 {
@@ -190,7 +190,7 @@ static int dnxhd_decode_header(DNXHDContext *ctx, const uint8_t *buf, int buf_si
 }
 
 static av_always_inline void dnxhd_decode_dct_block(DNXHDContext *ctx,
-                                                    DCTELEM *block, int n,
+                                                    int16_t *block, int n,
                                                     int qscale,
                                                     int index_bits,
                                                     int level_bias,
@@ -272,13 +272,13 @@ static av_always_inline void dnxhd_decode_dct_block(DNXHDContext *ctx,
     CLOSE_READER(bs, &ctx->gb);
 }
 
-static void dnxhd_decode_dct_block_8(DNXHDContext *ctx, DCTELEM *block,
+static void dnxhd_decode_dct_block_8(DNXHDContext *ctx, int16_t *block,
                                      int n, int qscale)
 {
     dnxhd_decode_dct_block(ctx, block, n, qscale, 4, 32, 6);
 }
 
-static void dnxhd_decode_dct_block_10(DNXHDContext *ctx, DCTELEM *block,
+static void dnxhd_decode_dct_block_10(DNXHDContext *ctx, int16_t *block,
                                       int n, int qscale)
 {
     dnxhd_decode_dct_block(ctx, block, n, qscale, 6, 8, 4);
diff --git a/libavcodec/dnxhdenc.c b/libavcodec/dnxhdenc.c
index 690e64f869..5fbf46b281 100644
--- a/libavcodec/dnxhdenc.c
+++ b/libavcodec/dnxhdenc.c
@@ -51,7 +51,7 @@ static const AVClass class = {
 
 #define LAMBDA_FRAC_BITS 10
 
-static void dnxhd_8bit_get_pixels_8x4_sym(DCTELEM *av_restrict block, const uint8_t *pixels, int line_size)
+static void dnxhd_8bit_get_pixels_8x4_sym(int16_t *av_restrict block, const uint8_t *pixels, int line_size)
 {
     int i;
     for (i = 0; i < 4; i++) {
@@ -68,7 +68,7 @@ static void dnxhd_8bit_get_pixels_8x4_sym(DCTELEM *av_restrict block, const uint
     memcpy(block + 24, block - 32, sizeof(*block) * 8);
 }
 
-static av_always_inline void dnxhd_10bit_get_pixels_8x4_sym(DCTELEM *av_restrict block, const uint8_t *pixels, int line_size)
+static av_always_inline void dnxhd_10bit_get_pixels_8x4_sym(int16_t *av_restrict block, const uint8_t *pixels, int line_size)
 {
     int i;
     const uint16_t* pixels16 = (const uint16_t*)pixels;
@@ -88,7 +88,7 @@ static av_always_inline void dnxhd_10bit_get_pixels_8x4_sym(DCTELEM *av_restrict
     memcpy(block + 24, block - 32, sizeof(*block) * 8);
 }
 
-static int dnxhd_10bit_dct_quantize(MpegEncContext *ctx, DCTELEM *block,
+static int dnxhd_10bit_dct_quantize(MpegEncContext *ctx, int16_t *block,
                                     int n, int qscale, int *overflow)
 {
     const uint8_t *scantable= ctx->intra_scantable.scantable;
@@ -392,7 +392,7 @@ static av_always_inline void dnxhd_encode_dc(DNXHDEncContext *ctx, int diff)
              (ctx->cid_table->dc_codes[nbits]<<nbits) + (diff & ((1 << nbits) - 1)));
 }
 
-static av_always_inline void dnxhd_encode_block(DNXHDEncContext *ctx, DCTELEM *block, int last_index, int n)
+static av_always_inline void dnxhd_encode_block(DNXHDEncContext *ctx, int16_t *block, int last_index, int n)
 {
     int last_non_zero = 0;
     int slevel, i, j;
@@ -415,7 +415,7 @@ static av_always_inline void dnxhd_encode_block(DNXHDEncContext *ctx, DCTELEM *b
     put_bits(&ctx->m.pb, ctx->vlc_bits[0], ctx->vlc_codes[0]); // EOB
 }
 
-static av_always_inline void dnxhd_unquantize_c(DNXHDEncContext *ctx, DCTELEM *block, int n, int qscale, int last_index)
+static av_always_inline void dnxhd_unquantize_c(DNXHDEncContext *ctx, int16_t *block, int n, int qscale, int last_index)
 {
     const uint8_t *weight_matrix;
     int level;
@@ -456,7 +456,7 @@ static av_always_inline void dnxhd_unquantize_c(DNXHDEncContext *ctx, DCTELEM *b
     }
 }
 
-static av_always_inline int dnxhd_ssd_block(DCTELEM *qblock, DCTELEM *block)
+static av_always_inline int dnxhd_ssd_block(int16_t *qblock, int16_t *block)
 {
     int score = 0;
     int i;
@@ -465,7 +465,7 @@ static av_always_inline int dnxhd_ssd_block(DCTELEM *qblock, DCTELEM *block)
     return score;
 }
 
-static av_always_inline int dnxhd_calc_ac_bits(DNXHDEncContext *ctx, DCTELEM *block, int last_index)
+static av_always_inline int dnxhd_calc_ac_bits(DNXHDEncContext *ctx, int16_t *block, int last_index)
 {
     int last_non_zero = 0;
     int bits = 0;
@@ -527,7 +527,7 @@ static int dnxhd_calc_bits_thread(AVCodecContext *avctx, void *arg, int jobnr, i
     DNXHDEncContext *ctx = avctx->priv_data;
     int mb_y = jobnr, mb_x;
     int qscale = ctx->qscale;
-    LOCAL_ALIGNED_16(DCTELEM, block, [64]);
+    LOCAL_ALIGNED_16(int16_t, block, [64]);
     ctx = ctx->thread[threadnr];
 
     ctx->m.last_dc[0] =
@@ -544,7 +544,7 @@ static int dnxhd_calc_bits_thread(AVCodecContext *avctx, void *arg, int jobnr, i
         dnxhd_get_blocks(ctx, mb_x, mb_y);
 
         for (i = 0; i < 8; i++) {
-            DCTELEM *src_block = ctx->blocks[i];
+            int16_t *src_block = ctx->blocks[i];
             int overflow, nbits, diff, last_index;
             int n = dnxhd_switch_matrix(ctx, i);
 
@@ -593,7 +593,7 @@ static int dnxhd_encode_thread(AVCodecContext *avctx, void *arg, int jobnr, int
         dnxhd_get_blocks(ctx, mb_x, mb_y);
 
         for (i = 0; i < 8; i++) {
-            DCTELEM *block = ctx->blocks[i];
+            int16_t *block = ctx->blocks[i];
             int overflow, n = dnxhd_switch_matrix(ctx, i);
             int last_index = ctx->m.dct_quantize(&ctx->m, block, 4&(2*i), qscale, &overflow);
             //START_TIMER;
diff --git a/libavcodec/dnxhdenc.h b/libavcodec/dnxhdenc.h
index 36a42fb100..9b59b96d3e 100644
--- a/libavcodec/dnxhdenc.h
+++ b/libavcodec/dnxhdenc.h
@@ -64,7 +64,7 @@ typedef struct DNXHDEncContext {
     int nitris_compat;
     unsigned min_padding;
 
-    DECLARE_ALIGNED(16, DCTELEM, blocks)[8][64];
+    DECLARE_ALIGNED(16, int16_t, blocks)[8][64];
 
     int      (*qmatrix_c)     [64];
     int      (*qmatrix_l)     [64];
@@ -90,7 +90,7 @@ typedef struct DNXHDEncContext {
     RCCMPEntry *mb_cmp;
     RCEntry   (*mb_rc)[8160];
 
-    void (*get_pixels_8x4_sym)(DCTELEM * /*align 16*/, const uint8_t *, int);
+    void (*get_pixels_8x4_sym)(int16_t * /*align 16*/, const uint8_t *, int);
 } DNXHDEncContext;
 
 void ff_dnxhdenc_init_x86(DNXHDEncContext *ctx);
diff --git a/libavcodec/dsputil.c b/libavcodec/dsputil.c
index 98d4596d82..80a85a951c 100644
--- a/libavcodec/dsputil.c
+++ b/libavcodec/dsputil.c
@@ -351,7 +351,7 @@ static int sse16_c(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h)
     return s;
 }
 
-static void diff_pixels_c(DCTELEM *av_restrict block, const uint8_t *s1,
+static void diff_pixels_c(int16_t *av_restrict block, const uint8_t *s1,
                           const uint8_t *s2, int stride){
     int i;
 
@@ -371,8 +371,7 @@ static void diff_pixels_c(DCTELEM *av_restrict block, const uint8_t *s1,
     }
 }
 
-
-static void put_pixels_clamped_c(const DCTELEM *block, uint8_t *av_restrict pixels,
+static void put_pixels_clamped_c(const int16_t *block, uint8_t *av_restrict pixels,
                                  int line_size)
 {
     int i;
@@ -393,7 +392,7 @@ static void put_pixels_clamped_c(const DCTELEM *block, uint8_t *av_restrict pixe
     }
 }
 
-static void put_pixels_clamped4_c(const DCTELEM *block, uint8_t *av_restrict pixels,
+static void put_pixels_clamped4_c(const int16_t *block, uint8_t *av_restrict pixels,
                                  int line_size)
 {
     int i;
@@ -410,7 +409,7 @@ static void put_pixels_clamped4_c(const DCTELEM *block, uint8_t *av_restrict pix
     }
 }
 
-static void put_pixels_clamped2_c(const DCTELEM *block, uint8_t *av_restrict pixels,
+static void put_pixels_clamped2_c(const int16_t *block, uint8_t *av_restrict pixels,
                                  int line_size)
 {
     int i;
@@ -425,7 +424,7 @@ static void put_pixels_clamped2_c(const DCTELEM *block, uint8_t *av_restrict pix
     }
 }
 
-static void put_signed_pixels_clamped_c(const DCTELEM *block,
+static void put_signed_pixels_clamped_c(const int16_t *block,
                                         uint8_t *av_restrict pixels,
                                         int line_size)
 {
@@ -446,7 +445,7 @@ static void put_signed_pixels_clamped_c(const DCTELEM *block,
     }
 }
 
-static void add_pixels_clamped_c(const DCTELEM *block, uint8_t *av_restrict pixels,
+static void add_pixels_clamped_c(const int16_t *block, uint8_t *av_restrict pixels,
                                  int line_size)
 {
     int i;
@@ -466,7 +465,7 @@ static void add_pixels_clamped_c(const DCTELEM *block, uint8_t *av_restrict pixe
     }
 }
 
-static void add_pixels_clamped4_c(const DCTELEM *block, uint8_t *av_restrict pixels,
+static void add_pixels_clamped4_c(const int16_t *block, uint8_t *av_restrict pixels,
                           int line_size)
 {
     int i;
@@ -482,7 +481,7 @@ static void add_pixels_clamped4_c(const DCTELEM *block, uint8_t *av_restrict pix
     }
 }
 
-static void add_pixels_clamped2_c(const DCTELEM *block, uint8_t *av_restrict pixels,
+static void add_pixels_clamped2_c(const int16_t *block, uint8_t *av_restrict pixels,
                           int line_size)
 {
     int i;
@@ -496,7 +495,7 @@ static void add_pixels_clamped2_c(const DCTELEM *block, uint8_t *av_restrict pix
     }
 }
 
-static int sum_abs_dctelem_c(DCTELEM *block)
+static int sum_abs_dctelem_c(int16_t *block)
 {
     int sum=0, i;
     for(i=0; i<64; i++)
@@ -1834,10 +1833,10 @@ static void add_8x8basis_c(int16_t rem[64], int16_t basis[64], int scale){
  * @param scantable the used scantable, this is only used to speed the permutation up, the block is not
  *                  (inverse) permutated to scantable order!
  */
-void ff_block_permute(DCTELEM *block, uint8_t *permutation, const uint8_t *scantable, int last)
+void ff_block_permute(int16_t *block, uint8_t *permutation, const uint8_t *scantable, int last)
 {
     int i;
-    DCTELEM temp[64];
+    int16_t temp[64];
 
     if(last<=0) return;
     //if(permutation[1]==1) return; //FIXME it is ok but not clean and might fail for some permutations
@@ -2160,7 +2159,7 @@ static int hadamard8_intra8x8_c(/*MpegEncContext*/ void *s, uint8_t *src, uint8_
 
 static int dct_sad8x8_c(/*MpegEncContext*/ void *c, uint8_t *src1, uint8_t *src2, int stride, int h){
     MpegEncContext * const s= (MpegEncContext *)c;
-    LOCAL_ALIGNED_16(DCTELEM, temp, [64]);
+    LOCAL_ALIGNED_16(int16_t, temp, [64]);
 
     av_assert2(h==8);
 
@@ -2199,7 +2198,7 @@ static int dct_sad8x8_c(/*MpegEncContext*/ void *c, uint8_t *src1, uint8_t *src2
 
 static int dct264_sad8x8_c(/*MpegEncContext*/ void *c, uint8_t *src1, uint8_t *src2, int stride, int h){
     MpegEncContext * const s= (MpegEncContext *)c;
-    DCTELEM dct[8][8];
+    int16_t dct[8][8];
     int i;
     int sum=0;
 
@@ -2224,7 +2223,7 @@ static int dct264_sad8x8_c(/*MpegEncContext*/ void *c, uint8_t *src1, uint8_t *s
 
 static int dct_max8x8_c(/*MpegEncContext*/ void *c, uint8_t *src1, uint8_t *src2, int stride, int h){
     MpegEncContext * const s= (MpegEncContext *)c;
-    LOCAL_ALIGNED_16(DCTELEM, temp, [64]);
+    LOCAL_ALIGNED_16(int16_t, temp, [64]);
     int sum=0, i;
 
     av_assert2(h==8);
@@ -2240,8 +2239,8 @@ static int dct_max8x8_c(/*MpegEncContext*/ void *c, uint8_t *src1, uint8_t *src2
 
 static int quant_psnr8x8_c(/*MpegEncContext*/ void *c, uint8_t *src1, uint8_t *src2, int stride, int h){
     MpegEncContext * const s= (MpegEncContext *)c;
-    LOCAL_ALIGNED_16(DCTELEM, temp, [64*2]);
-    DCTELEM * const bak = temp+64;
+    LOCAL_ALIGNED_16(int16_t, temp, [64*2]);
+    int16_t * const bak = temp+64;
     int sum=0, i;
 
     av_assert2(h==8);
@@ -2249,7 +2248,7 @@ static int quant_psnr8x8_c(/*MpegEncContext*/ void *c, uint8_t *src1, uint8_t *s
 
     s->dsp.diff_pixels(temp, src1, src2, stride);
 
-    memcpy(bak, temp, 64*sizeof(DCTELEM));
+    memcpy(bak, temp, 64*sizeof(int16_t));
 
     s->block_last_index[0/*FIXME*/]= s->fast_dct_quantize(s, temp, 0/*FIXME*/, s->qscale, &i);
     s->dct_unquantize_inter(s, temp, 0, s->qscale);
@@ -2264,7 +2263,7 @@ static int quant_psnr8x8_c(/*MpegEncContext*/ void *c, uint8_t *src1, uint8_t *s
 static int rd8x8_c(/*MpegEncContext*/ void *c, uint8_t *src1, uint8_t *src2, int stride, int h){
     MpegEncContext * const s= (MpegEncContext *)c;
     const uint8_t *scantable= s->intra_scantable.permutated;
-    LOCAL_ALIGNED_16(DCTELEM, temp, [64]);
+    LOCAL_ALIGNED_16(int16_t, temp, [64]);
     LOCAL_ALIGNED_16(uint8_t, lsrc1, [64]);
     LOCAL_ALIGNED_16(uint8_t, lsrc2, [64]);
     int i, last, run, bits, level, distortion, start_i;
@@ -2340,7 +2339,7 @@ static int rd8x8_c(/*MpegEncContext*/ void *c, uint8_t *src1, uint8_t *src2, int
 static int bit8x8_c(/*MpegEncContext*/ void *c, uint8_t *src1, uint8_t *src2, int stride, int h){
     MpegEncContext * const s= (MpegEncContext *)c;
     const uint8_t *scantable= s->intra_scantable.permutated;
-    LOCAL_ALIGNED_16(DCTELEM, temp, [64]);
+    LOCAL_ALIGNED_16(int16_t, temp, [64]);
     int i, last, run, bits, level, start_i;
     const int esc_length= s->ac_esc_length;
     uint8_t * length;
@@ -2577,44 +2576,44 @@ static void vector_clip_int32_c(int32_t *dst, const int32_t *src, int32_t min,
     } while (len > 0);
 }
 
-static void ff_jref_idct_put(uint8_t *dest, int line_size, DCTELEM *block)
+static void ff_jref_idct_put(uint8_t *dest, int line_size, int16_t *block)
 {
     ff_j_rev_dct (block);
     put_pixels_clamped_c(block, dest, line_size);
 }
-static void ff_jref_idct_add(uint8_t *dest, int line_size, DCTELEM *block)
+static void ff_jref_idct_add(uint8_t *dest, int line_size, int16_t *block)
 {
     ff_j_rev_dct (block);
     add_pixels_clamped_c(block, dest, line_size);
 }
 
-static void ff_jref_idct4_put(uint8_t *dest, int line_size, DCTELEM *block)
+static void ff_jref_idct4_put(uint8_t *dest, int line_size, int16_t *block)
 {
     ff_j_rev_dct4 (block);
     put_pixels_clamped4_c(block, dest, line_size);
 }
-static void ff_jref_idct4_add(uint8_t *dest, int line_size, DCTELEM *block)
+static void ff_jref_idct4_add(uint8_t *dest, int line_size, int16_t *block)
 {
     ff_j_rev_dct4 (block);
     add_pixels_clamped4_c(block, dest, line_size);
 }
 
-static void ff_jref_idct2_put(uint8_t *dest, int line_size, DCTELEM *block)
+static void ff_jref_idct2_put(uint8_t *dest, int line_size, int16_t *block)
 {
     ff_j_rev_dct2 (block);
     put_pixels_clamped2_c(block, dest, line_size);
 }
-static void ff_jref_idct2_add(uint8_t *dest, int line_size, DCTELEM *block)
+static void ff_jref_idct2_add(uint8_t *dest, int line_size, int16_t *block)
 {
     ff_j_rev_dct2 (block);
     add_pixels_clamped2_c(block, dest, line_size);
 }
 
-static void ff_jref_idct1_put(uint8_t *dest, int line_size, DCTELEM *block)
+static void ff_jref_idct1_put(uint8_t *dest, int line_size, int16_t *block)
 {
     dest[0] = av_clip_uint8((block[0] + 4)>>3);
 }
-static void ff_jref_idct1_add(uint8_t *dest, int line_size, DCTELEM *block)
+static void ff_jref_idct1_add(uint8_t *dest, int line_size, int16_t *block)
 {
     dest[0] = av_clip_uint8(dest[0] + ((block[0] + 4)>>3));
 }
diff --git a/libavcodec/dsputil.h b/libavcodec/dsputil.h
index 0b27ebd54b..606555a8d4 100644
--- a/libavcodec/dsputil.h
+++ b/libavcodec/dsputil.h
@@ -36,37 +36,36 @@
 
 //#define DEBUG
 /* dct code */
-typedef short DCTELEM;
 
-void ff_fdct_ifast (DCTELEM *data);
-void ff_fdct_ifast248 (DCTELEM *data);
-void ff_jpeg_fdct_islow_8(DCTELEM *data);
-void ff_jpeg_fdct_islow_10(DCTELEM *data);
-void ff_fdct248_islow_8(DCTELEM *data);
-void ff_fdct248_islow_10(DCTELEM *data);
+void ff_fdct_ifast(int16_t *data);
+void ff_fdct_ifast248(int16_t *data);
+void ff_jpeg_fdct_islow_8(int16_t *data);
+void ff_jpeg_fdct_islow_10(int16_t *data);
+void ff_fdct248_islow_8(int16_t *data);
+void ff_fdct248_islow_10(int16_t *data);
 
-void ff_j_rev_dct (DCTELEM *data);
-void ff_j_rev_dct4 (DCTELEM *data);
-void ff_j_rev_dct2 (DCTELEM *data);
-void ff_j_rev_dct1 (DCTELEM *data);
+void ff_j_rev_dct(int16_t *data);
+void ff_j_rev_dct4(int16_t *data);
+void ff_j_rev_dct2(int16_t *data);
+void ff_j_rev_dct1(int16_t *data);
 
-void ff_fdct_mmx(DCTELEM *block);
-void ff_fdct_mmxext(DCTELEM *block);
-void ff_fdct_sse2(DCTELEM *block);
+void ff_fdct_mmx(int16_t *block);
+void ff_fdct_mmxext(int16_t *block);
+void ff_fdct_sse2(int16_t *block);
 
 #define H264_IDCT(depth) \
-void ff_h264_idct8_add_ ## depth ## _c(uint8_t *dst, DCTELEM *block, int stride);\
-void ff_h264_idct_add_ ## depth ## _c(uint8_t *dst, DCTELEM *block, int stride);\
-void ff_h264_idct8_dc_add_ ## depth ## _c(uint8_t *dst, DCTELEM *block, int stride);\
-void ff_h264_idct_dc_add_ ## depth ## _c(uint8_t *dst, DCTELEM *block, int stride);\
-void ff_h264_idct_add16_ ## depth ## _c(uint8_t *dst, const int *blockoffset, DCTELEM *block, int stride, const uint8_t nnzc[6*8]);\
-void ff_h264_idct_add16intra_ ## depth ## _c(uint8_t *dst, const int *blockoffset, DCTELEM *block, int stride, const uint8_t nnzc[6*8]);\
-void ff_h264_idct8_add4_ ## depth ## _c(uint8_t *dst, const int *blockoffset, DCTELEM *block, int stride, const uint8_t nnzc[6*8]);\
-void ff_h264_idct_add8_422_ ## depth ## _c(uint8_t **dest, const int *blockoffset, DCTELEM *block, int stride, const uint8_t nnzc[6*8]);\
-void ff_h264_idct_add8_ ## depth ## _c(uint8_t **dest, const int *blockoffset, DCTELEM *block, int stride, const uint8_t nnzc[6*8]);\
-void ff_h264_luma_dc_dequant_idct_ ## depth ## _c(DCTELEM *output, DCTELEM *input, int qmul);\
-void ff_h264_chroma422_dc_dequant_idct_ ## depth ## _c(DCTELEM *block, int qmul);\
-void ff_h264_chroma_dc_dequant_idct_ ## depth ## _c(DCTELEM *block, int qmul);
+void ff_h264_idct8_add_ ## depth ## _c(uint8_t *dst, int16_t *block, int stride);\
+void ff_h264_idct_add_ ## depth ## _c(uint8_t *dst, int16_t *block, int stride);\
+void ff_h264_idct8_dc_add_ ## depth ## _c(uint8_t *dst, int16_t *block, int stride);\
+void ff_h264_idct_dc_add_ ## depth ## _c(uint8_t *dst, int16_t *block, int stride);\
+void ff_h264_idct_add16_ ## depth ## _c(uint8_t *dst, const int *blockoffset, int16_t *block, int stride, const uint8_t nnzc[6*8]);\
+void ff_h264_idct_add16intra_ ## depth ## _c(uint8_t *dst, const int *blockoffset, int16_t *block, int stride, const uint8_t nnzc[6*8]);\
+void ff_h264_idct8_add4_ ## depth ## _c(uint8_t *dst, const int *blockoffset, int16_t *block, int stride, const uint8_t nnzc[6*8]);\
+void ff_h264_idct_add8_422_ ## depth ## _c(uint8_t **dest, const int *blockoffset, int16_t *block, int stride, const uint8_t nnzc[6*8]);\
+void ff_h264_idct_add8_ ## depth ## _c(uint8_t **dest, const int *blockoffset, int16_t *block, int stride, const uint8_t nnzc[6*8]);\
+void ff_h264_luma_dc_dequant_idct_ ## depth ## _c(int16_t *output, int16_t *input, int qmul);\
+void ff_h264_chroma422_dc_dequant_idct_ ## depth ## _c(int16_t *block, int qmul);\
+void ff_h264_chroma_dc_dequant_idct_ ## depth ## _c(int16_t *block, int qmul);
 
 H264_IDCT( 8)
 H264_IDCT( 9)
@@ -74,8 +73,8 @@ H264_IDCT(10)
 H264_IDCT(12)
 H264_IDCT(14)
 
-void ff_svq3_luma_dc_dequant_idct_c(DCTELEM *output, DCTELEM *input, int qp);
-void ff_svq3_add_idct_c(uint8_t *dst, DCTELEM *block, int stride, int qp, int dc);
+void ff_svq3_luma_dc_dequant_idct_c(int16_t *output, int16_t *input, int qp);
+void ff_svq3_add_idct_c(uint8_t *dst, int16_t *block, int stride, int qp, int dc);
 
 /* encoding scans */
 extern const uint8_t ff_alternate_horizontal_scan[64];
@@ -135,11 +134,11 @@ could be reached easily ...
 */
 
 /*
-void get_pixels_c(DCTELEM *block, const uint8_t *pixels, int line_size);
-void diff_pixels_c(DCTELEM *block, const uint8_t *s1, const uint8_t *s2, int stride);
-void put_pixels_clamped_c(const DCTELEM *block, uint8_t *pixels, int line_size);
-void add_pixels_clamped_c(const DCTELEM *block, uint8_t *pixels, int line_size);
-void clear_blocks_c(DCTELEM *blocks);
+void get_pixels_c(int16_t *block, const uint8_t *pixels, int line_size);
+void diff_pixels_c(int16_t *block, const uint8_t *s1, const uint8_t *s2, int stride);
+void put_pixels_clamped_c(const int16_t *block, uint8_t *pixels, int line_size);
+void add_pixels_clamped_c(const int16_t *block, uint8_t *pixels, int line_size);
+void clear_blocks_c(int16_t *blocks);
 */
 
 /* add and put pixel (decoding) */
@@ -212,14 +211,14 @@ typedef struct DSPContext {
     int dct_bits;
 
     /* pixel ops : interface with DCT */
-    void (*get_pixels)(DCTELEM *block/*align 16*/, const uint8_t *pixels/*align 8*/, int line_size);
-    void (*diff_pixels)(DCTELEM *block/*align 16*/, const uint8_t *s1/*align 8*/, const uint8_t *s2/*align 8*/, int stride);
-    void (*put_pixels_clamped)(const DCTELEM *block/*align 16*/, uint8_t *pixels/*align 8*/, int line_size);
-    void (*put_signed_pixels_clamped)(const DCTELEM *block/*align 16*/, uint8_t *pixels/*align 8*/, int line_size);
-    void (*add_pixels_clamped)(const DCTELEM *block/*align 16*/, uint8_t *pixels/*align 8*/, int line_size);
-    void (*add_pixels8)(uint8_t *pixels, DCTELEM *block, int line_size);
-    void (*add_pixels4)(uint8_t *pixels, DCTELEM *block, int line_size);
-    int (*sum_abs_dctelem)(DCTELEM *block/*align 16*/);
+    void (*get_pixels)(int16_t *block/*align 16*/, const uint8_t *pixels/*align 8*/, int line_size);
+    void (*diff_pixels)(int16_t *block/*align 16*/, const uint8_t *s1/*align 8*/, const uint8_t *s2/*align 8*/, int stride);
+    void (*put_pixels_clamped)(const int16_t *block/*align 16*/, uint8_t *pixels/*align 8*/, int line_size);
+    void (*put_signed_pixels_clamped)(const int16_t *block/*align 16*/, uint8_t *pixels/*align 8*/, int line_size);
+    void (*add_pixels_clamped)(const int16_t *block/*align 16*/, uint8_t *pixels/*align 8*/, int line_size);
+    void (*add_pixels8)(uint8_t *pixels, int16_t *block, int line_size);
+    void (*add_pixels4)(uint8_t *pixels, int16_t *block, int line_size);
+    int (*sum_abs_dctelem)(int16_t *block/*align 16*/);
     /**
      * translational global motion compensation.
      */
@@ -229,8 +228,8 @@ typedef struct DSPContext {
      */
     void (*gmc )(uint8_t *dst/*align 8*/, uint8_t *src/*align 1*/, int stride, int h, int ox, int oy,
                     int dxx, int dxy, int dyx, int dyy, int shift, int r, int width, int height);
-    void (*clear_block)(DCTELEM *block/*align 16*/);
-    void (*clear_blocks)(DCTELEM *blocks/*align 16*/);
+    void (*clear_block)(int16_t *block/*align 16*/);
+    void (*clear_blocks)(int16_t *blocks/*align 16*/);
     int (*pix_sum)(uint8_t * pix, int line_size);
     int (*pix_norm1)(uint8_t * pix, int line_size);
 // 16x16 8x8 4x4 2x2 16x8 8x4 4x2 8x16 4x8 2x4
@@ -362,24 +361,24 @@ typedef struct DSPContext {
     void (*vector_clipf)(float *dst /* align 16 */, const float *src /* align 16 */, float min, float max, int len /* align 16 */);
 
     /* (I)DCT */
-    void (*fdct)(DCTELEM *block/* align 16*/);
-    void (*fdct248)(DCTELEM *block/* align 16*/);
+    void (*fdct)(int16_t *block/* align 16*/);
+    void (*fdct248)(int16_t *block/* align 16*/);
 
     /* IDCT really*/
-    void (*idct)(DCTELEM *block/* align 16*/);
+    void (*idct)(int16_t *block/* align 16*/);
 
     /**
      * block -> idct -> clip to unsigned 8 bit -> dest.
      * (-1392, 0, 0, ...) -> idct -> (-174, -174, ...) -> put -> (0, 0, ...)
      * @param line_size size in bytes of a horizontal line of dest
      */
-    void (*idct_put)(uint8_t *dest/*align 8*/, int line_size, DCTELEM *block/*align 16*/);
+    void (*idct_put)(uint8_t *dest/*align 8*/, int line_size, int16_t *block/*align 16*/);
 
     /**
      * block -> idct -> add dest -> clip to unsigned 8 bit -> dest.
      * @param line_size size in bytes of a horizontal line of dest
      */
-    void (*idct_add)(uint8_t *dest/*align 8*/, int line_size, DCTELEM *block/*align 16*/);
+    void (*idct_add)(uint8_t *dest/*align 8*/, int line_size, int16_t *block/*align 16*/);
 
     /**
      * idct input permutation.
@@ -470,7 +469,7 @@ int ff_check_alignment(void);
  * permute block according to permuatation.
  * @param last last non zero element in scantable order
  */
-void ff_block_permute(DCTELEM *block, uint8_t *permutation, const uint8_t *scantable, int last);
+void ff_block_permute(int16_t *block, uint8_t *permutation, const uint8_t *scantable, int last);
 
 void ff_set_cmp(DSPContext* c, me_cmp_func *cmp, int type);
 
diff --git a/libavcodec/dsputil_template.c b/libavcodec/dsputil_template.c
index 5162304d05..3c98086a64 100644
--- a/libavcodec/dsputil_template.c
+++ b/libavcodec/dsputil_template.c
@@ -114,7 +114,7 @@ static void FUNCC(draw_edges)(uint8_t *p_buf, int p_wrap, int width, int height,
 }
 
 #define DCTELEM_FUNCS(dctcoef, suffix)                                  \
-static void FUNCC(get_pixels ## suffix)(DCTELEM *av_restrict _block,    \
+static void FUNCC(get_pixels ## suffix)(int16_t *av_restrict _block,    \
                                         const uint8_t *_pixels,         \
                                         int line_size)                  \
 {                                                                       \
@@ -138,7 +138,7 @@ static void FUNCC(get_pixels ## suffix)(DCTELEM *av_restrict _block,    \
 }                                                                       \
                                                                         \
 static void FUNCC(add_pixels8 ## suffix)(uint8_t *av_restrict _pixels,  \
-                                         DCTELEM *_block,               \
+                                         int16_t *_block,               \
                                          int line_size)                 \
 {                                                                       \
     int i;                                                              \
@@ -161,7 +161,7 @@ static void FUNCC(add_pixels8 ## suffix)(uint8_t *av_restrict _pixels,  \
 }                                                                       \
                                                                         \
 static void FUNCC(add_pixels4 ## suffix)(uint8_t *av_restrict _pixels,  \
-                                         DCTELEM *_block,               \
+                                         int16_t *_block,               \
                                          int line_size)                 \
 {                                                                       \
     int i;                                                              \
@@ -179,20 +179,20 @@ static void FUNCC(add_pixels4 ## suffix)(uint8_t *av_restrict _pixels,  \
     }                                                                   \
 }                                                                       \
                                                                         \
-static void FUNCC(clear_block ## suffix)(DCTELEM *block)                \
+static void FUNCC(clear_block ## suffix)(int16_t *block)                \
 {                                                                       \
     memset(block, 0, sizeof(dctcoef)*64);                               \
 }                                                                       \
                                                                         \
 /**                                                                     \
- * memset(blocks, 0, sizeof(DCTELEM)*6*64)                              \
+ * memset(blocks, 0, sizeof(int16_t)*6*64)                              \
  */                                                                     \
-static void FUNCC(clear_blocks ## suffix)(DCTELEM *blocks)              \
+static void FUNCC(clear_blocks ## suffix)(int16_t *blocks)              \
 {                                                                       \
     memset(blocks, 0, sizeof(dctcoef)*6*64);                            \
 }
 
-DCTELEM_FUNCS(DCTELEM, _16)
+DCTELEM_FUNCS(int16_t, _16)
 #if BIT_DEPTH > 8
 DCTELEM_FUNCS(dctcoef, _32)
 #endif
diff --git a/libavcodec/dv.c b/libavcodec/dv.c
index 18c90c2b6c..e2a00a6c14 100644
--- a/libavcodec/dv.c
+++ b/libavcodec/dv.c
@@ -417,7 +417,7 @@ typedef struct EncBlockInfo {
     int      cur_ac;
     int      cno;
     int      dct_mode;
-    DCTELEM  mb[64];
+    int16_t  mb[64];
     uint8_t  next[64];
     uint8_t  sign[64];
     uint8_t  partial_bit_count;
@@ -506,7 +506,7 @@ static av_always_inline int dv_init_enc_block(EncBlockInfo* bi, uint8_t *data, i
 {
     const int *weight;
     const uint8_t* zigzag_scan;
-    LOCAL_ALIGNED_16(DCTELEM, blk, [64]);
+    LOCAL_ALIGNED_16(int16_t, blk, [64]);
     int i, area;
     /* We offer two different methods for class number assignment: the
        method suggested in SMPTE 314M Table 22, and an improved
diff --git a/libavcodec/dvdata.h b/libavcodec/dvdata.h
index b1fa44ae39..97c20e3a1a 100644
--- a/libavcodec/dvdata.h
+++ b/libavcodec/dvdata.h
@@ -40,9 +40,9 @@ typedef struct DVVideoContext {
 
     uint8_t  dv_zigzag[2][64];
 
-    void (*get_pixels)(DCTELEM *block, const uint8_t *pixels, int line_size);
-    void (*fdct[2])(DCTELEM *block);
-    void (*idct_put[2])(uint8_t *dest, int line_size, DCTELEM *block);
+    void (*get_pixels)(int16_t *block, const uint8_t *pixels, int line_size);
+    void (*fdct[2])(int16_t *block);
+    void (*idct_put[2])(uint8_t *dest, int line_size, int16_t *block);
     me_cmp_func ildct_cmp;
 } DVVideoContext;
 
diff --git a/libavcodec/dvdec.c b/libavcodec/dvdec.c
index fd00cc18d6..bb60fc0b13 100644
--- a/libavcodec/dvdec.c
+++ b/libavcodec/dvdec.c
@@ -49,7 +49,7 @@ typedef struct BlockInfo {
     const uint32_t *factor_table;
     const uint8_t *scan_table;
     uint8_t pos; /* position in block */
-    void (*idct_put)(uint8_t *dest, int line_size, DCTELEM *block);
+    void (*idct_put)(uint8_t *dest, int line_size, int16_t *block);
     uint8_t partial_bit_count;
     uint32_t partial_bit_buffer;
     int shift_offset;
@@ -58,7 +58,7 @@ typedef struct BlockInfo {
 static const int dv_iweight_bits = 14;
 
 /* decode AC coefficients */
-static void dv_decode_ac(GetBitContext *gb, BlockInfo *mb, DCTELEM *block)
+static void dv_decode_ac(GetBitContext *gb, BlockInfo *mb, int16_t *block)
 {
     int last_index = gb->size_in_bits;
     const uint8_t  *scan_table   = mb->scan_table;
@@ -136,14 +136,14 @@ static int dv_decode_video_segment(AVCodecContext *avctx, void *arg)
     int quant, dc, dct_mode, class1, j;
     int mb_index, mb_x, mb_y, last_index;
     int y_stride, linesize;
-    DCTELEM *block, *block1;
+    int16_t *block, *block1;
     int c_offset;
     uint8_t *y_ptr;
     const uint8_t *buf_ptr;
     PutBitContext pb, vs_pb;
     GetBitContext gb;
     BlockInfo mb_data[5 * DV_MAX_BPM], *mb, *mb1;
-    LOCAL_ALIGNED_16(DCTELEM, sblock, [5*DV_MAX_BPM], [64]);
+    LOCAL_ALIGNED_16(int16_t, sblock, [5*DV_MAX_BPM], [64]);
     LOCAL_ALIGNED_16(uint8_t, mb_bit_buffer, [  80 + FF_INPUT_BUFFER_PADDING_SIZE]); /* allow some slack */
     LOCAL_ALIGNED_16(uint8_t, vs_bit_buffer, [5*80 + FF_INPUT_BUFFER_PADDING_SIZE]); /* allow some slack */
     const int log2_blocksize = 3-s->avctx->lowres;
diff --git a/libavcodec/eaidct.c b/libavcodec/eaidct.c
index 175c1c47c0..6ba1faf614 100644
--- a/libavcodec/eaidct.c
+++ b/libavcodec/eaidct.c
@@ -64,7 +64,7 @@
 #define MUNGE_8BIT(x) av_clip_uint8((x)>>4)
 #define IDCT_ROW(dest,src) IDCT_TRANSFORM(dest,0,1,2,3,4,5,6,7,0,1,2,3,4,5,6,7,MUNGE_8BIT,src)
 
-static inline void ea_idct_col(DCTELEM *dest, const DCTELEM *src) {
+static inline void ea_idct_col(int16_t *dest, const int16_t *src) {
     if ((src[8]|src[16]|src[24]|src[32]|src[40]|src[48]|src[56])==0) {
         dest[0]  =
         dest[8]  =
@@ -78,9 +78,9 @@ static inline void ea_idct_col(DCTELEM *dest, const DCTELEM *src) {
         IDCT_COL(dest, src);
 }
 
-void ff_ea_idct_put_c(uint8_t *dest, int linesize, DCTELEM *block) {
+void ff_ea_idct_put_c(uint8_t *dest, int linesize, int16_t *block) {
     int i;
-    DCTELEM temp[64];
+    int16_t temp[64];
     block[0] += 4;
     for (i=0; i<8; i++)
         ea_idct_col(&temp[i], &block[i]);
diff --git a/libavcodec/eaidct.h b/libavcodec/eaidct.h
index a436673e88..6b9ec1c91c 100644
--- a/libavcodec/eaidct.h
+++ b/libavcodec/eaidct.h
@@ -20,8 +20,7 @@
 #define AVCODEC_EAIDCT_H
 
 #include <stdint.h>
-#include "dsputil.h"
 
-void ff_ea_idct_put_c(uint8_t *dest, int linesize, DCTELEM *block);
+void ff_ea_idct_put_c(uint8_t *dest, int linesize, int16_t *block);
 
 #endif /* AVCODEC_EAIDCT_H */
diff --git a/libavcodec/eamad.c b/libavcodec/eamad.c
index 2f9559fe8e..75285dccec 100644
--- a/libavcodec/eamad.c
+++ b/libavcodec/eamad.c
@@ -51,7 +51,7 @@ typedef struct MadContext {
     GetBitContext gb;
     void *bitstream_buf;
     unsigned int bitstream_buf_size;
-    DECLARE_ALIGNED(16, DCTELEM, block)[64];
+    DECLARE_ALIGNED(16, int16_t, block)[64];
     ScanTable scantable;
     uint16_t quant_matrix[64];
     int mb_x;
@@ -102,7 +102,7 @@ static inline void comp_block(MadContext *t, int mb_x, int mb_y,
     }
 }
 
-static inline void idct_put(MadContext *t, DCTELEM *block, int mb_x, int mb_y, int j)
+static inline void idct_put(MadContext *t, int16_t *block, int mb_x, int mb_y, int j)
 {
     if (j < 4) {
         ff_ea_idct_put_c(
@@ -116,7 +116,7 @@ static inline void idct_put(MadContext *t, DCTELEM *block, int mb_x, int mb_y, i
     }
 }
 
-static inline int decode_block_intra(MadContext *s, DCTELEM * block)
+static inline int decode_block_intra(MadContext *s, int16_t * block)
 {
     int level, i, j, run;
     RLTable *rl = &ff_rl_mpeg1;
diff --git a/libavcodec/eatgq.c b/libavcodec/eatgq.c
index c38b3cdf9c..fba6e831cf 100644
--- a/libavcodec/eatgq.c
+++ b/libavcodec/eatgq.c
@@ -43,7 +43,7 @@ typedef struct TgqContext {
     int width,height;
     ScanTable scantable;
     int qtable[64];
-    DECLARE_ALIGNED(16, DCTELEM, block)[6][64];
+    DECLARE_ALIGNED(16, int16_t, block)[6][64];
     GetByteContext gb;
 } TgqContext;
 
@@ -58,7 +58,7 @@ static av_cold int tgq_decode_init(AVCodecContext *avctx){
     return 0;
 }
 
-static void tgq_decode_block(TgqContext *s, DCTELEM block[64], GetBitContext *gb){
+static void tgq_decode_block(TgqContext *s, int16_t block[64], GetBitContext *gb){
     uint8_t *perm = s->scantable.permutated;
     int i,j,value;
     block[0] = get_sbits(gb,8) * s->qtable[0];
@@ -103,7 +103,7 @@ static void tgq_decode_block(TgqContext *s, DCTELEM block[64], GetBitContext *gb
     block[0] += 128<<4;
 }
 
-static void tgq_idct_put_mb(TgqContext *s, DCTELEM (*block)[64], int mb_x, int mb_y){
+static void tgq_idct_put_mb(TgqContext *s, int16_t (*block)[64], int mb_x, int mb_y){
     int linesize= s->frame.linesize[0];
     uint8_t *dest_y  = s->frame.data[0] + (mb_y * 16* linesize            ) + mb_x * 16;
     uint8_t *dest_cb = s->frame.data[1] + (mb_y * 8 * s->frame.linesize[1]) + mb_x * 8;
diff --git a/libavcodec/eatqi.c b/libavcodec/eatqi.c
index 306331f359..6f535efa7a 100644
--- a/libavcodec/eatqi.c
+++ b/libavcodec/eatqi.c
@@ -40,7 +40,7 @@ typedef struct TqiContext {
     AVFrame frame;
     void *bitstream_buf;
     unsigned int bitstream_buf_size;
-    DECLARE_ALIGNED(16, DCTELEM, block)[6][64];
+    DECLARE_ALIGNED(16, int16_t, block)[6][64];
 } TqiContext;
 
 static av_cold int tqi_decode_init(AVCodecContext *avctx)
@@ -58,7 +58,7 @@ static av_cold int tqi_decode_init(AVCodecContext *avctx)
     return 0;
 }
 
-static int tqi_decode_mb(MpegEncContext *s, DCTELEM (*block)[64])
+static int tqi_decode_mb(MpegEncContext *s, int16_t (*block)[64])
 {
     int n;
     s->dsp.clear_blocks(block[0]);
@@ -69,7 +69,7 @@ static int tqi_decode_mb(MpegEncContext *s, DCTELEM (*block)[64])
     return 0;
 }
 
-static inline void tqi_idct_put(TqiContext *t, DCTELEM (*block)[64])
+static inline void tqi_idct_put(TqiContext *t, int16_t (*block)[64])
 {
     MpegEncContext *s = &t->s;
     int linesize= t->frame.linesize[0];
diff --git a/libavcodec/faandct.c b/libavcodec/faandct.c
index 1379394aa9..0219e20eab 100644
--- a/libavcodec/faandct.c
+++ b/libavcodec/faandct.c
@@ -64,7 +64,7 @@ B6*B0, B6*B1, B6*B2, B6*B3, B6*B4, B6*B5, B6*B6, B6*B7,
 B7*B0, B7*B1, B7*B2, B7*B3, B7*B4, B7*B5, B7*B6, B7*B7,
 };
 
-static av_always_inline void row_fdct(FLOAT temp[64], DCTELEM * data)
+static av_always_inline void row_fdct(FLOAT temp[64], int16_t *data)
 {
     FLOAT tmp0, tmp1, tmp2, tmp3, tmp4, tmp5, tmp6, tmp7;
     FLOAT tmp10, tmp11, tmp12, tmp13;
@@ -119,7 +119,7 @@ static av_always_inline void row_fdct(FLOAT temp[64], DCTELEM * data)
     }
 }
 
-void ff_faandct(DCTELEM * data)
+void ff_faandct(int16_t *data)
 {
     FLOAT tmp0, tmp1, tmp2, tmp3, tmp4, tmp5, tmp6, tmp7;
     FLOAT tmp10, tmp11, tmp12, tmp13;
@@ -179,7 +179,7 @@ void ff_faandct(DCTELEM * data)
     }
 }
 
-void ff_faandct248(DCTELEM * data)
+void ff_faandct248(int16_t *data)
 {
     FLOAT tmp0, tmp1, tmp2, tmp3, tmp4, tmp5, tmp6, tmp7;
     FLOAT tmp10, tmp11, tmp12, tmp13;
diff --git a/libavcodec/faandct.h b/libavcodec/faandct.h
index a12dcaf92f..c5ef96dcf1 100644
--- a/libavcodec/faandct.h
+++ b/libavcodec/faandct.h
@@ -29,9 +29,9 @@
 #ifndef AVCODEC_FAANDCT_H
 #define AVCODEC_FAANDCT_H
 
-#include "dsputil.h"
+#include <stdint.h>
 
-void ff_faandct(DCTELEM * data);
-void ff_faandct248(DCTELEM * data);
+void ff_faandct(int16_t *data);
+void ff_faandct248(int16_t *data);
 
 #endif /* AVCODEC_FAANDCT_H */
diff --git a/libavcodec/faanidct.c b/libavcodec/faanidct.c
index 1e9bfaf48a..5f34fa508f 100644
--- a/libavcodec/faanidct.c
+++ b/libavcodec/faanidct.c
@@ -47,7 +47,7 @@ B6*B0/8, B6*B1/8, B6*B2/8, B6*B3/8, B6*B4/8, B6*B5/8, B6*B6/8, B6*B7/8,
 B7*B0/8, B7*B1/8, B7*B2/8, B7*B3/8, B7*B4/8, B7*B5/8, B7*B6/8, B7*B7/8,
 };
 
-static inline void p8idct(DCTELEM data[64], FLOAT temp[64], uint8_t *dest, int stride, int x, int y, int type){
+static inline void p8idct(int16_t data[64], FLOAT temp[64], uint8_t *dest, int stride, int x, int y, int type){
     int i;
     FLOAT av_unused tmp0;
     FLOAT s04, d04, s17, d17, s26, d26, s53, d53;
@@ -129,7 +129,7 @@ static inline void p8idct(DCTELEM data[64], FLOAT temp[64], uint8_t *dest, int s
     }
 }
 
-void ff_faanidct(DCTELEM block[64]){
+void ff_faanidct(int16_t block[64]){
     FLOAT temp[64];
     int i;
 
@@ -142,7 +142,7 @@ void ff_faanidct(DCTELEM block[64]){
     p8idct(block, temp, NULL, 0, 8, 1, 1);
 }
 
-void ff_faanidct_add(uint8_t *dest, int line_size, DCTELEM block[64]){
+void ff_faanidct_add(uint8_t *dest, int line_size, int16_t block[64]){
     FLOAT temp[64];
     int i;
 
@@ -155,7 +155,7 @@ void ff_faanidct_add(uint8_t *dest, int line_size, DCTELEM block[64]){
     p8idct(NULL , temp, dest, line_size, 8, 1, 2);
 }
 
-void ff_faanidct_put(uint8_t *dest, int line_size, DCTELEM block[64]){
+void ff_faanidct_put(uint8_t *dest, int line_size, int16_t block[64]){
     FLOAT temp[64];
     int i;
 
diff --git a/libavcodec/faanidct.h b/libavcodec/faanidct.h
index 4cf1189286..4cd2c7858f 100644
--- a/libavcodec/faanidct.h
+++ b/libavcodec/faanidct.h
@@ -23,10 +23,9 @@
 #define AVCODEC_FAANIDCT_H
 
 #include <stdint.h>
-#include "dsputil.h"
 
-void ff_faanidct(DCTELEM block[64]);
-void ff_faanidct_add(uint8_t *dest, int line_size, DCTELEM block[64]);
-void ff_faanidct_put(uint8_t *dest, int line_size, DCTELEM block[64]);
+void ff_faanidct(int16_t block[64]);
+void ff_faanidct_add(uint8_t *dest, int line_size, int16_t block[64]);
+void ff_faanidct_put(uint8_t *dest, int line_size, int16_t block[64]);
 
 #endif /* AVCODEC_FAANIDCT_H */
diff --git a/libavcodec/h261dec.c b/libavcodec/h261dec.c
index 7cca7bb82e..22fa031a51 100644
--- a/libavcodec/h261dec.c
+++ b/libavcodec/h261dec.c
@@ -48,7 +48,7 @@ static VLC h261_mtype_vlc;
 static VLC h261_mv_vlc;
 static VLC h261_cbp_vlc;
 
-static int h261_decode_block(H261Context * h, DCTELEM * block, int n, int coded);
+static int h261_decode_block(H261Context * h, int16_t * block, int n, int coded);
 
 static av_cold void h261_decode_init_vlc(H261Context *h){
     static int done = 0;
@@ -366,7 +366,7 @@ intra:
  * Decode a macroblock.
  * @return <0 if an error occurred
  */
-static int h261_decode_block(H261Context * h, DCTELEM * block,
+static int h261_decode_block(H261Context * h, int16_t * block,
                              int n, int coded)
 {
     MpegEncContext * const s = &h->s;
diff --git a/libavcodec/h261enc.c b/libavcodec/h261enc.c
index 29bee5d46e..edf01f99b5 100644
--- a/libavcodec/h261enc.c
+++ b/libavcodec/h261enc.c
@@ -35,7 +35,7 @@
 
 extern uint8_t ff_h261_rl_table_store[2][2*MAX_RUN + MAX_LEVEL + 3];
 
-static void h261_encode_block(H261Context * h, DCTELEM * block,
+static void h261_encode_block(H261Context * h, int16_t * block,
                               int n);
 
 int ff_h261_get_picture_format(int width, int height){
@@ -144,7 +144,7 @@ static void h261_encode_motion(H261Context * h, int val){
 }
 
 static inline int get_cbp(MpegEncContext * s,
-                      DCTELEM block[6][64])
+                      int16_t block[6][64])
 {
     int i, cbp;
     cbp= 0;
@@ -155,7 +155,7 @@ static inline int get_cbp(MpegEncContext * s,
     return cbp;
 }
 void ff_h261_encode_mb(MpegEncContext * s,
-         DCTELEM block[6][64],
+         int16_t block[6][64],
          int motion_x, int motion_y)
 {
     H261Context * h = (H261Context *)s;
@@ -256,7 +256,7 @@ void ff_h261_encode_init(MpegEncContext *s){
  * @param block the 8x8 block
  * @param n block index (0-3 are luma, 4-5 are chroma)
  */
-static void h261_encode_block(H261Context * h, DCTELEM * block, int n){
+static void h261_encode_block(H261Context * h, int16_t * block, int n){
     MpegEncContext * const s = &h->s;
     int level, run, i, j, last_index, last_non_zero, sign, slevel, code;
     RLTable *rl;
diff --git a/libavcodec/h263.c b/libavcodec/h263.c
index 5b77106d7d..2390e0d945 100644
--- a/libavcodec/h263.c
+++ b/libavcodec/h263.c
@@ -226,7 +226,7 @@ void ff_h263_loop_filter(MpegEncContext * s){
     }
 }
 
-void ff_h263_pred_acdc(MpegEncContext * s, DCTELEM *block, int n)
+void ff_h263_pred_acdc(MpegEncContext * s, int16_t *block, int n)
 {
     int x, y, wrap, a, c, pred_dc, scale, i;
     int16_t *dc_val, *ac_val, *ac_val1;
diff --git a/libavcodec/h263.h b/libavcodec/h263.h
index 34d79a088d..3cb06f387a 100644
--- a/libavcodec/h263.h
+++ b/libavcodec/h263.h
@@ -73,7 +73,7 @@ int ff_h263_decode_frame(AVCodecContext *avctx,
                              AVPacket *avpkt);
 int ff_h263_decode_end(AVCodecContext *avctx);
 void ff_h263_encode_mb(MpegEncContext *s,
-                       DCTELEM block[6][64],
+                       int16_t block[6][64],
                        int motion_x, int motion_y);
 void ff_h263_encode_picture_header(MpegEncContext *s, int picture_number);
 void ff_h263_encode_gob_header(MpegEncContext * s, int mb_line);
@@ -89,7 +89,7 @@ int ff_h263_decode_mba(MpegEncContext *s);
 void ff_h263_encode_mba(MpegEncContext *s);
 void ff_init_qscale_tab(MpegEncContext *s);
 int ff_h263_pred_dc(MpegEncContext * s, int n, int16_t **dc_val_ptr);
-void ff_h263_pred_acdc(MpegEncContext * s, DCTELEM *block, int n);
+void ff_h263_pred_acdc(MpegEncContext * s, int16_t *block, int n);
 
 
 /**
@@ -99,7 +99,7 @@ void ff_h263_show_pict_info(MpegEncContext *s);
 
 int ff_intel_h263_decode_picture_header(MpegEncContext *s);
 int ff_h263_decode_mb(MpegEncContext *s,
-                      DCTELEM block[6][64]);
+                      int16_t block[6][64]);
 
 /**
  * Return the value of the 3bit "source format" syntax element.
@@ -144,7 +144,7 @@ static inline void ff_h263_encode_motion_vector(MpegEncContext * s, int x, int y
 }
 
 static inline int get_p_cbp(MpegEncContext * s,
-                      DCTELEM block[6][64],
+                      int16_t block[6][64],
                       int motion_x, int motion_y){
     int cbp, i;
 
diff --git a/libavcodec/h264.c b/libavcodec/h264.c
index 34d16baec8..89d2a6b540 100644
--- a/libavcodec/h264.c
+++ b/libavcodec/h264.c
@@ -1769,7 +1769,7 @@ static av_always_inline void xchg_mb_border(H264Context *h, uint8_t *src_y,
     }
 }
 
-static av_always_inline int dctcoef_get(DCTELEM *mb, int high_bit_depth,
+static av_always_inline int dctcoef_get(int16_t *mb, int high_bit_depth,
                                         int index)
 {
     if (high_bit_depth) {
@@ -1778,7 +1778,7 @@ static av_always_inline int dctcoef_get(DCTELEM *mb, int high_bit_depth,
         return AV_RN16A(mb + index);
 }
 
-static av_always_inline void dctcoef_set(DCTELEM *mb, int high_bit_depth,
+static av_always_inline void dctcoef_set(int16_t *mb, int high_bit_depth,
                                          int index, int value)
 {
     if (high_bit_depth) {
@@ -1797,8 +1797,8 @@ static av_always_inline void hl_decode_mb_predict_luma(H264Context *h,
                                                        uint8_t *dest_y, int p)
 {
     MpegEncContext *const s = &h->s;
-    void (*idct_add)(uint8_t *dst, DCTELEM *block, int stride);
-    void (*idct_dc_add)(uint8_t *dst, DCTELEM *block, int stride);
+    void (*idct_add)(uint8_t *dst, int16_t *block, int stride);
+    void (*idct_dc_add)(uint8_t *dst, int16_t *block, int stride);
     int i;
     int qscale = p == 0 ? s->qscale : h->chroma_qp[p - 1];
     block_offset += 16 * p;
@@ -1914,7 +1914,7 @@ static av_always_inline void hl_decode_mb_idct_luma(H264Context *h, int mb_type,
                                                     uint8_t *dest_y, int p)
 {
     MpegEncContext *const s = &h->s;
-    void (*idct_add)(uint8_t *dst, DCTELEM *block, int stride);
+    void (*idct_add)(uint8_t *dst, int16_t *block, int stride);
     int i;
     block_offset += 16 * p;
     if (!IS_INTRA4x4(mb_type)) {
diff --git a/libavcodec/h264.h b/libavcodec/h264.h
index e7ad07db8e..6989178af2 100644
--- a/libavcodec/h264.h
+++ b/libavcodec/h264.h
@@ -29,7 +29,6 @@
 #define AVCODEC_H264_H
 
 #include "libavutil/intreadwrite.h"
-#include "dsputil.h"
 #include "cabac.h"
 #include "mpegvideo.h"
 #include "h264dsp.h"
@@ -390,9 +389,9 @@ typedef struct H264Context {
     GetBitContext *intra_gb_ptr;
     GetBitContext *inter_gb_ptr;
 
-    DECLARE_ALIGNED(16, DCTELEM, mb)[16 * 48 * 2]; ///< as a dct coeffecient is int32_t in high depth, we need to reserve twice the space.
-    DECLARE_ALIGNED(16, DCTELEM, mb_luma_dc)[3][16 * 2];
-    DCTELEM mb_padding[256 * 2];        ///< as mb is addressed by scantable[i] and scantable is uint8_t we can either check that i is not too large or ensure that there is some unused stuff after mb
+    DECLARE_ALIGNED(16, int16_t, mb)[16 * 48 * 2]; ///< as a dct coeffecient is int32_t in high depth, we need to reserve twice the space.
+    DECLARE_ALIGNED(16, int16_t, mb_luma_dc)[3][16 * 2];
+    int16_t mb_padding[256 * 2];        ///< as mb is addressed by scantable[i] and scantable is uint8_t we can either check that i is not too large or ensure that there is some unused stuff after mb
 
     /**
      * Cabac
diff --git a/libavcodec/h264_cabac.c b/libavcodec/h264_cabac.c
index 76a648143d..bbca26ab4d 100644
--- a/libavcodec/h264_cabac.c
+++ b/libavcodec/h264_cabac.c
@@ -1560,7 +1560,7 @@ static av_always_inline int get_cabac_cbf_ctx( H264Context *h, int cat, int idx,
 }
 
 static av_always_inline void
-decode_cabac_residual_internal(H264Context *h, DCTELEM *block,
+decode_cabac_residual_internal(H264Context *h, int16_t *block,
                                int cat, int n, const uint8_t *scantable,
                                const uint32_t *qmul, int max_coeff,
                                int is_dc, int chroma422)
@@ -1744,18 +1744,27 @@ decode_cabac_residual_internal(H264Context *h, DCTELEM *block,
 
 }
 
-static void decode_cabac_residual_dc_internal( H264Context *h, DCTELEM *block, int cat, int n, const uint8_t *scantable, int max_coeff ) {
+static void decode_cabac_residual_dc_internal(H264Context *h, int16_t *block,
+                                              int cat, int n,
+                                              const uint8_t *scantable,
+                                              int max_coeff)
+{
     decode_cabac_residual_internal(h, block, cat, n, scantable, NULL, max_coeff, 1, 0);
 }
 
-static void decode_cabac_residual_dc_internal_422(H264Context *h, DCTELEM *block,
+static void decode_cabac_residual_dc_internal_422(H264Context *h, int16_t *block,
                                                   int cat, int n, const uint8_t *scantable,
                                                   int max_coeff)
 {
     decode_cabac_residual_internal(h, block, cat, n, scantable, NULL, max_coeff, 1, 1);
 }
 
-static void decode_cabac_residual_nondc_internal( H264Context *h, DCTELEM *block, int cat, int n, const uint8_t *scantable, const uint32_t *qmul, int max_coeff ) {
+static void decode_cabac_residual_nondc_internal(H264Context *h, int16_t *block,
+                                                 int cat, int n,
+                                                 const uint8_t *scantable,
+                                                 const uint32_t *qmul,
+                                                 int max_coeff)
+{
     decode_cabac_residual_internal(h, block, cat, n, scantable, qmul, max_coeff, 0, 0);
 }
 
@@ -1771,7 +1780,12 @@ static void decode_cabac_residual_nondc_internal( H264Context *h, DCTELEM *block
  * because it allows improved constant propagation into get_cabac_cbf_ctx,
  * as well as because most blocks have zero CBFs. */
 
-static av_always_inline void decode_cabac_residual_dc( H264Context *h, DCTELEM *block, int cat, int n, const uint8_t *scantable, int max_coeff ) {
+static av_always_inline void decode_cabac_residual_dc(H264Context *h,
+                                                      int16_t *block,
+                                                      int cat, int n,
+                                                      const uint8_t *scantable,
+                                                      int max_coeff)
+{
     /* read coded block flag */
     if( get_cabac( &h->cabac, &h->cabac_state[get_cabac_cbf_ctx( h, cat, n, max_coeff, 1 ) ] ) == 0 ) {
         h->non_zero_count_cache[scan8[n]] = 0;
@@ -1781,7 +1795,7 @@ static av_always_inline void decode_cabac_residual_dc( H264Context *h, DCTELEM *
 }
 
 static av_always_inline void
-decode_cabac_residual_dc_422(H264Context *h, DCTELEM *block,
+decode_cabac_residual_dc_422(H264Context *h, int16_t *block,
                              int cat, int n, const uint8_t *scantable,
                              int max_coeff)
 {
@@ -1793,7 +1807,13 @@ decode_cabac_residual_dc_422(H264Context *h, DCTELEM *block,
     decode_cabac_residual_dc_internal_422(h, block, cat, n, scantable, max_coeff);
 }
 
-static av_always_inline void decode_cabac_residual_nondc( H264Context *h, DCTELEM *block, int cat, int n, const uint8_t *scantable, const uint32_t *qmul, int max_coeff ) {
+static av_always_inline void decode_cabac_residual_nondc(H264Context *h,
+                                                         int16_t *block,
+                                                         int cat, int n,
+                                                         const uint8_t *scantable,
+                                                         const uint32_t *qmul,
+                                                         int max_coeff)
+{
     /* read coded block flag */
     if( (cat != 5 || CHROMA444) && get_cabac( &h->cabac, &h->cabac_state[get_cabac_cbf_ctx( h, cat, n, max_coeff, 0 ) ] ) == 0 ) {
         if( max_coeff == 64 ) {
@@ -2361,7 +2381,7 @@ decode_intra_mb:
             if( cbp&0x20 ) {
                 int c, i, i8x8;
                 for( c = 0; c < 2; c++ ) {
-                    DCTELEM *mb = h->mb + (16*(16 + 16*c) << pixel_shift);
+                    int16_t *mb = h->mb + (16*(16 + 16*c) << pixel_shift);
                     qmul = h->dequant4_coeff[c+1+(IS_INTRA( mb_type ) ? 0:3)][h->chroma_qp[c]];
                     for (i8x8 = 0; i8x8 < 2; i8x8++) {
                         for (i = 0; i < 4; i++) {
diff --git a/libavcodec/h264_cavlc.c b/libavcodec/h264_cavlc.c
index cd1130a4bc..7ca2bb2a1f 100644
--- a/libavcodec/h264_cavlc.c
+++ b/libavcodec/h264_cavlc.c
@@ -442,7 +442,7 @@ static inline int get_level_prefix(GetBitContext *gb){
  * @param max_coeff number of coefficients in the block
  * @return <0 if an error occurred
  */
-static int decode_residual(H264Context *h, GetBitContext *gb, DCTELEM *block, int n, const uint8_t *scantable, const uint32_t *qmul, int max_coeff){
+static int decode_residual(H264Context *h, GetBitContext *gb, int16_t *block, int n, const uint8_t *scantable, const uint32_t *qmul, int max_coeff){
     MpegEncContext * const s = &h->s;
     static const int coeff_token_table_index[17]= {0, 0, 1, 1, 2, 2, 2, 2, 3, 3, 3, 3, 3, 3, 3, 3, 3};
     int level[16];
@@ -662,7 +662,7 @@ static av_always_inline int decode_luma_residual(H264Context *h, GetBitContext *
         for(i8x8=0; i8x8<4; i8x8++){
             if(cbp & (1<<i8x8)){
                 if(IS_8x8DCT(mb_type)){
-                    DCTELEM *buf = &h->mb[64*i8x8+256*p << pixel_shift];
+                    int16_t *buf = &h->mb[64*i8x8+256*p << pixel_shift];
                     uint8_t *nnz;
                     for(i4x4=0; i4x4<4; i4x4++){
                         const int index= i4x4 + 4*i8x8 + p*16;
@@ -1143,7 +1143,7 @@ decode_intra_mb:
             if(cbp&0x20){
                 for(chroma_idx=0; chroma_idx<2; chroma_idx++){
                     const uint32_t *qmul = h->dequant4_coeff[chroma_idx+1+(IS_INTRA( mb_type ) ? 0:3)][h->chroma_qp[chroma_idx]];
-                    DCTELEM *mb = h->mb + (16*(16 + 16*chroma_idx) << pixel_shift);
+                    int16_t *mb = h->mb + (16*(16 + 16*chroma_idx) << pixel_shift);
                     for (i8x8=0; i8x8<num_c8x8; i8x8++) {
                         for (i4x4=0; i4x4<4; i4x4++) {
                             const int index= 16 + 16*chroma_idx + 8*i8x8 + i4x4;
diff --git a/libavcodec/h264_mb_template.c b/libavcodec/h264_mb_template.c
index a11009412c..34e63dcc32 100644
--- a/libavcodec/h264_mb_template.c
+++ b/libavcodec/h264_mb_template.c
@@ -52,7 +52,7 @@ static av_noinline void FUNC(hl_decode_mb)(H264Context *h)
     const int transform_bypass = !SIMPLE && (s->qscale == 0 && h->sps.transform_bypass);
     /* is_h264 should always be true if SVQ3 is disabled. */
     const int is_h264 = !CONFIG_SVQ3_DECODER || SIMPLE || s->codec_id == AV_CODEC_ID_H264;
-    void (*idct_add)(uint8_t *dst, DCTELEM *block, int stride);
+    void (*idct_add)(uint8_t *dst, int16_t *block, int stride);
     const int block_h   = 16 >> s->chroma_y_shift;
     const int chroma422 = CHROMA422;
 
diff --git a/libavcodec/h264dsp.h b/libavcodec/h264dsp.h
index 45f81a09c8..f0eaacc79e 100644
--- a/libavcodec/h264dsp.h
+++ b/libavcodec/h264dsp.h
@@ -29,8 +29,6 @@
 
 #include <stdint.h>
 
-#include "dsputil.h"
-
 typedef void (*h264_weight_func)(uint8_t *block, int stride, int height,
                                  int log2_denom, int weight, int offset);
 typedef void (*h264_biweight_func)(uint8_t *dst, uint8_t *src,
@@ -80,29 +78,29 @@ typedef struct H264DSPContext {
 
     /* IDCT */
     void (*h264_idct_add)(uint8_t *dst /*align 4*/,
-                          DCTELEM *block /*align 16*/, int stride);
+                          int16_t *block /*align 16*/, int stride);
     void (*h264_idct8_add)(uint8_t *dst /*align 8*/,
-                           DCTELEM *block /*align 16*/, int stride);
+                           int16_t *block /*align 16*/, int stride);
     void (*h264_idct_dc_add)(uint8_t *dst /*align 4*/,
-                             DCTELEM *block /*align 16*/, int stride);
+                             int16_t *block /*align 16*/, int stride);
     void (*h264_idct8_dc_add)(uint8_t *dst /*align 8*/,
-                              DCTELEM *block /*align 16*/, int stride);
+                              int16_t *block /*align 16*/, int stride);
 
     void (*h264_idct_add16)(uint8_t *dst /*align 16*/, const int *blockoffset,
-                            DCTELEM *block /*align 16*/, int stride,
+                            int16_t *block /*align 16*/, int stride,
                             const uint8_t nnzc[15 * 8]);
     void (*h264_idct8_add4)(uint8_t *dst /*align 16*/, const int *blockoffset,
-                            DCTELEM *block /*align 16*/, int stride,
+                            int16_t *block /*align 16*/, int stride,
                             const uint8_t nnzc[15 * 8]);
     void (*h264_idct_add8)(uint8_t **dst /*align 16*/, const int *blockoffset,
-                           DCTELEM *block /*align 16*/, int stride,
+                           int16_t *block /*align 16*/, int stride,
                            const uint8_t nnzc[15 * 8]);
     void (*h264_idct_add16intra)(uint8_t *dst /*align 16*/, const int *blockoffset,
-                                 DCTELEM *block /*align 16*/,
+                                 int16_t *block /*align 16*/,
                                  int stride, const uint8_t nnzc[15 * 8]);
-    void (*h264_luma_dc_dequant_idct)(DCTELEM *output,
-                                      DCTELEM *input /*align 16*/, int qmul);
-    void (*h264_chroma_dc_dequant_idct)(DCTELEM *block, int qmul);
+    void (*h264_luma_dc_dequant_idct)(int16_t *output,
+                                      int16_t *input /*align 16*/, int qmul);
+    void (*h264_chroma_dc_dequant_idct)(int16_t *block, int qmul);
 } H264DSPContext;
 
 void ff_h264dsp_init(H264DSPContext *c, const int bit_depth,
diff --git a/libavcodec/h264idct_template.c b/libavcodec/h264idct_template.c
index 313732b466..4a029a081c 100644
--- a/libavcodec/h264idct_template.c
+++ b/libavcodec/h264idct_template.c
@@ -47,7 +47,7 @@ static const uint8_t scan8[16*3]={
 };
 #endif
 
-void FUNCC(ff_h264_idct_add)(uint8_t *_dst, DCTELEM *_block, int stride)
+void FUNCC(ff_h264_idct_add)(uint8_t *_dst, int16_t *_block, int stride)
 {
     int i;
     pixel *dst = (pixel*)_dst;
@@ -81,7 +81,7 @@ void FUNCC(ff_h264_idct_add)(uint8_t *_dst, DCTELEM *_block, int stride)
     }
 }
 
-void FUNCC(ff_h264_idct8_add)(uint8_t *_dst, DCTELEM *_block, int stride){
+void FUNCC(ff_h264_idct8_add)(uint8_t *_dst, int16_t *_block, int stride){
     int i;
     pixel *dst = (pixel*)_dst;
     dctcoef *block = (dctcoef*)_block;
@@ -154,10 +154,10 @@ void FUNCC(ff_h264_idct8_add)(uint8_t *_dst, DCTELEM *_block, int stride){
 }
 
 // assumes all AC coefs are 0
-void FUNCC(ff_h264_idct_dc_add)(uint8_t *p_dst, DCTELEM *block, int stride){
+void FUNCC(ff_h264_idct_dc_add)(uint8_t *_dst, int16_t *block, int stride){
     int i, j;
     int dc = (((dctcoef*)block)[0] + 32) >> 6;
-    pixel *dst = (pixel*)p_dst;
+    pixel *dst = (pixel*)_dst;
     stride >>= sizeof(pixel)-1;
     for( j = 0; j < 4; j++ )
     {
@@ -167,10 +167,10 @@ void FUNCC(ff_h264_idct_dc_add)(uint8_t *p_dst, DCTELEM *block, int stride){
     }
 }
 
-void FUNCC(ff_h264_idct8_dc_add)(uint8_t *p_dst, DCTELEM *block, int stride){
+void FUNCC(ff_h264_idct8_dc_add)(uint8_t *_dst, int16_t *block, int stride){
     int i, j;
     int dc = (((dctcoef*)block)[0] + 32) >> 6;
-    pixel *dst = (pixel*)p_dst;
+    pixel *dst = (pixel*)_dst;
     stride >>= sizeof(pixel)-1;
     for( j = 0; j < 8; j++ )
     {
@@ -180,7 +180,7 @@ void FUNCC(ff_h264_idct8_dc_add)(uint8_t *p_dst, DCTELEM *block, int stride){
     }
 }
 
-void FUNCC(ff_h264_idct_add16)(uint8_t *dst, const int *block_offset, DCTELEM *block, int stride, const uint8_t nnzc[15*8]){
+void FUNCC(ff_h264_idct_add16)(uint8_t *dst, const int *block_offset, int16_t *block, int stride, const uint8_t nnzc[15*8]){
     int i;
     for(i=0; i<16; i++){
         int nnz = nnzc[ scan8[i] ];
@@ -191,7 +191,7 @@ void FUNCC(ff_h264_idct_add16)(uint8_t *dst, const int *block_offset, DCTELEM *b
     }
 }
 
-void FUNCC(ff_h264_idct_add16intra)(uint8_t *dst, const int *block_offset, DCTELEM *block, int stride, const uint8_t nnzc[15*8]){
+void FUNCC(ff_h264_idct_add16intra)(uint8_t *dst, const int *block_offset, int16_t *block, int stride, const uint8_t nnzc[15*8]){
     int i;
     for(i=0; i<16; i++){
         if(nnzc[ scan8[i] ])             FUNCC(ff_h264_idct_add   )(dst + block_offset[i], block + i*16*sizeof(pixel), stride);
@@ -199,7 +199,7 @@ void FUNCC(ff_h264_idct_add16intra)(uint8_t *dst, const int *block_offset, DCTEL
     }
 }
 
-void FUNCC(ff_h264_idct8_add4)(uint8_t *dst, const int *block_offset, DCTELEM *block, int stride, const uint8_t nnzc[15*8]){
+void FUNCC(ff_h264_idct8_add4)(uint8_t *dst, const int *block_offset, int16_t *block, int stride, const uint8_t nnzc[15*8]){
     int i;
     for(i=0; i<16; i+=4){
         int nnz = nnzc[ scan8[i] ];
@@ -210,7 +210,7 @@ void FUNCC(ff_h264_idct8_add4)(uint8_t *dst, const int *block_offset, DCTELEM *b
     }
 }
 
-void FUNCC(ff_h264_idct_add8)(uint8_t **dest, const int *block_offset, DCTELEM *block, int stride, const uint8_t nnzc[15*8]){
+void FUNCC(ff_h264_idct_add8)(uint8_t **dest, const int *block_offset, int16_t *block, int stride, const uint8_t nnzc[15*8]){
     int i, j;
     for(j=1; j<3; j++){
         for(i=j*16; i<j*16+4; i++){
@@ -222,7 +222,7 @@ void FUNCC(ff_h264_idct_add8)(uint8_t **dest, const int *block_offset, DCTELEM *
     }
 }
 
-void FUNCC(ff_h264_idct_add8_422)(uint8_t **dest, const int *block_offset, DCTELEM *block, int stride, const uint8_t nnzc[15*8]){
+void FUNCC(ff_h264_idct_add8_422)(uint8_t **dest, const int *block_offset, int16_t *block, int stride, const uint8_t nnzc[15*8]){
     int i, j;
 
     for(j=1; j<3; j++){
@@ -248,13 +248,13 @@ void FUNCC(ff_h264_idct_add8_422)(uint8_t **dest, const int *block_offset, DCTEL
  * IDCT transforms the 16 dc values and dequantizes them.
  * @param qmul quantization parameter
  */
-void FUNCC(ff_h264_luma_dc_dequant_idct)(DCTELEM *p_output, DCTELEM *p_input, int qmul){
+void FUNCC(ff_h264_luma_dc_dequant_idct)(int16_t *_output, int16_t *_input, int qmul){
 #define stride 16
     int i;
     int temp[16];
     static const uint8_t x_offset[4]={0, 2*stride, 8*stride, 10*stride};
-    dctcoef *input = (dctcoef*)p_input;
-    dctcoef *output = (dctcoef*)p_output;
+    dctcoef *input = (dctcoef*)_input;
+    dctcoef *output = (dctcoef*)_output;
 
     for(i=0; i<4; i++){
         const int z0= input[4*i+0] + input[4*i+1];
@@ -283,7 +283,7 @@ void FUNCC(ff_h264_luma_dc_dequant_idct)(DCTELEM *p_output, DCTELEM *p_input, in
 #undef stride
 }
 
-void FUNCC(ff_h264_chroma422_dc_dequant_idct)(DCTELEM *_block, int qmul){
+void FUNCC(ff_h264_chroma422_dc_dequant_idct)(int16_t *_block, int qmul){
     const int stride= 16*2;
     const int xStride= 16;
     int i;
@@ -310,7 +310,7 @@ void FUNCC(ff_h264_chroma422_dc_dequant_idct)(DCTELEM *_block, int qmul){
     }
 }
 
-void FUNCC(ff_h264_chroma_dc_dequant_idct)(DCTELEM *_block, int qmul){
+void FUNCC(ff_h264_chroma_dc_dequant_idct)(int16_t *_block, int qmul){
     const int stride= 16*2;
     const int xStride= 16;
     int a,b,c,d,e;
diff --git a/libavcodec/h264pred.h b/libavcodec/h264pred.h
index 33f39447df..36b542b7be 100644
--- a/libavcodec/h264pred.h
+++ b/libavcodec/h264pred.h
@@ -28,8 +28,8 @@
 #ifndef AVCODEC_H264PRED_H
 #define AVCODEC_H264PRED_H
 
-#include "libavutil/common.h"
-#include "dsputil.h"
+#include <stddef.h>
+#include <stdint.h>
 
 /**
  * Prediction types
@@ -98,15 +98,15 @@ typedef struct H264PredContext {
     void(*pred16x16[4 + 3 + 2])(uint8_t *src, ptrdiff_t stride);
 
     void(*pred4x4_add[2])(uint8_t *pix /*align  4*/,
-                          const DCTELEM *block /*align 16*/, ptrdiff_t stride);
+                          const int16_t *block /*align 16*/, ptrdiff_t stride);
     void(*pred8x8l_add[2])(uint8_t *pix /*align  8*/,
-                           const DCTELEM *block /*align 16*/, ptrdiff_t stride);
+                           const int16_t *block /*align 16*/, ptrdiff_t stride);
     void(*pred8x8_add[3])(uint8_t *pix /*align  8*/,
                           const int *block_offset,
-                          const DCTELEM *block /*align 16*/, ptrdiff_t stride);
+                          const int16_t *block /*align 16*/, ptrdiff_t stride);
     void(*pred16x16_add[3])(uint8_t *pix /*align 16*/,
                             const int *block_offset,
-                            const DCTELEM *block /*align 16*/, ptrdiff_t stride);
+                            const int16_t *block /*align 16*/, ptrdiff_t stride);
 } H264PredContext;
 
 void ff_h264_pred_init(H264PredContext *h, int codec_id,
diff --git a/libavcodec/h264pred_template.c b/libavcodec/h264pred_template.c
index f08fdf4ba4..e78f2d423c 100644
--- a/libavcodec/h264pred_template.c
+++ b/libavcodec/h264pred_template.c
@@ -1132,7 +1132,7 @@ static void FUNCC(pred8x8l_horizontal_up)(uint8_t *_src, int has_topleft,
 #undef PL
 #undef SRC
 
-static void FUNCC(pred4x4_vertical_add)(uint8_t *_pix, const DCTELEM *_block,
+static void FUNCC(pred4x4_vertical_add)(uint8_t *_pix, const int16_t *_block,
                                         ptrdiff_t stride)
 {
     int i;
@@ -1151,7 +1151,7 @@ static void FUNCC(pred4x4_vertical_add)(uint8_t *_pix, const DCTELEM *_block,
     }
 }
 
-static void FUNCC(pred4x4_horizontal_add)(uint8_t *_pix, const DCTELEM *_block,
+static void FUNCC(pred4x4_horizontal_add)(uint8_t *_pix, const int16_t *_block,
                                           ptrdiff_t stride)
 {
     int i;
@@ -1169,7 +1169,7 @@ static void FUNCC(pred4x4_horizontal_add)(uint8_t *_pix, const DCTELEM *_block,
     }
 }
 
-static void FUNCC(pred8x8l_vertical_add)(uint8_t *_pix, const DCTELEM *_block,
+static void FUNCC(pred8x8l_vertical_add)(uint8_t *_pix, const int16_t *_block,
                                          ptrdiff_t stride)
 {
     int i;
@@ -1192,7 +1192,7 @@ static void FUNCC(pred8x8l_vertical_add)(uint8_t *_pix, const DCTELEM *_block,
     }
 }
 
-static void FUNCC(pred8x8l_horizontal_add)(uint8_t *_pix, const DCTELEM *_block,
+static void FUNCC(pred8x8l_horizontal_add)(uint8_t *_pix, const int16_t *_block,
                                            ptrdiff_t stride)
 {
     int i;
@@ -1215,7 +1215,7 @@ static void FUNCC(pred8x8l_horizontal_add)(uint8_t *_pix, const DCTELEM *_block,
 }
 
 static void FUNCC(pred16x16_vertical_add)(uint8_t *pix, const int *block_offset,
-                                          const DCTELEM *block,
+                                          const int16_t *block,
                                           ptrdiff_t stride)
 {
     int i;
@@ -1225,7 +1225,7 @@ static void FUNCC(pred16x16_vertical_add)(uint8_t *pix, const int *block_offset,
 
 static void FUNCC(pred16x16_horizontal_add)(uint8_t *pix,
                                             const int *block_offset,
-                                            const DCTELEM *block,
+                                            const int16_t *block,
                                             ptrdiff_t stride)
 {
     int i;
@@ -1234,7 +1234,7 @@ static void FUNCC(pred16x16_horizontal_add)(uint8_t *pix,
 }
 
 static void FUNCC(pred8x8_vertical_add)(uint8_t *pix, const int *block_offset,
-                                        const DCTELEM *block, ptrdiff_t stride)
+                                        const int16_t *block, ptrdiff_t stride)
 {
     int i;
     for(i=0; i<4; i++)
@@ -1242,7 +1242,7 @@ static void FUNCC(pred8x8_vertical_add)(uint8_t *pix, const int *block_offset,
 }
 
 static void FUNCC(pred8x16_vertical_add)(uint8_t *pix, const int *block_offset,
-                                         const DCTELEM *block, ptrdiff_t stride)
+                                         const int16_t *block, ptrdiff_t stride)
 {
     int i;
     for(i=0; i<4; i++)
@@ -1252,7 +1252,7 @@ static void FUNCC(pred8x16_vertical_add)(uint8_t *pix, const int *block_offset,
 }
 
 static void FUNCC(pred8x8_horizontal_add)(uint8_t *pix, const int *block_offset,
-                                          const DCTELEM *block,
+                                          const int16_t *block,
                                           ptrdiff_t stride)
 {
     int i;
@@ -1262,7 +1262,7 @@ static void FUNCC(pred8x8_horizontal_add)(uint8_t *pix, const int *block_offset,
 
 static void FUNCC(pred8x16_horizontal_add)(uint8_t *pix,
                                            const int *block_offset,
-                                           const DCTELEM *block, ptrdiff_t stride)
+                                           const int16_t *block, ptrdiff_t stride)
 {
     int i;
     for(i=0; i<4; i++)
diff --git a/libavcodec/ituh263dec.c b/libavcodec/ituh263dec.c
index 44b13aa5ed..ec9a4c4257 100644
--- a/libavcodec/ituh263dec.c
+++ b/libavcodec/ituh263dec.c
@@ -439,7 +439,7 @@ static void h263_decode_dquant(MpegEncContext *s){
     ff_set_qscale(s, s->qscale);
 }
 
-static int h263_decode_block(MpegEncContext * s, DCTELEM * block,
+static int h263_decode_block(MpegEncContext * s, int16_t * block,
                              int n, int coded)
 {
     int code, level, i, j, last, run;
@@ -564,7 +564,7 @@ not_coded:
 
 static int h263_skip_b_part(MpegEncContext *s, int cbp)
 {
-    LOCAL_ALIGNED_16(DCTELEM, dblock, [64]);
+    LOCAL_ALIGNED_16(int16_t, dblock, [64]);
     int i, mbi;
     int bli[6];
 
@@ -603,7 +603,7 @@ static int h263_get_modb(GetBitContext *gb, int pb_frame, int *cbpb)
 }
 
 int ff_h263_decode_mb(MpegEncContext *s,
-                      DCTELEM block[6][64])
+                      int16_t block[6][64])
 {
     int cbpc, cbpy, i, cbp, pred_x, pred_y, mx, my, dquant;
     int16_t *mot_val;
diff --git a/libavcodec/ituh263enc.c b/libavcodec/ituh263enc.c
index 3eb198f2ce..2c0f6dbc74 100644
--- a/libavcodec/ituh263enc.c
+++ b/libavcodec/ituh263enc.c
@@ -303,7 +303,7 @@ static const int dquant_code[5]= {1,0,9,2,3};
  * @param block the 8x8 block
  * @param n block index (0-3 are luma, 4-5 are chroma)
  */
-static void h263_encode_block(MpegEncContext * s, DCTELEM * block, int n)
+static void h263_encode_block(MpegEncContext * s, int16_t * block, int n)
 {
     int level, run, last, i, j, last_index, last_non_zero, sign, slevel, code;
     RLTable *rl;
@@ -452,7 +452,7 @@ static void h263p_encode_umotion(MpegEncContext * s, int val)
 }
 
 void ff_h263_encode_mb(MpegEncContext * s,
-                       DCTELEM block[6][64],
+                       int16_t block[6][64],
                        int motion_x, int motion_y)
 {
     int cbpc, cbpy, i, cbp, pred_x, pred_y;
diff --git a/libavcodec/jfdctfst.c b/libavcodec/jfdctfst.c
index 3e30e5d535..b2e41f107e 100644
--- a/libavcodec/jfdctfst.c
+++ b/libavcodec/jfdctfst.c
@@ -136,17 +136,17 @@
 #endif
 
 
-/* Multiply a DCTELEM variable by an int32_t constant, and immediately
- * descale to yield a DCTELEM result.
+/* Multiply a int16_t variable by an int32_t constant, and immediately
+ * descale to yield a int16_t result.
  */
 
-#define MULTIPLY(var,const)  ((DCTELEM) DESCALE((var) * (const), CONST_BITS))
+#define MULTIPLY(var,const)  ((int16_t) DESCALE((var) * (const), CONST_BITS))
 
-static av_always_inline void row_fdct(DCTELEM * data){
+static av_always_inline void row_fdct(int16_t * data){
   int tmp0, tmp1, tmp2, tmp3, tmp4, tmp5, tmp6, tmp7;
   int tmp10, tmp11, tmp12, tmp13;
   int z1, z2, z3, z4, z5, z11, z13;
-  DCTELEM *dataptr;
+  int16_t *dataptr;
   int ctr;
 
   /* Pass 1: process rows. */
@@ -205,12 +205,12 @@ static av_always_inline void row_fdct(DCTELEM * data){
  */
 
 GLOBAL(void)
-ff_fdct_ifast (DCTELEM * data)
+ff_fdct_ifast (int16_t * data)
 {
   int tmp0, tmp1, tmp2, tmp3, tmp4, tmp5, tmp6, tmp7;
   int tmp10, tmp11, tmp12, tmp13;
   int z1, z2, z3, z4, z5, z11, z13;
-  DCTELEM *dataptr;
+  int16_t *dataptr;
   int ctr;
 
   row_fdct(data);
@@ -271,12 +271,12 @@ ff_fdct_ifast (DCTELEM * data)
  */
 
 GLOBAL(void)
-ff_fdct_ifast248 (DCTELEM * data)
+ff_fdct_ifast248 (int16_t * data)
 {
   int tmp0, tmp1, tmp2, tmp3, tmp4, tmp5, tmp6, tmp7;
   int tmp10, tmp11, tmp12, tmp13;
   int z1;
-  DCTELEM *dataptr;
+  int16_t *dataptr;
   int ctr;
 
   row_fdct(data);
diff --git a/libavcodec/jfdctint_template.c b/libavcodec/jfdctint_template.c
index 5175390710..d2fe093b2b 100644
--- a/libavcodec/jfdctint_template.c
+++ b/libavcodec/jfdctint_template.c
@@ -184,12 +184,12 @@
 #endif
 
 
-static av_always_inline void FUNC(row_fdct)(DCTELEM *data)
+static av_always_inline void FUNC(row_fdct)(int16_t *data)
 {
   int tmp0, tmp1, tmp2, tmp3, tmp4, tmp5, tmp6, tmp7;
   int tmp10, tmp11, tmp12, tmp13;
   int z1, z2, z3, z4, z5;
-  DCTELEM *dataptr;
+  int16_t *dataptr;
   int ctr;
 
   /* Pass 1: process rows. */
@@ -216,13 +216,13 @@ static av_always_inline void FUNC(row_fdct)(DCTELEM *data)
     tmp11 = tmp1 + tmp2;
     tmp12 = tmp1 - tmp2;
 
-    dataptr[0] = (DCTELEM) ((tmp10 + tmp11) << PASS1_BITS);
-    dataptr[4] = (DCTELEM) ((tmp10 - tmp11) << PASS1_BITS);
+    dataptr[0] = (int16_t) ((tmp10 + tmp11) << PASS1_BITS);
+    dataptr[4] = (int16_t) ((tmp10 - tmp11) << PASS1_BITS);
 
     z1 = MULTIPLY(tmp12 + tmp13, FIX_0_541196100);
-    dataptr[2] = (DCTELEM) DESCALE(z1 + MULTIPLY(tmp13, FIX_0_765366865),
+    dataptr[2] = (int16_t) DESCALE(z1 + MULTIPLY(tmp13, FIX_0_765366865),
                                    CONST_BITS-PASS1_BITS);
-    dataptr[6] = (DCTELEM) DESCALE(z1 + MULTIPLY(tmp12, - FIX_1_847759065),
+    dataptr[6] = (int16_t) DESCALE(z1 + MULTIPLY(tmp12, - FIX_1_847759065),
                                    CONST_BITS-PASS1_BITS);
 
     /* Odd part per figure 8 --- note paper omits factor of sqrt(2).
@@ -248,10 +248,10 @@ static av_always_inline void FUNC(row_fdct)(DCTELEM *data)
     z3 += z5;
     z4 += z5;
 
-    dataptr[7] = (DCTELEM) DESCALE(tmp4 + z1 + z3, CONST_BITS-PASS1_BITS);
-    dataptr[5] = (DCTELEM) DESCALE(tmp5 + z2 + z4, CONST_BITS-PASS1_BITS);
-    dataptr[3] = (DCTELEM) DESCALE(tmp6 + z2 + z3, CONST_BITS-PASS1_BITS);
-    dataptr[1] = (DCTELEM) DESCALE(tmp7 + z1 + z4, CONST_BITS-PASS1_BITS);
+    dataptr[7] = (int16_t) DESCALE(tmp4 + z1 + z3, CONST_BITS-PASS1_BITS);
+    dataptr[5] = (int16_t) DESCALE(tmp5 + z2 + z4, CONST_BITS-PASS1_BITS);
+    dataptr[3] = (int16_t) DESCALE(tmp6 + z2 + z3, CONST_BITS-PASS1_BITS);
+    dataptr[1] = (int16_t) DESCALE(tmp7 + z1 + z4, CONST_BITS-PASS1_BITS);
 
     dataptr += DCTSIZE;         /* advance pointer to next row */
   }
@@ -262,12 +262,12 @@ static av_always_inline void FUNC(row_fdct)(DCTELEM *data)
  */
 
 GLOBAL(void)
-FUNC(ff_jpeg_fdct_islow)(DCTELEM *data)
+FUNC(ff_jpeg_fdct_islow)(int16_t *data)
 {
   int tmp0, tmp1, tmp2, tmp3, tmp4, tmp5, tmp6, tmp7;
   int tmp10, tmp11, tmp12, tmp13;
   int z1, z2, z3, z4, z5;
-  DCTELEM *dataptr;
+  int16_t *dataptr;
   int ctr;
 
   FUNC(row_fdct)(data);
@@ -344,12 +344,12 @@ FUNC(ff_jpeg_fdct_islow)(DCTELEM *data)
  * you do even part two times.
  */
 GLOBAL(void)
-FUNC(ff_fdct248_islow)(DCTELEM *data)
+FUNC(ff_fdct248_islow)(int16_t *data)
 {
   int tmp0, tmp1, tmp2, tmp3, tmp4, tmp5, tmp6, tmp7;
   int tmp10, tmp11, tmp12, tmp13;
   int z1;
-  DCTELEM *dataptr;
+  int16_t *dataptr;
   int ctr;
 
   FUNC(row_fdct)(data);
diff --git a/libavcodec/jrevdct.c b/libavcodec/jrevdct.c
index 395eb8c638..897842cf69 100644
--- a/libavcodec/jrevdct.c
+++ b/libavcodec/jrevdct.c
@@ -74,7 +74,7 @@
 
 #define RIGHT_SHIFT(x, n) ((x) >> (n))
 
-typedef DCTELEM DCTBLOCK[DCTSIZE2];
+typedef int16_t DCTBLOCK[DCTSIZE2];
 
 #define CONST_BITS 13
 
@@ -213,7 +213,7 @@ void ff_j_rev_dct(DCTBLOCK data)
   int32_t tmp10, tmp11, tmp12, tmp13;
   int32_t z1, z2, z3, z4, z5;
   int32_t d0, d1, d2, d3, d4, d5, d6, d7;
-  register DCTELEM *dataptr;
+  register int16_t *dataptr;
   int rowctr;
 
   /* Pass 1: process rows. */
@@ -249,7 +249,7 @@ void ff_j_rev_dct(DCTBLOCK data)
       /* AC terms all zero */
       if (d0) {
           /* Compute a 32 bit value to assign. */
-          DCTELEM dcval = (DCTELEM) (d0 << PASS1_BITS);
+          int16_t dcval = (int16_t) (d0 << PASS1_BITS);
           register int v = (dcval & 0xffff) | ((dcval << 16) & 0xffff0000);
 
           idataptr[0] = v;
@@ -574,14 +574,14 @@ void ff_j_rev_dct(DCTBLOCK data)
 }
     /* Final output stage: inputs are tmp10..tmp13, tmp0..tmp3 */
 
-    dataptr[0] = (DCTELEM) DESCALE(tmp10 + tmp3, CONST_BITS-PASS1_BITS);
-    dataptr[7] = (DCTELEM) DESCALE(tmp10 - tmp3, CONST_BITS-PASS1_BITS);
-    dataptr[1] = (DCTELEM) DESCALE(tmp11 + tmp2, CONST_BITS-PASS1_BITS);
-    dataptr[6] = (DCTELEM) DESCALE(tmp11 - tmp2, CONST_BITS-PASS1_BITS);
-    dataptr[2] = (DCTELEM) DESCALE(tmp12 + tmp1, CONST_BITS-PASS1_BITS);
-    dataptr[5] = (DCTELEM) DESCALE(tmp12 - tmp1, CONST_BITS-PASS1_BITS);
-    dataptr[3] = (DCTELEM) DESCALE(tmp13 + tmp0, CONST_BITS-PASS1_BITS);
-    dataptr[4] = (DCTELEM) DESCALE(tmp13 - tmp0, CONST_BITS-PASS1_BITS);
+    dataptr[0] = (int16_t) DESCALE(tmp10 + tmp3, CONST_BITS-PASS1_BITS);
+    dataptr[7] = (int16_t) DESCALE(tmp10 - tmp3, CONST_BITS-PASS1_BITS);
+    dataptr[1] = (int16_t) DESCALE(tmp11 + tmp2, CONST_BITS-PASS1_BITS);
+    dataptr[6] = (int16_t) DESCALE(tmp11 - tmp2, CONST_BITS-PASS1_BITS);
+    dataptr[2] = (int16_t) DESCALE(tmp12 + tmp1, CONST_BITS-PASS1_BITS);
+    dataptr[5] = (int16_t) DESCALE(tmp12 - tmp1, CONST_BITS-PASS1_BITS);
+    dataptr[3] = (int16_t) DESCALE(tmp13 + tmp0, CONST_BITS-PASS1_BITS);
+    dataptr[4] = (int16_t) DESCALE(tmp13 - tmp0, CONST_BITS-PASS1_BITS);
 
     dataptr += DCTSIZE;         /* advance pointer to next row */
   }
@@ -920,21 +920,21 @@ void ff_j_rev_dct(DCTBLOCK data)
 
     /* Final output stage: inputs are tmp10..tmp13, tmp0..tmp3 */
 
-    dataptr[DCTSIZE*0] = (DCTELEM) DESCALE(tmp10 + tmp3,
+    dataptr[DCTSIZE*0] = (int16_t) DESCALE(tmp10 + tmp3,
                                            CONST_BITS+PASS1_BITS+3);
-    dataptr[DCTSIZE*7] = (DCTELEM) DESCALE(tmp10 - tmp3,
+    dataptr[DCTSIZE*7] = (int16_t) DESCALE(tmp10 - tmp3,
                                            CONST_BITS+PASS1_BITS+3);
-    dataptr[DCTSIZE*1] = (DCTELEM) DESCALE(tmp11 + tmp2,
+    dataptr[DCTSIZE*1] = (int16_t) DESCALE(tmp11 + tmp2,
                                            CONST_BITS+PASS1_BITS+3);
-    dataptr[DCTSIZE*6] = (DCTELEM) DESCALE(tmp11 - tmp2,
+    dataptr[DCTSIZE*6] = (int16_t) DESCALE(tmp11 - tmp2,
                                            CONST_BITS+PASS1_BITS+3);
-    dataptr[DCTSIZE*2] = (DCTELEM) DESCALE(tmp12 + tmp1,
+    dataptr[DCTSIZE*2] = (int16_t) DESCALE(tmp12 + tmp1,
                                            CONST_BITS+PASS1_BITS+3);
-    dataptr[DCTSIZE*5] = (DCTELEM) DESCALE(tmp12 - tmp1,
+    dataptr[DCTSIZE*5] = (int16_t) DESCALE(tmp12 - tmp1,
                                            CONST_BITS+PASS1_BITS+3);
-    dataptr[DCTSIZE*3] = (DCTELEM) DESCALE(tmp13 + tmp0,
+    dataptr[DCTSIZE*3] = (int16_t) DESCALE(tmp13 + tmp0,
                                            CONST_BITS+PASS1_BITS+3);
-    dataptr[DCTSIZE*4] = (DCTELEM) DESCALE(tmp13 - tmp0,
+    dataptr[DCTSIZE*4] = (int16_t) DESCALE(tmp13 - tmp0,
                                            CONST_BITS+PASS1_BITS+3);
 
     dataptr++;                  /* advance pointer to next column */
@@ -951,7 +951,7 @@ void ff_j_rev_dct4(DCTBLOCK data)
   int32_t tmp10, tmp11, tmp12, tmp13;
   int32_t z1;
   int32_t d0, d2, d4, d6;
-  register DCTELEM *dataptr;
+  register int16_t *dataptr;
   int rowctr;
 
   /* Pass 1: process rows. */
@@ -983,7 +983,7 @@ void ff_j_rev_dct4(DCTBLOCK data)
       /* AC terms all zero */
       if (d0) {
           /* Compute a 32 bit value to assign. */
-          DCTELEM dcval = (DCTELEM) (d0 << PASS1_BITS);
+          int16_t dcval = (int16_t) (d0 << PASS1_BITS);
           register int v = (dcval & 0xffff) | ((dcval << 16) & 0xffff0000);
 
           idataptr[0] = v;
@@ -1045,10 +1045,10 @@ void ff_j_rev_dct4(DCTBLOCK data)
 
     /* Final output stage: inputs are tmp10..tmp13, tmp0..tmp3 */
 
-    dataptr[0] = (DCTELEM) DESCALE(tmp10, CONST_BITS-PASS1_BITS);
-    dataptr[1] = (DCTELEM) DESCALE(tmp11, CONST_BITS-PASS1_BITS);
-    dataptr[2] = (DCTELEM) DESCALE(tmp12, CONST_BITS-PASS1_BITS);
-    dataptr[3] = (DCTELEM) DESCALE(tmp13, CONST_BITS-PASS1_BITS);
+    dataptr[0] = (int16_t) DESCALE(tmp10, CONST_BITS-PASS1_BITS);
+    dataptr[1] = (int16_t) DESCALE(tmp11, CONST_BITS-PASS1_BITS);
+    dataptr[2] = (int16_t) DESCALE(tmp12, CONST_BITS-PASS1_BITS);
+    dataptr[3] = (int16_t) DESCALE(tmp13, CONST_BITS-PASS1_BITS);
 
     dataptr += DCTSTRIDE;       /* advance pointer to next row */
   }
diff --git a/libavcodec/mdec.c b/libavcodec/mdec.c
index 364e999513..68e850c60e 100644
--- a/libavcodec/mdec.c
+++ b/libavcodec/mdec.c
@@ -45,14 +45,14 @@ typedef struct MDECContext{
     int mb_width;
     int mb_height;
     int mb_x, mb_y;
-    DECLARE_ALIGNED(16, DCTELEM, block)[6][64];
+    DECLARE_ALIGNED(16, int16_t, block)[6][64];
     uint8_t *bitstream_buffer;
     unsigned int bitstream_buffer_size;
     int block_last_index[6];
 } MDECContext;
 
 //very similar to MPEG-1
-static inline int mdec_decode_block_intra(MDECContext *a, DCTELEM *block, int n)
+static inline int mdec_decode_block_intra(MDECContext *a, int16_t *block, int n)
 {
     int level, diff, i, j, run;
     int component;
@@ -119,7 +119,7 @@ static inline int mdec_decode_block_intra(MDECContext *a, DCTELEM *block, int n)
     return 0;
 }
 
-static inline int decode_mb(MDECContext *a, DCTELEM block[6][64]){
+static inline int decode_mb(MDECContext *a, int16_t block[6][64]){
     int i;
     const int block_index[6]= {5,4,0,1,2,3};
 
@@ -134,7 +134,7 @@ static inline int decode_mb(MDECContext *a, DCTELEM block[6][64]){
 }
 
 static inline void idct_put(MDECContext *a, int mb_x, int mb_y){
-    DCTELEM (*block)[64]= a->block;
+    int16_t (*block)[64]= a->block;
     int linesize= a->picture.linesize[0];
 
     uint8_t *dest_y  = a->picture.data[0] + (mb_y * 16* linesize              ) + mb_x * 16;
diff --git a/libavcodec/mimic.c b/libavcodec/mimic.c
index 880bbc0ddd..f0579fd031 100644
--- a/libavcodec/mimic.c
+++ b/libavcodec/mimic.c
@@ -47,7 +47,7 @@ typedef struct {
     AVFrame         buf_ptrs    [16];
     AVPicture       flipped_ptrs[16];
 
-    DECLARE_ALIGNED(16, DCTELEM, dct_block)[64];
+    DECLARE_ALIGNED(16, int16_t, dct_block)[64];
 
     GetBitContext   gb;
     ScanTable       scantable;
@@ -183,7 +183,7 @@ static const int8_t vlcdec_lookup[9][64] = {
 
 static int vlc_decode_block(MimicContext *ctx, int num_coeffs, int qscale)
 {
-    DCTELEM *block = ctx->dct_block;
+    int16_t *block = ctx->dct_block;
     unsigned int pos;
 
     ctx->dsp.clear_block(block);
diff --git a/libavcodec/mjpegdec.c b/libavcodec/mjpegdec.c
index 78a0453d57..6551b52630 100644
--- a/libavcodec/mjpegdec.c
+++ b/libavcodec/mjpegdec.c
@@ -498,7 +498,7 @@ static inline int mjpeg_decode_dc(MJpegDecodeContext *s, int dc_index)
 }
 
 /* decode block and dequantize */
-static int decode_block(MJpegDecodeContext *s, DCTELEM *block, int component,
+static int decode_block(MJpegDecodeContext *s, int16_t *block, int component,
                         int dc_index, int ac_index, int16_t *quant_matrix)
 {
     int code, i, j, level, val;
@@ -546,7 +546,7 @@ static int decode_block(MJpegDecodeContext *s, DCTELEM *block, int component,
     return 0;
 }
 
-static int decode_dc_progressive(MJpegDecodeContext *s, DCTELEM *block,
+static int decode_dc_progressive(MJpegDecodeContext *s, int16_t *block,
                                  int component, int dc_index,
                                  int16_t *quant_matrix, int Al)
 {
@@ -564,7 +564,7 @@ static int decode_dc_progressive(MJpegDecodeContext *s, DCTELEM *block,
 }
 
 /* decode block and dequantize - progressive JPEG version */
-static int decode_block_progressive(MJpegDecodeContext *s, DCTELEM *block,
+static int decode_block_progressive(MJpegDecodeContext *s, int16_t *block,
                                     uint8_t *last_nnz, int ac_index,
                                     int16_t *quant_matrix,
                                     int ss, int se, int Al, int *EOBRUN)
@@ -662,7 +662,7 @@ for (; ; i++) {                                                     \
 }
 
 /* decode block and dequantize - progressive JPEG refinement pass */
-static int decode_block_refinement(MJpegDecodeContext *s, DCTELEM *block,
+static int decode_block_refinement(MJpegDecodeContext *s, int16_t *block,
                                    uint8_t *last_nnz,
                                    int ac_index, int16_t *quant_matrix,
                                    int ss, int se, int Al, int *EOBRUN)
@@ -1081,7 +1081,7 @@ static int mjpeg_decode_scan(MJpegDecodeContext *s, int nb_components, int Ah,
                     } else {
                         int block_idx  = s->block_stride[c] * (v * mb_y + y) +
                                          (h * mb_x + x);
-                        DCTELEM *block = s->blocks[c][block_idx];
+                        int16_t *block = s->blocks[c][block_idx];
                         if (Ah)
                             block[0] += get_bits1(&s->gb) *
                                         s->quant_matrixes[s->quant_index[c]][0] << Al;
@@ -1139,7 +1139,7 @@ static int mjpeg_decode_scan_progressive_ac(MJpegDecodeContext *s, int ss,
     for (mb_y = 0; mb_y < s->mb_height; mb_y++) {
         uint8_t *ptr     = data + (mb_y * linesize * 8 >> s->avctx->lowres);
         int block_idx    = mb_y * s->block_stride[c];
-        DCTELEM (*block)[64] = &s->blocks[c][block_idx];
+        int16_t (*block)[64] = &s->blocks[c][block_idx];
         uint8_t *last_nnz    = &s->last_nnz[c][block_idx];
         for (mb_x = 0; mb_x < s->mb_width; mb_x++, block++, last_nnz++) {
                 int ret;
diff --git a/libavcodec/mjpegdec.h b/libavcodec/mjpegdec.h
index 6c6195b3d3..01f4323bc7 100644
--- a/libavcodec/mjpegdec.h
+++ b/libavcodec/mjpegdec.h
@@ -91,8 +91,8 @@ typedef struct MJpegDecodeContext {
     int got_picture;                                ///< we found a SOF and picture is valid, too.
     int linesize[MAX_COMPONENTS];                   ///< linesize << interlaced
     int8_t *qscale_table;
-    DECLARE_ALIGNED(16, DCTELEM, block)[64];
-    DCTELEM (*blocks[MAX_COMPONENTS])[64]; ///< intermediate sums (progressive mode)
+    DECLARE_ALIGNED(16, int16_t, block)[64];
+    int16_t (*blocks[MAX_COMPONENTS])[64]; ///< intermediate sums (progressive mode)
     uint8_t *last_nnz[MAX_COMPONENTS];
     uint64_t coefs_finished[MAX_COMPONENTS]; ///< bitmask of which coefs have been completely decoded (progressive mode)
     ScanTable scantable;
diff --git a/libavcodec/mjpegenc.c b/libavcodec/mjpegenc.c
index 7ac304295c..2a92dbdf73 100644
--- a/libavcodec/mjpegenc.c
+++ b/libavcodec/mjpegenc.c
@@ -397,7 +397,7 @@ void ff_mjpeg_encode_dc(MpegEncContext *s, int val,
     }
 }
 
-static void encode_block(MpegEncContext *s, DCTELEM *block, int n)
+static void encode_block(MpegEncContext *s, int16_t *block, int n)
 {
     int mant, nbits, code, i, j;
     int component, dc, run, last_index, val;
@@ -455,7 +455,7 @@ static void encode_block(MpegEncContext *s, DCTELEM *block, int n)
         put_bits(&s->pb, huff_size_ac[0], huff_code_ac[0]);
 }
 
-void ff_mjpeg_encode_mb(MpegEncContext *s, DCTELEM block[6][64])
+void ff_mjpeg_encode_mb(MpegEncContext *s, int16_t block[6][64])
 {
     int i;
     if (s->chroma_format == CHROMA_444) {
diff --git a/libavcodec/mjpegenc.h b/libavcodec/mjpegenc.h
index 06f77c6a15..ce0c1cce15 100644
--- a/libavcodec/mjpegenc.h
+++ b/libavcodec/mjpegenc.h
@@ -33,7 +33,8 @@
 #ifndef AVCODEC_MJPEGENC_H
 #define AVCODEC_MJPEGENC_H
 
-#include "dsputil.h"
+#include <stdint.h>
+
 #include "mpegvideo.h"
 
 typedef struct MJpegContext {
@@ -55,6 +56,6 @@ void ff_mjpeg_encode_picture_trailer(MpegEncContext *s);
 void ff_mjpeg_encode_stuffing(MpegEncContext *s);
 void ff_mjpeg_encode_dc(MpegEncContext *s, int val,
                         uint8_t *huff_size, uint16_t *huff_code);
-void ff_mjpeg_encode_mb(MpegEncContext *s, DCTELEM block[6][64]);
+void ff_mjpeg_encode_mb(MpegEncContext *s, int16_t block[6][64]);
 
 #endif /* AVCODEC_MJPEGENC_H */
diff --git a/libavcodec/mpeg12.c b/libavcodec/mpeg12.c
index d8e22a26af..c385853edc 100644
--- a/libavcodec/mpeg12.c
+++ b/libavcodec/mpeg12.c
@@ -81,7 +81,7 @@ static int mpeg_decode_motion(MpegEncContext *s, int fcode, int pred)
     return sign_extend(val, 5 + shift);
 }
 
-static inline int mpeg1_decode_block_intra(MpegEncContext *s, DCTELEM *block, int n)
+static inline int mpeg1_decode_block_intra(MpegEncContext *s, int16_t *block, int n)
 {
     int level, dc, diff, i, j, run;
     int component;
@@ -152,12 +152,12 @@ static inline int mpeg1_decode_block_intra(MpegEncContext *s, DCTELEM *block, in
    return 0;
 }
 
-int ff_mpeg1_decode_block_intra(MpegEncContext *s, DCTELEM *block, int n)
+int ff_mpeg1_decode_block_intra(MpegEncContext *s, int16_t *block, int n)
 {
     return mpeg1_decode_block_intra(s, block, n);
 }
 
-static inline int mpeg1_decode_block_inter(MpegEncContext *s, DCTELEM *block, int n)
+static inline int mpeg1_decode_block_inter(MpegEncContext *s, int16_t *block, int n)
 {
     int level, i, j, run;
     RLTable *rl = &ff_rl_mpeg1;
@@ -237,7 +237,7 @@ end:
  * Changing this would eat up any speed benefits it has.
  * Do not use "fast" flag if you need the code to be robust.
  */
-static inline int mpeg1_fast_decode_block_inter(MpegEncContext *s, DCTELEM *block, int n)
+static inline int mpeg1_fast_decode_block_inter(MpegEncContext *s, int16_t *block, int n)
 {
     int level, i, j, run;
     RLTable *rl = &ff_rl_mpeg1;
@@ -309,7 +309,7 @@ end:
 }
 
 
-static inline int mpeg2_decode_block_non_intra(MpegEncContext *s, DCTELEM *block, int n)
+static inline int mpeg2_decode_block_non_intra(MpegEncContext *s, int16_t *block, int n)
 {
     int level, i, j, run;
     RLTable *rl = &ff_rl_mpeg1;
@@ -394,7 +394,7 @@ end:
  * Do not use "fast" flag if you need the code to be robust.
  */
 static inline int mpeg2_fast_decode_block_non_intra(MpegEncContext *s,
-                                                    DCTELEM *block, int n)
+                                                    int16_t *block, int n)
 {
     int level, i, j, run;
     RLTable *rl = &ff_rl_mpeg1;
@@ -455,7 +455,7 @@ end:
 }
 
 
-static inline int mpeg2_decode_block_intra(MpegEncContext *s, DCTELEM *block, int n)
+static inline int mpeg2_decode_block_intra(MpegEncContext *s, int16_t *block, int n)
 {
     int level, dc, diff, i, j, run;
     int component;
@@ -538,7 +538,7 @@ static inline int mpeg2_decode_block_intra(MpegEncContext *s, DCTELEM *block, in
  * Changing this would eat up any speed benefits it has.
  * Do not use "fast" flag if you need the code to be robust.
  */
-static inline int mpeg2_fast_decode_block_intra(MpegEncContext *s, DCTELEM *block, int n)
+static inline int mpeg2_fast_decode_block_intra(MpegEncContext *s, int16_t *block, int n)
 {
     int level, dc, diff, j, run;
     int component;
@@ -737,7 +737,7 @@ static inline int get_qscale(MpegEncContext *s)
 
 static void exchange_uv(MpegEncContext *s)
 {
-    DCTELEM (*tmp)[64];
+    int16_t (*tmp)[64];
 
     tmp           = s->pblocks[4];
     s->pblocks[4] = s->pblocks[5];
@@ -750,7 +750,7 @@ static void exchange_uv(MpegEncContext *s)
 #define MT_16X8  2
 #define MT_DMV   3
 
-static int mpeg_decode_mb(MpegEncContext *s, DCTELEM block[12][64])
+static int mpeg_decode_mb(MpegEncContext *s, int16_t block[12][64])
 {
     int i, j, k, cbp, val, mb_type, motion_type;
     const int mb_block_count = 4 + (1 << s->chroma_format);
diff --git a/libavcodec/mpeg12.h b/libavcodec/mpeg12.h
index 6e60c8695f..1f369176a5 100644
--- a/libavcodec/mpeg12.h
+++ b/libavcodec/mpeg12.h
@@ -71,6 +71,6 @@ static inline int decode_dc(GetBitContext *gb, int component)
     return diff;
 }
 
-extern int ff_mpeg1_decode_block_intra(MpegEncContext *s, DCTELEM *block, int n);
+extern int ff_mpeg1_decode_block_intra(MpegEncContext *s, int16_t *block, int n);
 
 #endif /* AVCODEC_MPEG12_H */
diff --git a/libavcodec/mpeg12enc.c b/libavcodec/mpeg12enc.c
index f2b5c05f0d..4a54ed58ec 100644
--- a/libavcodec/mpeg12enc.c
+++ b/libavcodec/mpeg12enc.c
@@ -52,7 +52,7 @@ static const uint8_t svcd_scan_offset_placeholder[14] = {
 };
 
 static void mpeg1_encode_block(MpegEncContext *s,
-                         DCTELEM *block,
+                         int16_t *block,
                          int component);
 static void mpeg1_encode_motion(MpegEncContext *s, int val, int f_or_b_code);    // RAL: f_code parameter added
 
@@ -471,7 +471,7 @@ static inline void put_mb_modes(MpegEncContext *s, int n, int bits,
 }
 
 static av_always_inline void mpeg1_encode_mb_internal(MpegEncContext *s,
-                                                   DCTELEM block[6][64],
+                                                   int16_t block[6][64],
                                                    int motion_x, int motion_y,
                                                    int mb_block_count)
 {
@@ -695,7 +695,7 @@ static av_always_inline void mpeg1_encode_mb_internal(MpegEncContext *s,
     }
 }
 
-void ff_mpeg1_encode_mb(MpegEncContext *s, DCTELEM block[6][64], int motion_x, int motion_y)
+void ff_mpeg1_encode_mb(MpegEncContext *s, int16_t block[6][64], int motion_x, int motion_y)
 {
     if (s->chroma_format == CHROMA_420) mpeg1_encode_mb_internal(s, block, motion_x, motion_y, 6);
     else                                mpeg1_encode_mb_internal(s, block, motion_x, motion_y, 8);
@@ -878,7 +878,7 @@ static inline void encode_dc(MpegEncContext *s, int diff, int component)
 }
 
 static void mpeg1_encode_block(MpegEncContext *s,
-                               DCTELEM *block,
+                               int16_t *block,
                                int n)
 {
     int alevel, level, last_non_zero, dc, diff, i, j, run, last_index, sign;
diff --git a/libavcodec/mpeg4video.h b/libavcodec/mpeg4video.h
index 97c39b6ed6..400ce4d184 100644
--- a/libavcodec/mpeg4video.h
+++ b/libavcodec/mpeg4video.h
@@ -83,9 +83,9 @@ extern const uint16_t ff_mpeg4_resync_prefix[8];
 extern const uint8_t ff_mpeg4_dc_threshold[8];
 
 void ff_mpeg4_encode_mb(MpegEncContext *s,
-                        DCTELEM block[6][64],
+                        int16_t block[6][64],
                         int motion_x, int motion_y);
-void ff_mpeg4_pred_ac(MpegEncContext * s, DCTELEM *block, int n,
+void ff_mpeg4_pred_ac(MpegEncContext * s, int16_t *block, int n,
                       int dir);
 void ff_set_mpeg4_time(MpegEncContext * s);
 void ff_mpeg4_encode_picture_header(MpegEncContext *s, int picture_number);
diff --git a/libavcodec/mpeg4videodec.c b/libavcodec/mpeg4videodec.c
index acbc525a38..62106d9e30 100644
--- a/libavcodec/mpeg4videodec.c
+++ b/libavcodec/mpeg4videodec.c
@@ -53,7 +53,7 @@ static const int mb_type_b_map[4]= {
  * @param n block index (0-3 are luma, 4-5 are chroma)
  * @param dir the ac prediction direction
  */
-void ff_mpeg4_pred_ac(MpegEncContext * s, DCTELEM *block, int n,
+void ff_mpeg4_pred_ac(MpegEncContext * s, int16_t *block, int n,
                       int dir)
 {
     int i;
@@ -843,7 +843,7 @@ int ff_mpeg4_decode_partitions(MpegEncContext *s)
  * Decode a block.
  * @return <0 if an error occurred
  */
-static inline int mpeg4_decode_block(MpegEncContext * s, DCTELEM * block,
+static inline int mpeg4_decode_block(MpegEncContext * s, int16_t * block,
                               int n, int coded, int intra, int rvlc)
 {
     int level, i, last, run;
@@ -1089,7 +1089,7 @@ static inline int mpeg4_decode_block(MpegEncContext * s, DCTELEM * block,
  * decode partition C of one MB.
  * @return <0 if an error occurred
  */
-static int mpeg4_decode_partitioned_mb(MpegEncContext *s, DCTELEM block[6][64])
+static int mpeg4_decode_partitioned_mb(MpegEncContext *s, int16_t block[6][64])
 {
     int cbp, mb_type;
     const int xy= s->mb_x + s->mb_y*s->mb_stride;
@@ -1172,7 +1172,7 @@ static int mpeg4_decode_partitioned_mb(MpegEncContext *s, DCTELEM block[6][64])
 }
 
 static int mpeg4_decode_mb(MpegEncContext *s,
-                      DCTELEM block[6][64])
+                      int16_t block[6][64])
 {
     int cbpc, cbpy, i, cbp, pred_x, pred_y, mx, my, dquant;
     int16_t *mot_val;
diff --git a/libavcodec/mpeg4videoenc.c b/libavcodec/mpeg4videoenc.c
index 33115061fe..0cc4df9a07 100644
--- a/libavcodec/mpeg4videoenc.c
+++ b/libavcodec/mpeg4videoenc.c
@@ -59,7 +59,7 @@ max run: 29/41
  * Return the number of bits that encoding the 8x8 block in block would need.
  * @param[in]  block_last_index last index in scantable order that refers to a non zero element in block.
  */
-static inline int get_block_rate(MpegEncContext * s, DCTELEM block[64], int block_last_index, uint8_t scantable[64]){
+static inline int get_block_rate(MpegEncContext * s, int16_t block[64], int block_last_index, uint8_t scantable[64]){
     int last=0;
     int j;
     int rate=0;
@@ -91,7 +91,7 @@ static inline int get_block_rate(MpegEncContext * s, DCTELEM block[64], int bloc
  * @param[out] st scantable for each 8x8 block
  * @param[in] zigzag_last_index index referring to the last non zero coefficient in zigzag order
  */
-static inline void restore_ac_coeffs(MpegEncContext * s, DCTELEM block[6][64], const int dir[6], uint8_t *st[6], const int zigzag_last_index[6])
+static inline void restore_ac_coeffs(MpegEncContext * s, int16_t block[6][64], const int dir[6], uint8_t *st[6], const int zigzag_last_index[6])
 {
     int i, n;
     memcpy(s->block_last_index, zigzag_last_index, sizeof(int)*6);
@@ -122,7 +122,7 @@ static inline void restore_ac_coeffs(MpegEncContext * s, DCTELEM block[6][64], c
  * @param[out] st scantable for each 8x8 block
  * @param[out] zigzag_last_index index referring to the last non zero coefficient in zigzag order
  */
-static inline int decide_ac_pred(MpegEncContext * s, DCTELEM block[6][64], const int dir[6], uint8_t *st[6], int zigzag_last_index[6])
+static inline int decide_ac_pred(MpegEncContext * s, int16_t block[6][64], const int dir[6], uint8_t *st[6], int zigzag_last_index[6])
 {
     int score= 0;
     int i, n;
@@ -294,7 +294,7 @@ static inline int mpeg4_get_dc_length(int level, int n){
  * Encode an 8x8 block.
  * @param n block index (0-3 are luma, 4-5 are chroma)
  */
-static inline void mpeg4_encode_block(MpegEncContext * s, DCTELEM * block, int n, int intra_dc,
+static inline void mpeg4_encode_block(MpegEncContext * s, int16_t * block, int n, int intra_dc,
                                uint8_t *scan_table, PutBitContext *dc_pb, PutBitContext *ac_pb)
 {
     int i, last_non_zero;
@@ -345,7 +345,7 @@ static inline void mpeg4_encode_block(MpegEncContext * s, DCTELEM * block, int n
     }
 }
 
-static int mpeg4_get_block_length(MpegEncContext * s, DCTELEM * block, int n, int intra_dc,
+static int mpeg4_get_block_length(MpegEncContext * s, int16_t * block, int n, int intra_dc,
                                uint8_t *scan_table)
 {
     int i, last_non_zero;
@@ -396,7 +396,7 @@ static int mpeg4_get_block_length(MpegEncContext * s, DCTELEM * block, int n, in
     return len;
 }
 
-static inline void mpeg4_encode_blocks(MpegEncContext * s, DCTELEM block[6][64], int intra_dc[6],
+static inline void mpeg4_encode_blocks(MpegEncContext * s, int16_t block[6][64], int intra_dc[6],
                                uint8_t **scan_table, PutBitContext *dc_pb, PutBitContext *ac_pb){
     int i;
 
@@ -425,7 +425,7 @@ static inline void mpeg4_encode_blocks(MpegEncContext * s, DCTELEM block[6][64],
     }
 }
 
-static inline int get_b_cbp(MpegEncContext * s, DCTELEM block[6][64],
+static inline int get_b_cbp(MpegEncContext * s, int16_t block[6][64],
                             int motion_x, int motion_y, int mb_type)
 {
     int cbp = 0, i;
@@ -469,7 +469,7 @@ static inline int get_b_cbp(MpegEncContext * s, DCTELEM block[6][64],
 static const int dquant_code[5]= {1,0,9,2,3};
 
 void ff_mpeg4_encode_mb(MpegEncContext * s,
-                        DCTELEM block[6][64],
+                        int16_t block[6][64],
                         int motion_x, int motion_y)
 {
     int cbpc, cbpy, pred_x, pred_y;
diff --git a/libavcodec/mpegvideo.c b/libavcodec/mpegvideo.c
index a21054b5ff..e6e7337082 100644
--- a/libavcodec/mpegvideo.c
+++ b/libavcodec/mpegvideo.c
@@ -43,19 +43,19 @@
 //#include <assert.h>
 
 static void dct_unquantize_mpeg1_intra_c(MpegEncContext *s,
-                                   DCTELEM *block, int n, int qscale);
+                                   int16_t *block, int n, int qscale);
 static void dct_unquantize_mpeg1_inter_c(MpegEncContext *s,
-                                   DCTELEM *block, int n, int qscale);
+                                   int16_t *block, int n, int qscale);
 static void dct_unquantize_mpeg2_intra_c(MpegEncContext *s,
-                                   DCTELEM *block, int n, int qscale);
+                                   int16_t *block, int n, int qscale);
 static void dct_unquantize_mpeg2_intra_bitexact(MpegEncContext *s,
-                                   DCTELEM *block, int n, int qscale);
+                                   int16_t *block, int n, int qscale);
 static void dct_unquantize_mpeg2_inter_c(MpegEncContext *s,
-                                   DCTELEM *block, int n, int qscale);
+                                   int16_t *block, int n, int qscale);
 static void dct_unquantize_h263_intra_c(MpegEncContext *s,
-                                  DCTELEM *block, int n, int qscale);
+                                  int16_t *block, int n, int qscale);
 static void dct_unquantize_h263_inter_c(MpegEncContext *s,
-                                  DCTELEM *block, int n, int qscale);
+                                  int16_t *block, int n, int qscale);
 
 
 //#define DEBUG
@@ -392,7 +392,7 @@ int ff_alloc_picture(MpegEncContext *s, Picture *pic, int shared)
         }
         if (s->avctx->debug&FF_DEBUG_DCT_COEFF) {
             FF_ALLOCZ_OR_GOTO(s->avctx, pic->f.dct_coeff,
-                              64 * mb_array_size * sizeof(DCTELEM) * 6, fail)
+                              64 * mb_array_size * sizeof(int16_t) * 6, fail)
         }
         pic->f.qstride = s->mb_stride;
         FF_ALLOCZ_OR_GOTO(s->avctx, pic->f.pan_scan,
@@ -469,7 +469,7 @@ static int init_duplicate_context(MpegEncContext *s, MpegEncContext *base)
                               2 * 64 * sizeof(int), fail)
         }
     }
-    FF_ALLOCZ_OR_GOTO(s->avctx, s->blocks, 64 * 12 * 2 * sizeof(DCTELEM), fail)
+    FF_ALLOCZ_OR_GOTO(s->avctx, s->blocks, 64 * 12 * 2 * sizeof(int16_t), fail)
     s->block = s->blocks[0];
 
     for (i = 0; i < 12; i++) {
@@ -2434,7 +2434,7 @@ unhandled:
 
 /* put block[] to dest[] */
 static inline void put_dct(MpegEncContext *s,
-                           DCTELEM *block, int i, uint8_t *dest, int line_size, int qscale)
+                           int16_t *block, int i, uint8_t *dest, int line_size, int qscale)
 {
     s->dct_unquantize_intra(s, block, i, qscale);
     s->dsp.idct_put (dest, line_size, block);
@@ -2442,7 +2442,7 @@ static inline void put_dct(MpegEncContext *s,
 
 /* add block[] to dest[] */
 static inline void add_dct(MpegEncContext *s,
-                           DCTELEM *block, int i, uint8_t *dest, int line_size)
+                           int16_t *block, int i, uint8_t *dest, int line_size)
 {
     if (s->block_last_index[i] >= 0) {
         s->dsp.idct_add (dest, line_size, block);
@@ -2450,7 +2450,7 @@ static inline void add_dct(MpegEncContext *s,
 }
 
 static inline void add_dequant_dct(MpegEncContext *s,
-                           DCTELEM *block, int i, uint8_t *dest, int line_size, int qscale)
+                           int16_t *block, int i, uint8_t *dest, int line_size, int qscale)
 {
     if (s->block_last_index[i] >= 0) {
         s->dct_unquantize_inter(s, block, i, qscale);
@@ -2503,7 +2503,7 @@ void ff_clean_intra_table_entries(MpegEncContext *s)
    s->interlaced_dct : true if interlaced dct used (mpeg2)
  */
 static av_always_inline
-void MPV_decode_mb_internal(MpegEncContext *s, DCTELEM block[12][64],
+void MPV_decode_mb_internal(MpegEncContext *s, int16_t block[12][64],
                             int lowres_flag, int is_mpeg12)
 {
     const int mb_xy = s->mb_y * s->mb_stride + s->mb_x;
@@ -2515,7 +2515,7 @@ void MPV_decode_mb_internal(MpegEncContext *s, DCTELEM block[12][64],
     if(s->avctx->debug&FF_DEBUG_DCT_COEFF) {
        /* save DCT coefficients */
        int i,j;
-       DCTELEM *dct = &s->current_picture.f.dct_coeff[mb_xy * 64 * 6];
+       int16_t *dct = &s->current_picture.f.dct_coeff[mb_xy * 64 * 6];
        av_log(s->avctx, AV_LOG_DEBUG, "DCT coeffs of MB at %dx%d:\n", s->mb_x, s->mb_y);
        for(i=0; i<6; i++){
            for(j=0; j<64; j++){
@@ -2746,7 +2746,7 @@ skip_idct:
     }
 }
 
-void ff_MPV_decode_mb(MpegEncContext *s, DCTELEM block[12][64]){
+void ff_MPV_decode_mb(MpegEncContext *s, int16_t block[12][64]){
 #if !CONFIG_SMALL
     if(s->out_format == FMT_MPEG1) {
         if(s->avctx->lowres) MPV_decode_mb_internal(s, block, 1, 1);
@@ -2888,7 +2888,7 @@ void ff_mpeg_flush(AVCodecContext *avctx){
 }
 
 static void dct_unquantize_mpeg1_intra_c(MpegEncContext *s,
-                                   DCTELEM *block, int n, int qscale)
+                                   int16_t *block, int n, int qscale)
 {
     int i, level, nCoeffs;
     const uint16_t *quant_matrix;
@@ -2917,7 +2917,7 @@ static void dct_unquantize_mpeg1_intra_c(MpegEncContext *s,
 }
 
 static void dct_unquantize_mpeg1_inter_c(MpegEncContext *s,
-                                   DCTELEM *block, int n, int qscale)
+                                   int16_t *block, int n, int qscale)
 {
     int i, level, nCoeffs;
     const uint16_t *quant_matrix;
@@ -2946,7 +2946,7 @@ static void dct_unquantize_mpeg1_inter_c(MpegEncContext *s,
 }
 
 static void dct_unquantize_mpeg2_intra_c(MpegEncContext *s,
-                                   DCTELEM *block, int n, int qscale)
+                                   int16_t *block, int n, int qscale)
 {
     int i, level, nCoeffs;
     const uint16_t *quant_matrix;
@@ -2973,7 +2973,7 @@ static void dct_unquantize_mpeg2_intra_c(MpegEncContext *s,
 }
 
 static void dct_unquantize_mpeg2_intra_bitexact(MpegEncContext *s,
-                                   DCTELEM *block, int n, int qscale)
+                                   int16_t *block, int n, int qscale)
 {
     int i, level, nCoeffs;
     const uint16_t *quant_matrix;
@@ -3004,7 +3004,7 @@ static void dct_unquantize_mpeg2_intra_bitexact(MpegEncContext *s,
 }
 
 static void dct_unquantize_mpeg2_inter_c(MpegEncContext *s,
-                                   DCTELEM *block, int n, int qscale)
+                                   int16_t *block, int n, int qscale)
 {
     int i, level, nCoeffs;
     const uint16_t *quant_matrix;
@@ -3035,7 +3035,7 @@ static void dct_unquantize_mpeg2_inter_c(MpegEncContext *s,
 }
 
 static void dct_unquantize_h263_intra_c(MpegEncContext *s,
-                                  DCTELEM *block, int n, int qscale)
+                                  int16_t *block, int n, int qscale)
 {
     int i, level, qmul, qadd;
     int nCoeffs;
@@ -3069,7 +3069,7 @@ static void dct_unquantize_h263_intra_c(MpegEncContext *s,
 }
 
 static void dct_unquantize_h263_inter_c(MpegEncContext *s,
-                                  DCTELEM *block, int n, int qscale)
+                                  int16_t *block, int n, int qscale)
 {
     int i, level, qmul, qadd;
     int nCoeffs;
diff --git a/libavcodec/mpegvideo.h b/libavcodec/mpegvideo.h
index a2320ad405..03f4f9ff1e 100644
--- a/libavcodec/mpegvideo.h
+++ b/libavcodec/mpegvideo.h
@@ -681,39 +681,39 @@ typedef struct MpegEncContext {
 
     uint8_t *ptr_lastgob;
     int swap_uv;             //vcr2 codec is an MPEG-2 variant with U and V swapped
-    DCTELEM (*pblocks[12])[64];
+    int16_t (*pblocks[12])[64];
 
-    DCTELEM (*block)[64]; ///< points to one of the following blocks
-    DCTELEM (*blocks)[12][64]; // for HQ mode we need to keep the best block
-    int (*decode_mb)(struct MpegEncContext *s, DCTELEM block[6][64]); // used by some codecs to avoid a switch()
+    int16_t (*block)[64]; ///< points to one of the following blocks
+    int16_t (*blocks)[12][64]; // for HQ mode we need to keep the best block
+    int (*decode_mb)(struct MpegEncContext *s, int16_t block[6][64]); // used by some codecs to avoid a switch()
 #define SLICE_OK         0
 #define SLICE_ERROR     -1
 #define SLICE_END       -2 ///<end marker found
 #define SLICE_NOEND     -3 ///<no end marker or error found but mb count exceeded
 
     void (*dct_unquantize_mpeg1_intra)(struct MpegEncContext *s,
-                           DCTELEM *block/*align 16*/, int n, int qscale);
+                           int16_t *block/*align 16*/, int n, int qscale);
     void (*dct_unquantize_mpeg1_inter)(struct MpegEncContext *s,
-                           DCTELEM *block/*align 16*/, int n, int qscale);
+                           int16_t *block/*align 16*/, int n, int qscale);
     void (*dct_unquantize_mpeg2_intra)(struct MpegEncContext *s,
-                           DCTELEM *block/*align 16*/, int n, int qscale);
+                           int16_t *block/*align 16*/, int n, int qscale);
     void (*dct_unquantize_mpeg2_inter)(struct MpegEncContext *s,
-                           DCTELEM *block/*align 16*/, int n, int qscale);
+                           int16_t *block/*align 16*/, int n, int qscale);
     void (*dct_unquantize_h263_intra)(struct MpegEncContext *s,
-                           DCTELEM *block/*align 16*/, int n, int qscale);
+                           int16_t *block/*align 16*/, int n, int qscale);
     void (*dct_unquantize_h263_inter)(struct MpegEncContext *s,
-                           DCTELEM *block/*align 16*/, int n, int qscale);
+                           int16_t *block/*align 16*/, int n, int qscale);
     void (*dct_unquantize_h261_intra)(struct MpegEncContext *s,
-                           DCTELEM *block/*align 16*/, int n, int qscale);
+                           int16_t *block/*align 16*/, int n, int qscale);
     void (*dct_unquantize_h261_inter)(struct MpegEncContext *s,
-                           DCTELEM *block/*align 16*/, int n, int qscale);
+                           int16_t *block/*align 16*/, int n, int qscale);
     void (*dct_unquantize_intra)(struct MpegEncContext *s, // unquantizer to use (mpeg4 can use both)
-                           DCTELEM *block/*align 16*/, int n, int qscale);
+                           int16_t *block/*align 16*/, int n, int qscale);
     void (*dct_unquantize_inter)(struct MpegEncContext *s, // unquantizer to use (mpeg4 can use both)
-                           DCTELEM *block/*align 16*/, int n, int qscale);
-    int (*dct_quantize)(struct MpegEncContext *s, DCTELEM *block/*align 16*/, int n, int qscale, int *overflow);
-    int (*fast_dct_quantize)(struct MpegEncContext *s, DCTELEM *block/*align 16*/, int n, int qscale, int *overflow);
-    void (*denoise_dct)(struct MpegEncContext *s, DCTELEM *block);
+                           int16_t *block/*align 16*/, int n, int qscale);
+    int (*dct_quantize)(struct MpegEncContext *s, int16_t *block/*align 16*/, int n, int qscale, int *overflow);
+    int (*fast_dct_quantize)(struct MpegEncContext *s, int16_t *block/*align 16*/, int n, int qscale, int *overflow);
+    void (*denoise_dct)(struct MpegEncContext *s, int16_t *block);
 
     int mpv_flags;      ///< flags set by private options
     int quantizer_noise_shaping;
@@ -776,7 +776,7 @@ int ff_MPV_common_init(MpegEncContext *s);
 int ff_mpv_frame_size_alloc(MpegEncContext *s, int linesize);
 int ff_MPV_common_frame_size_change(MpegEncContext *s);
 void ff_MPV_common_end(MpegEncContext *s);
-void ff_MPV_decode_mb(MpegEncContext *s, DCTELEM block[12][64]);
+void ff_MPV_decode_mb(MpegEncContext *s, int16_t block[12][64]);
 int ff_MPV_frame_start(MpegEncContext *s, AVCodecContext *avctx);
 void ff_MPV_frame_end(MpegEncContext *s);
 int ff_MPV_encode_init(AVCodecContext *avctx);
@@ -796,7 +796,7 @@ void ff_print_debug_info(MpegEncContext *s, AVFrame *pict);
 void ff_write_quant_matrix(PutBitContext *pb, uint16_t *matrix);
 void ff_release_unused_pictures(MpegEncContext *s, int remove_current);
 int ff_find_unused_picture(MpegEncContext *s, int shared);
-void ff_denoise_dct(MpegEncContext *s, DCTELEM *block);
+void ff_denoise_dct(MpegEncContext *s, int16_t *block);
 int ff_update_duplicate_context(MpegEncContext *dst, MpegEncContext *src);
 int ff_MPV_lowest_referenced_row(MpegEncContext *s, int dir);
 void ff_MPV_report_decode_progress(MpegEncContext *s);
@@ -812,7 +812,7 @@ int ff_dct_common_init(MpegEncContext *s);
 int ff_dct_encode_init(MpegEncContext *s);
 void ff_convert_matrix(DSPContext *dsp, int (*qmat)[64], uint16_t (*qmat16)[2][64],
                        const uint16_t *quant_matrix, int bias, int qmin, int qmax, int intra);
-int ff_dct_quantize_c(MpegEncContext *s, DCTELEM *block, int n, int qscale, int *overflow);
+int ff_dct_quantize_c(MpegEncContext *s, int16_t *block, int n, int qscale, int *overflow);
 
 void ff_init_block_index(MpegEncContext *s);
 void ff_copy_picture(Picture *dst, Picture *src);
@@ -887,7 +887,7 @@ extern const uint8_t * const ff_mpeg2_dc_scale_table[4];
 
 void ff_mpeg1_encode_picture_header(MpegEncContext *s, int picture_number);
 void ff_mpeg1_encode_mb(MpegEncContext *s,
-                        DCTELEM block[6][64],
+                        int16_t block[6][64],
                         int motion_x, int motion_y);
 void ff_mpeg1_encode_init(MpegEncContext *s);
 void ff_mpeg1_encode_slice_header(MpegEncContext *s);
@@ -902,7 +902,7 @@ extern const uint8_t ff_h263_loop_filter_strength[32];
 void ff_h261_loop_filter(MpegEncContext *s);
 void ff_h261_reorder_mb_index(MpegEncContext* s);
 void ff_h261_encode_mb(MpegEncContext *s,
-                    DCTELEM block[6][64],
+                    int16_t block[6][64],
                     int motion_x, int motion_y);
 void ff_h261_encode_picture_header(MpegEncContext * s, int picture_number);
 void ff_h261_encode_init(MpegEncContext *s);
@@ -919,7 +919,7 @@ void ff_rv20_encode_picture_header(MpegEncContext *s, int picture_number);
 void ff_msmpeg4_encode_picture_header(MpegEncContext * s, int picture_number);
 void ff_msmpeg4_encode_ext_header(MpegEncContext * s);
 void ff_msmpeg4_encode_mb(MpegEncContext * s,
-                          DCTELEM block[6][64],
+                          int16_t block[6][64],
                           int motion_x, int motion_y);
 int ff_msmpeg4_decode_picture_header(MpegEncContext * s);
 int ff_msmpeg4_decode_ext_header(MpegEncContext * s, int buf_size);
@@ -927,14 +927,14 @@ int ff_msmpeg4_decode_init(AVCodecContext *avctx);
 void ff_msmpeg4_encode_init(MpegEncContext *s);
 int ff_wmv2_decode_picture_header(MpegEncContext * s);
 int ff_wmv2_decode_secondary_picture_header(MpegEncContext * s);
-void ff_wmv2_add_mb(MpegEncContext *s, DCTELEM block[6][64], uint8_t *dest_y, uint8_t *dest_cb, uint8_t *dest_cr);
+void ff_wmv2_add_mb(MpegEncContext *s, int16_t block[6][64], uint8_t *dest_y, uint8_t *dest_cb, uint8_t *dest_cr);
 void ff_mspel_motion(MpegEncContext *s,
                                uint8_t *dest_y, uint8_t *dest_cb, uint8_t *dest_cr,
                                uint8_t **ref_picture, op_pixels_func (*pix_op)[4],
                                int motion_x, int motion_y, int h);
 int ff_wmv2_encode_picture_header(MpegEncContext * s, int picture_number);
 void ff_wmv2_encode_mb(MpegEncContext * s,
-                       DCTELEM block[6][64],
+                       int16_t block[6][64],
                        int motion_x, int motion_y);
 
 #endif /* AVCODEC_MPEGVIDEO_H */
diff --git a/libavcodec/mpegvideo_enc.c b/libavcodec/mpegvideo_enc.c
index 1e4c58d9ce..ef9e0f398d 100644
--- a/libavcodec/mpegvideo_enc.c
+++ b/libavcodec/mpegvideo_enc.c
@@ -52,10 +52,10 @@
 //#include <assert.h>
 
 static int encode_picture(MpegEncContext *s, int picture_number);
-static int dct_quantize_refine(MpegEncContext *s, DCTELEM *block, int16_t *weight, DCTELEM *orig, int n, int qscale);
+static int dct_quantize_refine(MpegEncContext *s, int16_t *block, int16_t *weight, int16_t *orig, int n, int qscale);
 static int sse_mb(MpegEncContext *s);
-static void denoise_dct_c(MpegEncContext *s, DCTELEM *block);
-static int dct_quantize_trellis_c(MpegEncContext *s, DCTELEM *block, int n, int qscale, int *overflow);
+static void denoise_dct_c(MpegEncContext *s, int16_t *block);
+static int dct_quantize_trellis_c(MpegEncContext *s, int16_t *block, int n, int qscale, int *overflow);
 
 //#define DEBUG
 
@@ -1727,7 +1727,7 @@ static inline void dct_single_coeff_elimination(MpegEncContext *s,
     int score = 0;
     int run = 0;
     int i;
-    DCTELEM *block = s->block[n];
+    int16_t *block = s->block[n];
     const int last_index = s->block_last_index[n];
     int skip_dc;
 
@@ -1767,7 +1767,7 @@ static inline void dct_single_coeff_elimination(MpegEncContext *s,
         s->block_last_index[n] = -1;
 }
 
-static inline void clip_coeffs(MpegEncContext *s, DCTELEM *block,
+static inline void clip_coeffs(MpegEncContext *s, int16_t *block,
                                int last_index)
 {
     int i;
@@ -1832,7 +1832,7 @@ static av_always_inline void encode_mb_internal(MpegEncContext *s,
                                                 int mb_block_count)
 {
     int16_t weight[12][64];
-    DCTELEM orig[12][64];
+    int16_t orig[12][64];
     const int mb_x = s->mb_x;
     const int mb_y = s->mb_y;
     int i;
@@ -2093,7 +2093,7 @@ static av_always_inline void encode_mb_internal(MpegEncContext *s,
                 get_visual_weight(weight[7], ptr_cr + uv_dct_offset,
                                   wrap_c);
         }
-        memcpy(orig[0], s->block[0], sizeof(DCTELEM) * 64 * mb_block_count);
+        memcpy(orig[0], s->block[0], sizeof(int16_t) * 64 * mb_block_count);
     }
 
     /* DCT & quantize */
@@ -2839,7 +2839,7 @@ static int encode_thread(AVCodecContext *c, void *arg){
                     if(best_s.mv_type==MV_TYPE_16X16){ //FIXME move 4mv after QPRD
                         const int last_qp= backup_s.qscale;
                         int qpi, qp, dc[6];
-                        DCTELEM ac[6][16];
+                        int16_t ac[6][16];
                         const int mvdir= (best_s.mv_dir&MV_DIR_BACKWARD) ? 1 : 0;
                         static const int dquant_tab[4]={-1,1,-2,2};
 
@@ -2864,7 +2864,7 @@ static int encode_thread(AVCodecContext *c, void *arg){
                             if(s->mb_intra && s->dc_val[0]){
                                 for(i=0; i<6; i++){
                                     dc[i]= s->dc_val[0][ s->block_index[i] ];
-                                    memcpy(ac[i], s->ac_val[0][s->block_index[i]], sizeof(DCTELEM)*16);
+                                    memcpy(ac[i], s->ac_val[0][s->block_index[i]], sizeof(int16_t)*16);
                                 }
                             }
 
@@ -2874,7 +2874,7 @@ static int encode_thread(AVCodecContext *c, void *arg){
                                 if(s->mb_intra && s->dc_val[0]){
                                     for(i=0; i<6; i++){
                                         s->dc_val[0][ s->block_index[i] ]= dc[i];
-                                        memcpy(s->ac_val[0][s->block_index[i]], ac[i], sizeof(DCTELEM)*16);
+                                        memcpy(s->ac_val[0][s->block_index[i]], ac[i], sizeof(int16_t)*16);
                                     }
                                 }
                             }
@@ -3479,7 +3479,7 @@ static int encode_picture(MpegEncContext *s, int picture_number)
     return 0;
 }
 
-static void denoise_dct_c(MpegEncContext *s, DCTELEM *block){
+static void denoise_dct_c(MpegEncContext *s, int16_t *block){
     const int intra= s->mb_intra;
     int i;
 
@@ -3504,7 +3504,7 @@ static void denoise_dct_c(MpegEncContext *s, DCTELEM *block){
 }
 
 static int dct_quantize_trellis_c(MpegEncContext *s,
-                                  DCTELEM *block, int n,
+                                  int16_t *block, int n,
                                   int qscale, int *overflow){
     const int *qmat;
     const uint8_t *scantable= s->intra_scantable.scantable;
@@ -3611,7 +3611,7 @@ static int dct_quantize_trellis_c(MpegEncContext *s,
     *overflow= s->max_qcoeff < max; //overflow might have happened
 
     if(last_non_zero < start_i){
-        memset(block + start_i, 0, (64-start_i)*sizeof(DCTELEM));
+        memset(block + start_i, 0, (64-start_i)*sizeof(int16_t));
         return last_non_zero;
     }
 
@@ -3743,7 +3743,7 @@ static int dct_quantize_trellis_c(MpegEncContext *s,
 
     dc= FFABS(block[0]);
     last_non_zero= last_i - 1;
-    memset(block + start_i, 0, (64-start_i)*sizeof(DCTELEM));
+    memset(block + start_i, 0, (64-start_i)*sizeof(int16_t));
 
     if(last_non_zero < start_i)
         return last_non_zero;
@@ -3818,10 +3818,10 @@ static void build_basis(uint8_t *perm){
 }
 
 static int dct_quantize_refine(MpegEncContext *s, //FIXME breaks denoise?
-                        DCTELEM *block, int16_t *weight, DCTELEM *orig,
+                        int16_t *block, int16_t *weight, int16_t *orig,
                         int n, int qscale){
     int16_t rem[64];
-    LOCAL_ALIGNED_16(DCTELEM, d1, [64]);
+    LOCAL_ALIGNED_16(int16_t, d1, [64]);
     const uint8_t *scantable= s->intra_scantable.scantable;
     const uint8_t *perm_scantable= s->intra_scantable.permutated;
 //    unsigned int threshold1, threshold2;
@@ -4191,7 +4191,7 @@ STOP_TIMER("iterative search")
 }
 
 int ff_dct_quantize_c(MpegEncContext *s,
-                        DCTELEM *block, int n,
+                        int16_t *block, int n,
                         int qscale, int *overflow)
 {
     int i, j, level, last_non_zero, q, start_i;
diff --git a/libavcodec/mpegvideo_xvmc.c b/libavcodec/mpegvideo_xvmc.c
index 6247e6240c..30a37e911a 100644
--- a/libavcodec/mpegvideo_xvmc.c
+++ b/libavcodec/mpegvideo_xvmc.c
@@ -44,7 +44,7 @@ void ff_xvmc_init_block(MpegEncContext *s)
     struct xvmc_pix_fmt *render = (struct xvmc_pix_fmt*)s->current_picture.f.data[2];
     assert(render && render->xvmc_id == AV_XVMC_ID);
 
-    s->block = (DCTELEM (*)[64])(render->data_blocks + render->next_free_data_block_num * 64);
+    s->block = (int16_t (*)[64])(render->data_blocks + render->next_free_data_block_num * 64);
 }
 
 /**
diff --git a/libavcodec/msmpeg4.c b/libavcodec/msmpeg4.c
index 00fac8c4b2..278c5fcfdc 100644
--- a/libavcodec/msmpeg4.c
+++ b/libavcodec/msmpeg4.c
@@ -404,7 +404,7 @@ static int msmpeg4v2_decode_motion(MpegEncContext * s, int pred, int f_code)
     return val;
 }
 
-static int msmpeg4v12_decode_mb(MpegEncContext *s, DCTELEM block[6][64])
+static int msmpeg4v12_decode_mb(MpegEncContext *s, int16_t block[6][64])
 {
     int cbp, code, i;
     uint32_t * const mb_type_ptr = &s->current_picture.f.mb_type[s->mb_x + s->mb_y*s->mb_stride];
@@ -494,7 +494,7 @@ static int msmpeg4v12_decode_mb(MpegEncContext *s, DCTELEM block[6][64])
     return 0;
 }
 
-static int msmpeg4v34_decode_mb(MpegEncContext *s, DCTELEM block[6][64])
+static int msmpeg4v34_decode_mb(MpegEncContext *s, int16_t block[6][64])
 {
     int cbp, code, i;
     uint8_t *coded_val;
@@ -938,7 +938,7 @@ static int msmpeg4_decode_dc(MpegEncContext * s, int n, int *dir_ptr)
 }
 
 //#define ERROR_DETAILS
-int ff_msmpeg4_decode_block(MpegEncContext * s, DCTELEM * block,
+int ff_msmpeg4_decode_block(MpegEncContext * s, int16_t * block,
                               int n, int coded, const uint8_t *scan_table)
 {
     int level, i, last, run, run_diff;
diff --git a/libavcodec/msmpeg4.h b/libavcodec/msmpeg4.h
index abc414cfdd..a723888192 100644
--- a/libavcodec/msmpeg4.h
+++ b/libavcodec/msmpeg4.h
@@ -26,7 +26,6 @@
 
 #include "config.h"
 #include "avcodec.h"
-#include "dsputil.h"
 #include "mpegvideo.h"
 #include "msmpeg4data.h"
 #include "put_bits.h"
@@ -45,17 +44,17 @@ extern VLC ff_inter_intra_vlc;
 
 void ff_msmpeg4_code012(PutBitContext *pb, int n);
 void ff_msmpeg4_common_init(MpegEncContext *s);
-void ff_msmpeg4_encode_block(MpegEncContext * s, DCTELEM * block, int n);
+void ff_msmpeg4_encode_block(MpegEncContext * s, int16_t * block, int n);
 void ff_msmpeg4_handle_slices(MpegEncContext *s);
 void ff_msmpeg4_encode_motion(MpegEncContext * s, int mx, int my);
 int ff_msmpeg4_coded_block_pred(MpegEncContext * s, int n,
                                 uint8_t **coded_block_ptr);
 int ff_msmpeg4_decode_motion(MpegEncContext * s, int *mx_ptr, int *my_ptr);
-int ff_msmpeg4_decode_block(MpegEncContext * s, DCTELEM * block,
+int ff_msmpeg4_decode_block(MpegEncContext * s, int16_t * block,
                             int n, int coded, const uint8_t *scan_table);
 int ff_msmpeg4_pred_dc(MpegEncContext *s, int n,
                        int16_t **dc_val_ptr, int *dir_ptr);
-int ff_wmv2_decode_mb(MpegEncContext *s, DCTELEM block[6][64]);
+int ff_wmv2_decode_mb(MpegEncContext *s, int16_t block[6][64]);
 
 #define CONFIG_MSMPEG4_DECODER (CONFIG_MSMPEG4V1_DECODER || \
                                 CONFIG_MSMPEG4V2_DECODER || \
diff --git a/libavcodec/msmpeg4enc.c b/libavcodec/msmpeg4enc.c
index 71b84df739..82e6646fba 100644
--- a/libavcodec/msmpeg4enc.c
+++ b/libavcodec/msmpeg4enc.c
@@ -369,7 +369,7 @@ static void msmpeg4v2_encode_motion(MpegEncContext * s, int val)
 }
 
 void ff_msmpeg4_encode_mb(MpegEncContext * s,
-                          DCTELEM block[6][64],
+                          int16_t block[6][64],
                           int motion_x, int motion_y)
 {
     int cbp, coded_cbp, i;
@@ -570,7 +570,7 @@ static void msmpeg4_encode_dc(MpegEncContext * s, int level, int n, int *dir_ptr
 /* Encoding of a block. Very similar to MPEG4 except for a different
    escape coding (same as H263) and more vlc tables.
  */
-void ff_msmpeg4_encode_block(MpegEncContext * s, DCTELEM * block, int n)
+void ff_msmpeg4_encode_block(MpegEncContext * s, int16_t * block, int n)
 {
     int level, run, last, i, j, last_index;
     int last_non_zero, sign, slevel;
diff --git a/libavcodec/ppc/dsputil_altivec.c b/libavcodec/ppc/dsputil_altivec.c
index 149cbf522a..ab8e3b7fb1 100644
--- a/libavcodec/ppc/dsputil_altivec.c
+++ b/libavcodec/ppc/dsputil_altivec.c
@@ -476,7 +476,7 @@ static int pix_sum_altivec(uint8_t * pix, int line_size)
     return s;
 }
 
-static void get_pixels_altivec(DCTELEM *av_restrict block, const uint8_t *pixels, int line_size)
+static void get_pixels_altivec(int16_t *restrict block, const uint8_t *pixels, int line_size)
 {
     int i;
     vector unsigned char perm = vec_lvsl(0, pixels);
@@ -502,7 +502,7 @@ static void get_pixels_altivec(DCTELEM *av_restrict block, const uint8_t *pixels
     }
 }
 
-static void diff_pixels_altivec(DCTELEM *av_restrict block, const uint8_t *s1,
+static void diff_pixels_altivec(int16_t *restrict block, const uint8_t *s1,
         const uint8_t *s2, int stride)
 {
     int i;
@@ -576,7 +576,7 @@ static void diff_pixels_altivec(DCTELEM *av_restrict block, const uint8_t *s1,
 }
 
 
-static void clear_block_altivec(DCTELEM *block) {
+static void clear_block_altivec(int16_t *block) {
     LOAD_ZERO;
     vec_st(zero_s16v,   0, block);
     vec_st(zero_s16v,  16, block);
diff --git a/libavcodec/ppc/dsputil_ppc.c b/libavcodec/ppc/dsputil_ppc.c
index 1537d92643..50aa188829 100644
--- a/libavcodec/ppc/dsputil_ppc.c
+++ b/libavcodec/ppc/dsputil_ppc.c
@@ -47,7 +47,7 @@ distinguish, and use dcbz (32 bytes) or dcbzl (one cache line) as required.
 see <http://developer.apple.com/technotes/tn/tn2087.html>
 and <http://developer.apple.com/technotes/tn/tn2086.html>
 */
-static void clear_blocks_dcbz32_ppc(DCTELEM *blocks)
+static void clear_blocks_dcbz32_ppc(int16_t *blocks)
 {
     register int misal = ((unsigned long)blocks & 0x00000010);
     register int i = 0;
@@ -58,7 +58,7 @@ static void clear_blocks_dcbz32_ppc(DCTELEM *blocks)
         ((unsigned long*)blocks)[3] = 0L;
         i += 16;
     }
-    for ( ; i < sizeof(DCTELEM)*6*64-31 ; i += 32) {
+    for ( ; i < sizeof(int16_t)*6*64-31 ; i += 32) {
         __asm__ volatile("dcbz %0,%1" : : "b" (blocks), "r" (i) : "memory");
     }
     if (misal) {
@@ -73,7 +73,7 @@ static void clear_blocks_dcbz32_ppc(DCTELEM *blocks)
 /* same as above, when dcbzl clear a whole 128B cache line
    i.e. the PPC970 aka G5 */
 #if HAVE_DCBZL
-static void clear_blocks_dcbz128_ppc(DCTELEM *blocks)
+static void clear_blocks_dcbz128_ppc(int16_t *blocks)
 {
     register int misal = ((unsigned long)blocks & 0x0000007f);
     register int i = 0;
@@ -81,17 +81,17 @@ static void clear_blocks_dcbz128_ppc(DCTELEM *blocks)
         // we could probably also optimize this case,
         // but there's not much point as the machines
         // aren't available yet (2003-06-26)
-        memset(blocks, 0, sizeof(DCTELEM)*6*64);
+        memset(blocks, 0, sizeof(int16_t)*6*64);
     }
     else
-        for ( ; i < sizeof(DCTELEM)*6*64 ; i += 128) {
+        for ( ; i < sizeof(int16_t)*6*64 ; i += 128) {
             __asm__ volatile("dcbzl %0,%1" : : "b" (blocks), "r" (i) : "memory");
         }
 }
 #else
-static void clear_blocks_dcbz128_ppc(DCTELEM *blocks)
+static void clear_blocks_dcbz128_ppc(int16_t *blocks)
 {
-    memset(blocks, 0, sizeof(DCTELEM)*6*64);
+    memset(blocks, 0, sizeof(int16_t)*6*64);
 }
 #endif
 
diff --git a/libavcodec/ppc/h264_altivec.c b/libavcodec/ppc/h264_altivec.c
index be54b48303..619f326d55 100644
--- a/libavcodec/ppc/h264_altivec.c
+++ b/libavcodec/ppc/h264_altivec.c
@@ -315,7 +315,7 @@ H264_MC(avg_, 16, altivec)
     va_u32 = vec_splat((vec_u32)va_u8, 0);                  \
     vec_ste(va_u32, element, (uint32_t*)dst);
 
-static void ff_h264_idct_add_altivec(uint8_t *dst, DCTELEM *block, int stride)
+static void ff_h264_idct_add_altivec(uint8_t *dst, int16_t *block, int stride)
 {
     vec_s16 va0, va1, va2, va3;
     vec_s16 vz0, vz1, vz2, vz3;
@@ -429,7 +429,7 @@ static void ff_h264_idct_add_altivec(uint8_t *dst, DCTELEM *block, int stride)
     vec_st( hv, 0, dest );                                     \
  }
 
-static void ff_h264_idct8_add_altivec( uint8_t *dst, DCTELEM *dct, int stride ) {
+static void ff_h264_idct8_add_altivec( uint8_t *dst, int16_t *dct, int stride ) {
     vec_s16 s0, s1, s2, s3, s4, s5, s6, s7;
     vec_s16 d0, d1, d2, d3, d4, d5, d6, d7;
     vec_s16 idct0, idct1, idct2, idct3, idct4, idct5, idct6, idct7;
@@ -473,7 +473,7 @@ static void ff_h264_idct8_add_altivec( uint8_t *dst, DCTELEM *dct, int stride )
     ALTIVEC_STORE_SUM_CLIP(&dst[7*stride], idct7, perm_ldv, perm_stv, sel);
 }
 
-static av_always_inline void h264_idct_dc_add_internal(uint8_t *dst, DCTELEM *block, int stride, int size)
+static av_always_inline void h264_idct_dc_add_internal(uint8_t *dst, int16_t *block, int stride, int size)
 {
     vec_s16 dc16;
     vec_u8 dcplus, dcminus, v0, v1, v2, v3, aligner;
@@ -518,17 +518,17 @@ static av_always_inline void h264_idct_dc_add_internal(uint8_t *dst, DCTELEM *bl
     }
 }
 
-static void h264_idct_dc_add_altivec(uint8_t *dst, DCTELEM *block, int stride)
+static void h264_idct_dc_add_altivec(uint8_t *dst, int16_t *block, int stride)
 {
     h264_idct_dc_add_internal(dst, block, stride, 4);
 }
 
-static void ff_h264_idct8_dc_add_altivec(uint8_t *dst, DCTELEM *block, int stride)
+static void ff_h264_idct8_dc_add_altivec(uint8_t *dst, int16_t *block, int stride)
 {
     h264_idct_dc_add_internal(dst, block, stride, 8);
 }
 
-static void ff_h264_idct_add16_altivec(uint8_t *dst, const int *block_offset, DCTELEM *block, int stride, const uint8_t nnzc[15*8]){
+static void ff_h264_idct_add16_altivec(uint8_t *dst, const int *block_offset, int16_t *block, int stride, const uint8_t nnzc[15*8]){
     int i;
     for(i=0; i<16; i++){
         int nnz = nnzc[ scan8[i] ];
@@ -539,7 +539,7 @@ static void ff_h264_idct_add16_altivec(uint8_t *dst, const int *block_offset, DC
     }
 }
 
-static void ff_h264_idct_add16intra_altivec(uint8_t *dst, const int *block_offset, DCTELEM *block, int stride, const uint8_t nnzc[15*8]){
+static void ff_h264_idct_add16intra_altivec(uint8_t *dst, const int *block_offset, int16_t *block, int stride, const uint8_t nnzc[15*8]){
     int i;
     for(i=0; i<16; i++){
         if(nnzc[ scan8[i] ]) ff_h264_idct_add_altivec(dst + block_offset[i], block + i*16, stride);
@@ -547,7 +547,7 @@ static void ff_h264_idct_add16intra_altivec(uint8_t *dst, const int *block_offse
     }
 }
 
-static void ff_h264_idct8_add4_altivec(uint8_t *dst, const int *block_offset, DCTELEM *block, int stride, const uint8_t nnzc[15*8]){
+static void ff_h264_idct8_add4_altivec(uint8_t *dst, const int *block_offset, int16_t *block, int stride, const uint8_t nnzc[15*8]){
     int i;
     for(i=0; i<16; i+=4){
         int nnz = nnzc[ scan8[i] ];
@@ -558,7 +558,7 @@ static void ff_h264_idct8_add4_altivec(uint8_t *dst, const int *block_offset, DC
     }
 }
 
-static void ff_h264_idct_add8_altivec(uint8_t **dest, const int *block_offset, DCTELEM *block, int stride, const uint8_t nnzc[15*8]){
+static void ff_h264_idct_add8_altivec(uint8_t **dest, const int *block_offset, int16_t *block, int stride, const uint8_t nnzc[15*8]){
     int i, j;
     for (j = 1; j < 3; j++) {
         for(i = j * 16; i < j * 16 + 4; i++){
diff --git a/libavcodec/ppc/mpegvideo_altivec.c b/libavcodec/ppc/mpegvideo_altivec.c
index bbc102bc21..6a1a06537f 100644
--- a/libavcodec/ppc/mpegvideo_altivec.c
+++ b/libavcodec/ppc/mpegvideo_altivec.c
@@ -35,7 +35,7 @@
 /* AltiVec version of dct_unquantize_h263
    this code assumes `block' is 16 bytes-aligned */
 static void dct_unquantize_h263_altivec(MpegEncContext *s,
-                                 DCTELEM *block, int n, int qscale)
+                                 int16_t *block, int n, int qscale)
 {
     int i, level, qmul, qadd;
     int nCoeffs;
diff --git a/libavcodec/ppc/vc1dsp_altivec.c b/libavcodec/ppc/vc1dsp_altivec.c
index 33e87afc3a..c1a0055cf1 100644
--- a/libavcodec/ppc/vc1dsp_altivec.c
+++ b/libavcodec/ppc/vc1dsp_altivec.c
@@ -129,7 +129,7 @@ do { \
 
 /** Do inverse transform on 8x8 block
 */
-static void vc1_inv_trans_8x8_altivec(DCTELEM block[64])
+static void vc1_inv_trans_8x8_altivec(int16_t block[64])
 {
     vector signed short src0, src1, src2, src3, src4, src5, src6, src7;
     vector signed int s0, s1, s2, s3, s4, s5, s6, s7;
@@ -224,7 +224,7 @@ static void vc1_inv_trans_8x8_altivec(DCTELEM block[64])
 
 /** Do inverse transform on 8x4 part of block
 */
-static void vc1_inv_trans_8x4_altivec(uint8_t *dest, int stride, DCTELEM *block)
+static void vc1_inv_trans_8x4_altivec(uint8_t *dest, int stride, int16_t *block)
 {
     vector signed short src0, src1, src2, src3, src4, src5, src6, src7;
     vector signed int s0, s1, s2, s3, s4, s5, s6, s7;
diff --git a/libavcodec/ppc/vp3dsp_altivec.c b/libavcodec/ppc/vp3dsp_altivec.c
index 6271dea560..54e291e483 100644
--- a/libavcodec/ppc/vp3dsp_altivec.c
+++ b/libavcodec/ppc/vp3dsp_altivec.c
@@ -116,7 +116,7 @@ static inline vec_s16 M16(vec_s16 a, vec_s16 C)
 #define ADD8(a) vec_add(a, eight)
 #define SHIFT4(a) vec_sra(a, four)
 
-static void vp3_idct_put_altivec(uint8_t *dst, int stride, DCTELEM block[64])
+static void vp3_idct_put_altivec(uint8_t *dst, int stride, int16_t block[64])
 {
     vec_u8 t;
     IDCT_START
@@ -145,7 +145,7 @@ static void vp3_idct_put_altivec(uint8_t *dst, int stride, DCTELEM block[64])
     memset(block, 0, sizeof(*block) * 64);
 }
 
-static void vp3_idct_add_altivec(uint8_t *dst, int stride, DCTELEM block[64])
+static void vp3_idct_add_altivec(uint8_t *dst, int stride, int16_t block[64])
 {
     LOAD_ZERO;
     vec_u8 t, vdst;
diff --git a/libavcodec/proresdec2.c b/libavcodec/proresdec2.c
index 3b3175a41a..4634e704e2 100644
--- a/libavcodec/proresdec2.c
+++ b/libavcodec/proresdec2.c
@@ -294,10 +294,10 @@ static int decode_picture_header(AVCodecContext *avctx, const uint8_t *buf, cons
 
 static const uint8_t dc_codebook[7] = { 0x04, 0x28, 0x28, 0x4D, 0x4D, 0x70, 0x70};
 
-static av_always_inline void decode_dc_coeffs(GetBitContext *gb, DCTELEM *out,
+static av_always_inline void decode_dc_coeffs(GetBitContext *gb, int16_t *out,
                                               int blocks_per_slice)
 {
-    DCTELEM prev_dc;
+    int16_t prev_dc;
     int code, i, sign;
 
     OPEN_READER(re, gb);
@@ -325,7 +325,7 @@ static const uint8_t run_to_cb[16] = { 0x06, 0x06, 0x05, 0x05, 0x04, 0x29, 0x29,
 static const uint8_t lev_to_cb[10] = { 0x04, 0x0A, 0x05, 0x06, 0x04, 0x28, 0x28, 0x28, 0x28, 0x4C };
 
 static av_always_inline void decode_ac_coeffs(AVCodecContext *avctx, GetBitContext *gb,
-                                              DCTELEM *out, int blocks_per_slice)
+                                              int16_t *out, int blocks_per_slice)
 {
     ProresContext *ctx = avctx->priv_data;
     int block_mask, sign;
@@ -372,8 +372,8 @@ static void decode_slice_luma(AVCodecContext *avctx, SliceContext *slice,
                               const int16_t *qmat)
 {
     ProresContext *ctx = avctx->priv_data;
-    LOCAL_ALIGNED_16(DCTELEM, blocks, [8*4*64]);
-    DCTELEM *block;
+    LOCAL_ALIGNED_16(int16_t, blocks, [8*4*64]);
+    int16_t *block;
     GetBitContext gb;
     int i, blocks_per_slice = slice->mb_count<<2;
 
@@ -402,8 +402,8 @@ static void decode_slice_chroma(AVCodecContext *avctx, SliceContext *slice,
                                 const int16_t *qmat, int log2_blocks_per_mb)
 {
     ProresContext *ctx = avctx->priv_data;
-    LOCAL_ALIGNED_16(DCTELEM, blocks, [8*4*64]);
-    DCTELEM *block;
+    LOCAL_ALIGNED_16(int16_t, blocks, [8*4*64]);
+    int16_t *block;
     GetBitContext gb;
     int i, j, blocks_per_slice = slice->mb_count << log2_blocks_per_mb;
 
diff --git a/libavcodec/proresdec_lgpl.c b/libavcodec/proresdec_lgpl.c
index a25ecca3c9..5c53882ce3 100644
--- a/libavcodec/proresdec_lgpl.c
+++ b/libavcodec/proresdec_lgpl.c
@@ -34,6 +34,7 @@
 
 #include "libavutil/intmath.h"
 #include "avcodec.h"
+#include "dsputil.h"
 #include "internal.h"
 #include "proresdata.h"
 #include "proresdsp.h"
@@ -45,7 +46,7 @@ typedef struct {
     int x_pos, y_pos;
     int slice_width;
     int prev_slice_sf;               ///< scalefactor of the previous decoded slice
-    DECLARE_ALIGNED(16, DCTELEM, blocks)[8 * 4 * 64];
+    DECLARE_ALIGNED(16, int16_t, blocks)[8 * 4 * 64];
     DECLARE_ALIGNED(16, int16_t, qmat_luma_scaled)[64];
     DECLARE_ALIGNED(16, int16_t, qmat_chroma_scaled)[64];
 } ProresThreadData;
@@ -340,10 +341,10 @@ static inline int decode_vlc_codeword(GetBitContext *gb, unsigned codebook)
 /**
  * Decode DC coefficients for all blocks in a slice.
  */
-static inline void decode_dc_coeffs(GetBitContext *gb, DCTELEM *out,
+static inline void decode_dc_coeffs(GetBitContext *gb, int16_t *out,
                                     int nblocks)
 {
-    DCTELEM prev_dc;
+    int16_t prev_dc;
     int     i, sign;
     int16_t delta;
     unsigned int code;
@@ -368,7 +369,7 @@ static inline void decode_dc_coeffs(GetBitContext *gb, DCTELEM *out,
 /**
  * Decode AC coefficients for all blocks in a slice.
  */
-static inline void decode_ac_coeffs(GetBitContext *gb, DCTELEM *out,
+static inline void decode_ac_coeffs(GetBitContext *gb, int16_t *out,
                                     int blocks_per_slice,
                                     int plane_size_factor,
                                     const uint8_t *scan)
@@ -421,7 +422,7 @@ static void decode_slice_plane(ProresContext *ctx, ProresThreadData *td,
                                const int16_t *qmat, int is_chroma)
 {
     GetBitContext gb;
-    DCTELEM *block_ptr;
+    int16_t *block_ptr;
     int mb_num, blocks_per_slice;
 
     blocks_per_slice = mbs_per_slice * blocks_per_mb;
diff --git a/libavcodec/proresdsp.c b/libavcodec/proresdsp.c
index 9e063b0974..dd6d72511e 100644
--- a/libavcodec/proresdsp.c
+++ b/libavcodec/proresdsp.c
@@ -20,6 +20,7 @@
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  */
 
+#include "dsputil.h"
 #include "proresdsp.h"
 #include "simple_idct.h"
 #include "libavutil/common.h"
@@ -34,7 +35,7 @@
 /**
  * Add bias value, clamp and output pixels of a slice
  */
-static void put_pixels(uint16_t *dst, int stride, const DCTELEM *in)
+static void put_pixels(uint16_t *dst, int stride, const int16_t *in)
 {
     int x, y, src_offset, dst_offset;
 
@@ -47,7 +48,7 @@ static void put_pixels(uint16_t *dst, int stride, const DCTELEM *in)
     }
 }
 
-static void prores_idct_put_c(uint16_t *out, int linesize, DCTELEM *block, const int16_t *qmat)
+static void prores_idct_put_c(uint16_t *out, int linesize, int16_t *block, const int16_t *qmat)
 {
     ff_prores_idct(block, qmat);
     put_pixels(out, linesize >> 1, block);
@@ -55,7 +56,7 @@ static void prores_idct_put_c(uint16_t *out, int linesize, DCTELEM *block, const
 #endif
 
 #if CONFIG_PRORES_KOSTYA_ENCODER
-static void prores_fdct_c(const uint16_t *src, int linesize, DCTELEM *block)
+static void prores_fdct_c(const uint16_t *src, int linesize, int16_t *block)
 {
     int x, y;
     const uint16_t *tsrc = src;
diff --git a/libavcodec/proresdsp.h b/libavcodec/proresdsp.h
index 09541d9642..a9df91851f 100644
--- a/libavcodec/proresdsp.h
+++ b/libavcodec/proresdsp.h
@@ -23,7 +23,7 @@
 #ifndef AVCODEC_PRORESDSP_H
 #define AVCODEC_PRORESDSP_H
 
-#include "dsputil.h"
+#include <stdint.h>
 
 #define PRORES_BITS_PER_SAMPLE 10 ///< output precision of prores decoder
 
@@ -32,8 +32,8 @@ typedef struct ProresDSPContext {
     uint8_t idct_permutation[64];
     int dct_permutation_type;
     uint8_t dct_permutation[64];
-    void (* idct_put) (uint16_t *out, int linesize, DCTELEM *block, const int16_t *qmat);
-    void (* fdct) (const uint16_t *src, int linesize, DCTELEM *block);
+    void (* idct_put) (uint16_t *out, int linesize, int16_t *block, const int16_t *qmat);
+    void (* fdct) (const uint16_t *src, int linesize, int16_t *block);
 } ProresDSPContext;
 
 void ff_proresdsp_init(ProresDSPContext *dsp, AVCodecContext *avctx);
diff --git a/libavcodec/proresenc_anatoliy.c b/libavcodec/proresenc_anatoliy.c
index 8c71ca5aea..1e73757d73 100644
--- a/libavcodec/proresenc_anatoliy.c
+++ b/libavcodec/proresenc_anatoliy.c
@@ -198,7 +198,7 @@ static av_always_inline int get_level(int val)
 
 static const uint8_t dc_codebook[7] = { 0x04, 0x28, 0x28, 0x4D, 0x4D, 0x70, 0x70};
 
-static void encode_dc_coeffs(PutBitContext *pb, DCTELEM *in,
+static void encode_dc_coeffs(PutBitContext *pb, int16_t *in,
         int blocks_per_slice, int *qmat)
 {
     int prev_dc, code;
@@ -230,7 +230,7 @@ static const uint8_t lev_to_cb[10] = { 0x04, 0x0A, 0x05, 0x06, 0x04, 0x28,
         0x28, 0x28, 0x28, 0x4C };
 
 static void encode_ac_coeffs(AVCodecContext *avctx, PutBitContext *pb,
-        DCTELEM *in, int blocks_per_slice, int *qmat)
+        int16_t *in, int blocks_per_slice, int *qmat)
 {
     int prev_run = 4;
     int prev_level = 2;
@@ -260,7 +260,7 @@ static void encode_ac_coeffs(AVCodecContext *avctx, PutBitContext *pb,
     }
 }
 
-static void get(uint8_t *pixels, int stride, DCTELEM* block)
+static void get(uint8_t *pixels, int stride, int16_t* block)
 {
     int16_t *p = (int16_t*)pixels;
     int i, j;
@@ -275,7 +275,7 @@ static void get(uint8_t *pixels, int stride, DCTELEM* block)
     }
 }
 
-static void fdct_get(uint8_t *pixels, int stride, DCTELEM* block)
+static void fdct_get(uint8_t *pixels, int stride, int16_t* block)
 {
     get(pixels, stride, block);
     ff_jpeg_fdct_islow_10(block);
@@ -285,7 +285,7 @@ static int encode_slice_plane(AVCodecContext *avctx, int mb_count,
         uint8_t *src, int src_stride, uint8_t *buf, unsigned buf_size,
         int *qmat, int chroma)
 {
-    DECLARE_ALIGNED(16, DCTELEM, blocks)[DEFAULT_SLICE_MB_WIDTH << 8], *block;
+    DECLARE_ALIGNED(16, int16_t, blocks)[DEFAULT_SLICE_MB_WIDTH << 8], *block;
     int i, blocks_per_slice;
     PutBitContext pb;
 
diff --git a/libavcodec/proresenc_kostya.c b/libavcodec/proresenc_kostya.c
index 0462c74b13..8da13ac822 100644
--- a/libavcodec/proresenc_kostya.c
+++ b/libavcodec/proresenc_kostya.c
@@ -25,6 +25,7 @@
 
 #include "libavutil/opt.h"
 #include "avcodec.h"
+#include "dsputil.h"
 #include "put_bits.h"
 #include "bytestream.h"
 #include "internal.h"
@@ -170,7 +171,7 @@ struct TrellisNode {
 #define MAX_STORED_Q 16
 
 typedef struct ProresThreadData {
-    DECLARE_ALIGNED(16, DCTELEM, blocks)[MAX_PLANES][64 * 4 * MAX_MBS_PER_SLICE];
+    DECLARE_ALIGNED(16, int16_t, blocks)[MAX_PLANES][64 * 4 * MAX_MBS_PER_SLICE];
     DECLARE_ALIGNED(16, uint16_t, emu_buf)[16 * 16];
     int16_t custom_q[64];
     struct TrellisNode *nodes;
@@ -178,7 +179,7 @@ typedef struct ProresThreadData {
 
 typedef struct ProresContext {
     AVClass *class;
-    DECLARE_ALIGNED(16, DCTELEM, blocks)[MAX_PLANES][64 * 4 * MAX_MBS_PER_SLICE];
+    DECLARE_ALIGNED(16, int16_t, blocks)[MAX_PLANES][64 * 4 * MAX_MBS_PER_SLICE];
     DECLARE_ALIGNED(16, uint16_t, emu_buf)[16*16];
     int16_t quants[MAX_STORED_Q][64];
     int16_t custom_q[64];
@@ -213,7 +214,7 @@ typedef struct ProresContext {
 
 static void get_slice_data(ProresContext *ctx, const uint16_t *src,
                            int linesize, int x, int y, int w, int h,
-                           DCTELEM *blocks, uint16_t *emu_buf,
+                           int16_t *blocks, uint16_t *emu_buf,
                            int mbs_per_slice, int blocks_per_mb, int is_chroma)
 {
     const uint16_t *esrc;
@@ -317,7 +318,7 @@ static inline void encode_vlc_codeword(PutBitContext *pb, unsigned codebook, int
 #define GET_SIGN(x)  ((x) >> 31)
 #define MAKE_CODE(x) (((x) << 1) ^ GET_SIGN(x))
 
-static void encode_dcs(PutBitContext *pb, DCTELEM *blocks,
+static void encode_dcs(PutBitContext *pb, int16_t *blocks,
                        int blocks_per_slice, int scale)
 {
     int i;
@@ -343,7 +344,7 @@ static void encode_dcs(PutBitContext *pb, DCTELEM *blocks,
     }
 }
 
-static void encode_acs(PutBitContext *pb, DCTELEM *blocks,
+static void encode_acs(PutBitContext *pb, int16_t *blocks,
                        int blocks_per_slice,
                        int plane_size_factor,
                        const uint8_t *scan, const int16_t *qmat)
@@ -379,7 +380,7 @@ static void encode_acs(PutBitContext *pb, DCTELEM *blocks,
 
 static int encode_slice_plane(ProresContext *ctx, PutBitContext *pb,
                               const uint16_t *src, int linesize,
-                              int mbs_per_slice, DCTELEM *blocks,
+                              int mbs_per_slice, int16_t *blocks,
                               int blocks_per_mb, int plane_size_factor,
                               const int16_t *qmat)
 {
@@ -481,7 +482,7 @@ static inline int estimate_vlc(unsigned codebook, int val)
     }
 }
 
-static int estimate_dcs(int *error, DCTELEM *blocks, int blocks_per_slice,
+static int estimate_dcs(int *error, int16_t *blocks, int blocks_per_slice,
                         int scale)
 {
     int i;
@@ -512,7 +513,7 @@ static int estimate_dcs(int *error, DCTELEM *blocks, int blocks_per_slice,
     return bits;
 }
 
-static int estimate_acs(int *error, DCTELEM *blocks, int blocks_per_slice,
+static int estimate_acs(int *error, int16_t *blocks, int blocks_per_slice,
                         int plane_size_factor,
                         const uint8_t *scan, const int16_t *qmat)
 {
diff --git a/libavcodec/rtjpeg.c b/libavcodec/rtjpeg.c
index 7797a655c1..2bd4416fdf 100644
--- a/libavcodec/rtjpeg.c
+++ b/libavcodec/rtjpeg.c
@@ -44,7 +44,7 @@
  * aligned this could be done faster in a different way, e.g. as it is done
  * in MPlayer libmpcodecs/native/rtjpegn.c.
  */
-static inline int get_block(GetBitContext *gb, DCTELEM *block, const uint8_t *scan,
+static inline int get_block(GetBitContext *gb, int16_t *block, const uint8_t *scan,
                             const uint32_t *quant) {
     int coeff, i, n;
     int8_t ac;
@@ -61,7 +61,7 @@ static inline int get_block(GetBitContext *gb, DCTELEM *block, const uint8_t *sc
 
     // normally we would only need to clear the (63 - coeff) last values,
     // but since we do not know where they are we just clear the whole block
-    memset(block, 0, 64 * sizeof(DCTELEM));
+    memset(block, 0, 64 * sizeof(int16_t));
 
     // 2 bits per coefficient
     while (coeff) {
@@ -121,7 +121,7 @@ int ff_rtjpeg_decode_frame_yuv420(RTJpegContext *c, AVFrame *f,
     if (res > 0) \
         c->dsp->idct_put(dst, stride, block); \
 } while (0)
-            DCTELEM *block = c->block;
+            int16_t *block = c->block;
             BLOCK(c->lquant, y1, f->linesize[0]);
             y1 += 8;
             BLOCK(c->lquant, y1, f->linesize[0]);
diff --git a/libavcodec/rtjpeg.h b/libavcodec/rtjpeg.h
index 0fd24a356c..294b06c697 100644
--- a/libavcodec/rtjpeg.h
+++ b/libavcodec/rtjpeg.h
@@ -35,7 +35,7 @@ typedef struct RTJpegContext {
     uint8_t scan[64];
     uint32_t lquant[64];
     uint32_t cquant[64];
-    DECLARE_ALIGNED(16, DCTELEM, block)[64];
+    DECLARE_ALIGNED(16, int16_t, block)[64];
 } RTJpegContext;
 
 void ff_rtjpeg_decode_init(RTJpegContext *c, DSPContext *dsp,
diff --git a/libavcodec/rv34.c b/libavcodec/rv34.c
index 73b6cb78d9..7a9778a101 100644
--- a/libavcodec/rv34.c
+++ b/libavcodec/rv34.c
@@ -217,7 +217,7 @@ static int rv34_decode_cbp(GetBitContext *gb, RV34VLC *vlc, int table)
 /**
  * Get one coefficient value from the bitstream and store it.
  */
-static inline void decode_coeff(DCTELEM *dst, int coef, int esc, GetBitContext *gb, VLC* vlc, int q)
+static inline void decode_coeff(int16_t *dst, int coef, int esc, GetBitContext *gb, VLC* vlc, int q)
 {
     if(coef){
         if(coef == esc){
@@ -237,7 +237,7 @@ static inline void decode_coeff(DCTELEM *dst, int coef, int esc, GetBitContext *
 /**
  * Decode 2x2 subblock of coefficients.
  */
-static inline void decode_subblock(DCTELEM *dst, int code, const int is_block2, GetBitContext *gb, VLC *vlc, int q)
+static inline void decode_subblock(int16_t *dst, int code, const int is_block2, GetBitContext *gb, VLC *vlc, int q)
 {
     int flags = modulo_three_table[code];
 
@@ -255,13 +255,13 @@ static inline void decode_subblock(DCTELEM *dst, int code, const int is_block2,
 /**
  * Decode a single coefficient.
  */
-static inline void decode_subblock1(DCTELEM *dst, int code, GetBitContext *gb, VLC *vlc, int q)
+static inline void decode_subblock1(int16_t *dst, int code, GetBitContext *gb, VLC *vlc, int q)
 {
     int coeff = modulo_three_table[code] >> 6;
     decode_coeff(dst, coeff, 3, gb, vlc, q);
 }
 
-static inline void decode_subblock3(DCTELEM *dst, int code, GetBitContext *gb, VLC *vlc,
+static inline void decode_subblock3(int16_t *dst, int code, GetBitContext *gb, VLC *vlc,
                                     int q_dc, int q_ac1, int q_ac2)
 {
     int flags = modulo_three_table[code];
@@ -283,7 +283,7 @@ static inline void decode_subblock3(DCTELEM *dst, int code, GetBitContext *gb, V
  *  o--o
  */
 
-static int rv34_decode_block(DCTELEM *dst, GetBitContext *gb, RV34VLC *rvlc, int fc, int sc, int q_dc, int q_ac1, int q_ac2)
+static int rv34_decode_block(int16_t *dst, GetBitContext *gb, RV34VLC *rvlc, int fc, int sc, int q_dc, int q_ac1, int q_ac2)
 {
     int code, pattern, has_ac = 1;
 
@@ -995,7 +995,7 @@ static inline void rv34_process_block(RV34DecContext *r,
                                       int fc, int sc, int q_dc, int q_ac)
 {
     MpegEncContext *s = &r->s;
-    DCTELEM *ptr = s->block[0];
+    int16_t *ptr = s->block[0];
     int has_ac = rv34_decode_block(ptr, &s->gb, r->cur_vlcs,
                                    fc, sc, q_dc, q_ac, q_ac);
     if(has_ac){
@@ -1008,13 +1008,13 @@ static inline void rv34_process_block(RV34DecContext *r,
 
 static void rv34_output_i16x16(RV34DecContext *r, int8_t *intra_types, int cbp)
 {
-    LOCAL_ALIGNED_16(DCTELEM, block16, [16]);
+    LOCAL_ALIGNED_16(int16_t, block16, [16]);
     MpegEncContext *s    = &r->s;
     GetBitContext  *gb   = &s->gb;
     int             q_dc = rv34_qscale_tab[ r->luma_dc_quant_i[s->qscale] ],
                     q_ac = rv34_qscale_tab[s->qscale];
     uint8_t        *dst  = s->dest[0];
-    DCTELEM        *ptr  = s->block[0];
+    int16_t        *ptr  = s->block[0];
     int i, j, itype, has_ac;
 
     memset(block16, 0, 16 * sizeof(*block16));
@@ -1180,7 +1180,7 @@ static int rv34_decode_inter_macroblock(RV34DecContext *r, int8_t *intra_types)
     MpegEncContext *s   = &r->s;
     GetBitContext  *gb  = &s->gb;
     uint8_t        *dst = s->dest[0];
-    DCTELEM        *ptr = s->block[0];
+    int16_t        *ptr = s->block[0];
     int          mb_pos = s->mb_x + s->mb_y * s->mb_stride;
     int cbp, cbp2;
     int q_dc, q_ac, has_ac;
@@ -1220,7 +1220,7 @@ static int rv34_decode_inter_macroblock(RV34DecContext *r, int8_t *intra_types)
 
     if(r->is16){
         // Only for RV34_MB_P_MIX16x16
-        LOCAL_ALIGNED_16(DCTELEM, block16, [16]);
+        LOCAL_ALIGNED_16(int16_t, block16, [16]);
         memset(block16, 0, 16 * sizeof(*block16));
         q_dc = rv34_qscale_tab[ r->luma_dc_quant_p[s->qscale] ];
         q_ac = rv34_qscale_tab[s->qscale];
diff --git a/libavcodec/rv34dsp.c b/libavcodec/rv34dsp.c
index 86a2ffd58e..676e823787 100644
--- a/libavcodec/rv34dsp.c
+++ b/libavcodec/rv34dsp.c
@@ -33,7 +33,7 @@
  * @{
  */
 
-static av_always_inline void rv34_row_transform(int temp[16], DCTELEM *block)
+static av_always_inline void rv34_row_transform(int temp[16], int16_t *block)
 {
     int i;
 
@@ -54,12 +54,12 @@ static av_always_inline void rv34_row_transform(int temp[16], DCTELEM *block)
  * Real Video 3.0/4.0 inverse transform + sample reconstruction
  * Code is almost the same as in SVQ3, only scaling is different.
  */
-static void rv34_idct_add_c(uint8_t *dst, ptrdiff_t stride, DCTELEM *block){
+static void rv34_idct_add_c(uint8_t *dst, ptrdiff_t stride, int16_t *block){
     int      temp[16];
     int      i;
 
     rv34_row_transform(temp, block);
-    memset(block, 0, 16*sizeof(DCTELEM));
+    memset(block, 0, 16*sizeof(int16_t));
 
     for(i = 0; i < 4; i++){
         const int z0 = 13*(temp[4*0+i] +    temp[4*2+i]) + 0x200;
@@ -82,7 +82,7 @@ static void rv34_idct_add_c(uint8_t *dst, ptrdiff_t stride, DCTELEM *block){
  * Code is almost the same as rv34_inv_transform()
  * but final coefficients are multiplied by 1.5 and have no rounding.
  */
-static void rv34_inv_transform_noround_c(DCTELEM *block){
+static void rv34_inv_transform_noround_c(int16_t *block){
     int temp[16];
     int i;
 
@@ -115,9 +115,9 @@ static void rv34_idct_dc_add_c(uint8_t *dst, ptrdiff_t stride, int dc)
     }
 }
 
-static void rv34_inv_transform_dc_noround_c(DCTELEM *block)
+static void rv34_inv_transform_dc_noround_c(int16_t *block)
 {
-    DCTELEM dc = (13 * 13 * 3 * block[0]) >> 11;
+    int16_t dc = (13 * 13 * 3 * block[0]) >> 11;
     int i, j;
 
     for (i = 0; i < 4; i++, block += 4)
diff --git a/libavcodec/rv34dsp.h b/libavcodec/rv34dsp.h
index f0263b12fe..e04a47d35d 100644
--- a/libavcodec/rv34dsp.h
+++ b/libavcodec/rv34dsp.h
@@ -34,9 +34,9 @@ typedef void (*rv40_weight_func)(uint8_t *dst/*align width (8 or 16)*/,
                                  uint8_t *src2/*align width (8 or 16)*/,
                                  int w1, int w2, ptrdiff_t stride);
 
-typedef void (*rv34_inv_transform_func)(DCTELEM *block);
+typedef void (*rv34_inv_transform_func)(int16_t *block);
 
-typedef void (*rv34_idct_add_func)(uint8_t *dst, ptrdiff_t stride, DCTELEM *block);
+typedef void (*rv34_idct_add_func)(uint8_t *dst, ptrdiff_t stride, int16_t *block);
 typedef void (*rv34_idct_dc_add_func)(uint8_t *dst, ptrdiff_t stride,
                                       int   dc);
 
diff --git a/libavcodec/sh4/dsputil_sh4.c b/libavcodec/sh4/dsputil_sh4.c
index 7deab41d62..5296bcf1b7 100644
--- a/libavcodec/sh4/dsputil_sh4.c
+++ b/libavcodec/sh4/dsputil_sh4.c
@@ -47,12 +47,12 @@ static void memzero_align8(void *dst,size_t size)
         fp_single_leave(fpscr);
 }
 
-static void clear_blocks_sh4(DCTELEM *blocks)
+static void clear_blocks_sh4(int16_t *blocks)
 {
-        memzero_align8(blocks,sizeof(DCTELEM)*6*64);
+        memzero_align8(blocks,sizeof(int16_t)*6*64);
 }
 
-static void idct_put(uint8_t *dest, int line_size, DCTELEM *block)
+static void idct_put(uint8_t *dest, int line_size, int16_t *block)
 {
         int i;
         uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;
@@ -70,7 +70,7 @@ static void idct_put(uint8_t *dest, int line_size, DCTELEM *block)
                 block+=8;
         }
 }
-static void idct_add(uint8_t *dest, int line_size, DCTELEM *block)
+static void idct_add(uint8_t *dest, int line_size, int16_t *block)
 {
         int i;
         uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;
diff --git a/libavcodec/sh4/dsputil_sh4.h b/libavcodec/sh4/dsputil_sh4.h
index f2cb11e378..1c041ae27d 100644
--- a/libavcodec/sh4/dsputil_sh4.h
+++ b/libavcodec/sh4/dsputil_sh4.h
@@ -22,7 +22,7 @@
 #include "libavcodec/avcodec.h"
 #include "libavcodec/dsputil.h"
 
-void ff_idct_sh4(DCTELEM *block);
+void ff_idct_sh4(int16_t *block);
 void ff_dsputil_init_align(DSPContext* c, AVCodecContext *avctx);
 
 #endif /* AVCODEC_SH4_DSPUTIL_SH4_H */
diff --git a/libavcodec/sh4/idct_sh4.c b/libavcodec/sh4/idct_sh4.c
index e54c330857..654a167710 100644
--- a/libavcodec/sh4/idct_sh4.c
+++ b/libavcodec/sh4/idct_sh4.c
@@ -89,7 +89,7 @@ static const float odd_table[] __attribute__ ((aligned(8))) = {
 
 //optimized
 
-void ff_idct_sh4(DCTELEM *block)
+void ff_idct_sh4(int16_t *block)
 {
         DEFREG;
 
diff --git a/libavcodec/simple_idct.c b/libavcodec/simple_idct.c
index 293185074b..2cbe568c20 100644
--- a/libavcodec/simple_idct.c
+++ b/libavcodec/simple_idct.c
@@ -50,7 +50,7 @@
    and the butterfly must be multiplied by 0.5 * sqrt(2.0) */
 #define C_SHIFT (4+1+12)
 
-static inline void idct4col_put(uint8_t *dest, int line_size, const DCTELEM *col)
+static inline void idct4col_put(uint8_t *dest, int line_size, const int16_t *col)
 {
     int c0, c1, c2, c3, a0, a1, a2, a3;
 
@@ -86,10 +86,10 @@ static inline void idct4col_put(uint8_t *dest, int line_size, const DCTELEM *col
 /* XXX: I think a 1.0/sqrt(2) normalization should be needed to
    compensate the extra butterfly stage - I don't have the full DV
    specification */
-void ff_simple_idct248_put(uint8_t *dest, int line_size, DCTELEM *block)
+void ff_simple_idct248_put(uint8_t *dest, int line_size, int16_t *block)
 {
     int i;
-    DCTELEM *ptr;
+    int16_t *ptr;
 
     /* butterfly */
     ptr = block;
@@ -129,7 +129,7 @@ void ff_simple_idct248_put(uint8_t *dest, int line_size, DCTELEM *block)
 #define C2 C_FIX(0.2705980501)
 #define C3 C_FIX(0.5)
 #define C_SHIFT (4+1+12)
-static inline void idct4col_add(uint8_t *dest, int line_size, const DCTELEM *col)
+static inline void idct4col_add(uint8_t *dest, int line_size, const int16_t *col)
 {
     int c0, c1, c2, c3, a0, a1, a2, a3;
 
@@ -156,7 +156,7 @@ static inline void idct4col_add(uint8_t *dest, int line_size, const DCTELEM *col
 #define R2 R_FIX(0.2705980501)
 #define R3 R_FIX(0.5)
 #define R_SHIFT 11
-static inline void idct4row(DCTELEM *row)
+static inline void idct4row(int16_t *row)
 {
     int c0, c1, c2, c3, a0, a1, a2, a3;
 
@@ -174,7 +174,7 @@ static inline void idct4row(DCTELEM *row)
     row[3]= (c0 - c1) >> R_SHIFT;
 }
 
-void ff_simple_idct84_add(uint8_t *dest, int line_size, DCTELEM *block)
+void ff_simple_idct84_add(uint8_t *dest, int line_size, int16_t *block)
 {
     int i;
 
@@ -189,7 +189,7 @@ void ff_simple_idct84_add(uint8_t *dest, int line_size, DCTELEM *block)
     }
 }
 
-void ff_simple_idct48_add(uint8_t *dest, int line_size, DCTELEM *block)
+void ff_simple_idct48_add(uint8_t *dest, int line_size, int16_t *block)
 {
     int i;
 
@@ -204,7 +204,7 @@ void ff_simple_idct48_add(uint8_t *dest, int line_size, DCTELEM *block)
     }
 }
 
-void ff_simple_idct44_add(uint8_t *dest, int line_size, DCTELEM *block)
+void ff_simple_idct44_add(uint8_t *dest, int line_size, int16_t *block)
 {
     int i;
 
@@ -219,7 +219,7 @@ void ff_simple_idct44_add(uint8_t *dest, int line_size, DCTELEM *block)
     }
 }
 
-void ff_prores_idct(DCTELEM *block, const int16_t *qmat)
+void ff_prores_idct(int16_t *block, const int16_t *qmat)
 {
     int i;
 
diff --git a/libavcodec/simple_idct.h b/libavcodec/simple_idct.h
index 64d3c2ac31..3fec5e0087 100644
--- a/libavcodec/simple_idct.h
+++ b/libavcodec/simple_idct.h
@@ -29,30 +29,29 @@
 #define AVCODEC_SIMPLE_IDCT_H
 
 #include <stdint.h>
-#include "dsputil.h"
 
-void ff_simple_idct_put_8(uint8_t *dest, int line_size, DCTELEM *block);
-void ff_simple_idct_add_8(uint8_t *dest, int line_size, DCTELEM *block);
-void ff_simple_idct_8(DCTELEM *block);
+void ff_simple_idct_put_8(uint8_t *dest, int line_size, int16_t *block);
+void ff_simple_idct_add_8(uint8_t *dest, int line_size, int16_t *block);
+void ff_simple_idct_8(int16_t *block);
 
-void ff_simple_idct_put_10(uint8_t *dest, int line_size, DCTELEM *block);
-void ff_simple_idct_add_10(uint8_t *dest, int line_size, DCTELEM *block);
-void ff_simple_idct_10(DCTELEM *block);
+void ff_simple_idct_put_10(uint8_t *dest, int line_size, int16_t *block);
+void ff_simple_idct_add_10(uint8_t *dest, int line_size, int16_t *block);
+void ff_simple_idct_10(int16_t *block);
 /**
  * Special version of ff_simple_idct_10() which does dequantization
  * and scales by a factor of 2 more between the two IDCTs to account
  * for larger scale of input coefficients.
  */
-void ff_prores_idct(DCTELEM *block, const int16_t *qmat);
+void ff_prores_idct(int16_t *block, const int16_t *qmat);
 
 void ff_simple_idct_mmx(int16_t *block);
 void ff_simple_idct_add_mmx(uint8_t *dest, int line_size, int16_t *block);
 void ff_simple_idct_put_mmx(uint8_t *dest, int line_size, int16_t *block);
 
-void ff_simple_idct248_put(uint8_t *dest, int line_size, DCTELEM *block);
+void ff_simple_idct248_put(uint8_t *dest, int line_size, int16_t *block);
 
-void ff_simple_idct84_add(uint8_t *dest, int line_size, DCTELEM *block);
-void ff_simple_idct48_add(uint8_t *dest, int line_size, DCTELEM *block);
-void ff_simple_idct44_add(uint8_t *dest, int line_size, DCTELEM *block);
+void ff_simple_idct84_add(uint8_t *dest, int line_size, int16_t *block);
+void ff_simple_idct48_add(uint8_t *dest, int line_size, int16_t *block);
+void ff_simple_idct44_add(uint8_t *dest, int line_size, int16_t *block);
 
 #endif /* AVCODEC_SIMPLE_IDCT_H */
diff --git a/libavcodec/simple_idct_template.c b/libavcodec/simple_idct_template.c
index b67893c2b2..dabfbda8a0 100644
--- a/libavcodec/simple_idct_template.c
+++ b/libavcodec/simple_idct_template.c
@@ -85,7 +85,7 @@
 
 #endif
 
-static inline void FUNC(idctRowCondDC)(DCTELEM *row, int extra_shift)
+static inline void FUNC(idctRowCondDC)(int16_t *row, int extra_shift)
 {
     int a0, a1, a2, a3, b0, b1, b2, b3;
 
@@ -221,7 +221,7 @@ static inline void FUNC(idctRowCondDC)(DCTELEM *row, int extra_shift)
     } while (0)
 
 static inline void FUNC(idctSparseColPut)(pixel *dest, int line_size,
-                                          DCTELEM *col)
+                                          int16_t *col)
 {
     int a0, a1, a2, a3, b0, b1, b2, b3;
 
@@ -245,7 +245,7 @@ static inline void FUNC(idctSparseColPut)(pixel *dest, int line_size,
 }
 
 static inline void FUNC(idctSparseColAdd)(pixel *dest, int line_size,
-                                          DCTELEM *col)
+                                          int16_t *col)
 {
     int a0, a1, a2, a3, b0, b1, b2, b3;
 
@@ -268,7 +268,7 @@ static inline void FUNC(idctSparseColAdd)(pixel *dest, int line_size,
     dest[0] = av_clip_pixel(dest[0] + ((a0 - b0) >> COL_SHIFT));
 }
 
-static inline void FUNC(idctSparseCol)(DCTELEM *col)
+static inline void FUNC(idctSparseCol)(int16_t *col)
 {
     int a0, a1, a2, a3, b0, b1, b2, b3;
 
@@ -284,7 +284,7 @@ static inline void FUNC(idctSparseCol)(DCTELEM *col)
     col[56] = ((a0 - b0) >> COL_SHIFT);
 }
 
-void FUNC(ff_simple_idct_put)(uint8_t *dest_, int line_size, DCTELEM *block)
+void FUNC(ff_simple_idct_put)(uint8_t *dest_, int line_size, int16_t *block)
 {
     pixel *dest = (pixel *)dest_;
     int i;
@@ -298,7 +298,7 @@ void FUNC(ff_simple_idct_put)(uint8_t *dest_, int line_size, DCTELEM *block)
         FUNC(idctSparseColPut)(dest + i, line_size, block + i);
 }
 
-void FUNC(ff_simple_idct_add)(uint8_t *dest_, int line_size, DCTELEM *block)
+void FUNC(ff_simple_idct_add)(uint8_t *dest_, int line_size, int16_t *block)
 {
     pixel *dest = (pixel *)dest_;
     int i;
@@ -312,7 +312,7 @@ void FUNC(ff_simple_idct_add)(uint8_t *dest_, int line_size, DCTELEM *block)
         FUNC(idctSparseColAdd)(dest + i, line_size, block + i);
 }
 
-void FUNC(ff_simple_idct)(DCTELEM *block)
+void FUNC(ff_simple_idct)(int16_t *block)
 {
     int i;
 
diff --git a/libavcodec/sparc/dsputil_vis.h b/libavcodec/sparc/dsputil_vis.h
index e1cbcb49e3..d7e2e63ab9 100644
--- a/libavcodec/sparc/dsputil_vis.h
+++ b/libavcodec/sparc/dsputil_vis.h
@@ -20,10 +20,9 @@
 #define AVCODEC_SPARC_DSPUTIL_VIS_H
 
 #include <stdint.h>
-#include "libavcodec/dsputil.h"
 
-void ff_simple_idct_put_vis(uint8_t *dest, int line_size, DCTELEM *data);
-void ff_simple_idct_add_vis(uint8_t *dest, int line_size, DCTELEM *data);
-void ff_simple_idct_vis(DCTELEM *data);
+void ff_simple_idct_put_vis(uint8_t *dest, int line_size, int16_t *data);
+void ff_simple_idct_add_vis(uint8_t *dest, int line_size, int16_t *data);
+void ff_simple_idct_vis(int16_t *data);
 
 #endif /* AVCODEC_SPARC_DSPUTIL_VIS_H */
diff --git a/libavcodec/sparc/simple_idct_vis.c b/libavcodec/sparc/simple_idct_vis.c
index 45ca010588..46345a6b25 100644
--- a/libavcodec/sparc/simple_idct_vis.c
+++ b/libavcodec/sparc/simple_idct_vis.c
@@ -388,7 +388,7 @@ static const DECLARE_ALIGNED(8, uint16_t, expand)[4] = {
         "st %%f14, [%12+" dest "] \n\t"\
 
 
-void ff_simple_idct_vis(DCTELEM *data) {
+void ff_simple_idct_vis(int16_t *data) {
     int out1, out2, out3, out4;
     DECLARE_ALIGNED(8, int16_t, temp)[8*8];
 
@@ -428,7 +428,7 @@ void ff_simple_idct_vis(DCTELEM *data) {
     );
 }
 
-void ff_simple_idct_put_vis(uint8_t *dest, int line_size, DCTELEM *data) {
+void ff_simple_idct_put_vis(uint8_t *dest, int line_size, int16_t *data) {
     int out1, out2, out3, out4, out5;
     int r1, r2, r3, r4, r5, r6, r7;
 
@@ -478,7 +478,7 @@ void ff_simple_idct_put_vis(uint8_t *dest, int line_size, DCTELEM *data) {
     );
 }
 
-void ff_simple_idct_add_vis(uint8_t *dest, int line_size, DCTELEM *data) {
+void ff_simple_idct_add_vis(uint8_t *dest, int line_size, int16_t *data) {
     int out1, out2, out3, out4, out5, out6;
     int r1, r2, r3, r4, r5, r6, r7;
 
diff --git a/libavcodec/svq3.c b/libavcodec/svq3.c
index c12c75f51b..05832bfb56 100644
--- a/libavcodec/svq3.c
+++ b/libavcodec/svq3.c
@@ -139,7 +139,7 @@ static const uint32_t svq3_dequant_coeff[32] = {
     61694, 68745, 77615, 89113, 100253, 109366, 126635, 141533
 };
 
-void ff_svq3_luma_dc_dequant_idct_c(DCTELEM *output, DCTELEM *input, int qp)
+void ff_svq3_luma_dc_dequant_idct_c(int16_t *output, int16_t *input, int qp)
 {
     const int qmul = svq3_dequant_coeff[qp];
 #define stride 16
@@ -174,7 +174,7 @@ void ff_svq3_luma_dc_dequant_idct_c(DCTELEM *output, DCTELEM *input, int qp)
 }
 #undef stride
 
-void ff_svq3_add_idct_c(uint8_t *dst, DCTELEM *block,
+void ff_svq3_add_idct_c(uint8_t *dst, int16_t *block,
                         int stride, int qp, int dc)
 {
     const int qmul = svq3_dequant_coeff[qp];
@@ -212,7 +212,7 @@ void ff_svq3_add_idct_c(uint8_t *dst, DCTELEM *block,
     }
 }
 
-static inline int svq3_decode_block(GetBitContext *gb, DCTELEM *block,
+static inline int svq3_decode_block(GetBitContext *gb, int16_t *block,
                                     int index, const int type)
 {
     static const uint8_t *const scan_patterns[4] =
diff --git a/libavcodec/vc1.h b/libavcodec/vc1.h
index 4ed8923469..e32a507a65 100644
--- a/libavcodec/vc1.h
+++ b/libavcodec/vc1.h
@@ -384,7 +384,7 @@ typedef struct VC1Context{
     int bi_type;
     int x8_type;
 
-    DCTELEM (*block)[6][64];
+    int16_t (*block)[6][64];
     int n_allocated_blks, cur_blk_idx, left_blk_idx, topleft_blk_idx, top_blk_idx;
     uint32_t *cbp_base, *cbp;
     uint8_t *is_intra_base, *is_intra;
diff --git a/libavcodec/vc1dec.c b/libavcodec/vc1dec.c
index e0d3d7a200..ee485427aa 100644
--- a/libavcodec/vc1dec.c
+++ b/libavcodec/vc1dec.c
@@ -2556,7 +2556,7 @@ static void vc1_decode_ac_coeff(VC1Context *v, int *last, int *skip,
  * @param coded are AC coeffs present or not
  * @param codingset set of VLC to decode data
  */
-static int vc1_decode_i_block(VC1Context *v, DCTELEM block[64], int n,
+static int vc1_decode_i_block(VC1Context *v, int16_t block[64], int n,
                               int coded, int codingset)
 {
     GetBitContext *gb = &v->s.gb;
@@ -2719,7 +2719,7 @@ not_coded:
  * @param codingset set of VLC to decode data
  * @param mquant quantizer value for this macroblock
  */
-static int vc1_decode_i_block_adv(VC1Context *v, DCTELEM block[64], int n,
+static int vc1_decode_i_block_adv(VC1Context *v, int16_t block[64], int n,
                                   int coded, int codingset, int mquant)
 {
     GetBitContext *gb = &v->s.gb;
@@ -2931,7 +2931,7 @@ static int vc1_decode_i_block_adv(VC1Context *v, DCTELEM block[64], int n,
  * @param mquant block quantizer
  * @param codingset set of VLC to decode data
  */
-static int vc1_decode_intra_block(VC1Context *v, DCTELEM block[64], int n,
+static int vc1_decode_intra_block(VC1Context *v, int16_t block[64], int n,
                                   int coded, int mquant, int codingset)
 {
     GetBitContext *gb = &v->s.gb;
@@ -3141,7 +3141,7 @@ static int vc1_decode_intra_block(VC1Context *v, DCTELEM block[64], int n,
 
 /** Decode P block
  */
-static int vc1_decode_p_block(VC1Context *v, DCTELEM block[64], int n,
+static int vc1_decode_p_block(VC1Context *v, int16_t block[64], int n,
                               int mquant, int ttmb, int first_block,
                               uint8_t *dst, int linesize, int skip_block,
                               int *ttmb_out)
@@ -4523,7 +4523,7 @@ static void vc1_decode_i_blocks_adv(VC1Context *v)
         s->mb_x = 0;
         ff_init_block_index(s);
         for (;s->mb_x < s->mb_width; s->mb_x++) {
-            DCTELEM (*block)[64] = v->block[v->cur_blk_idx];
+            int16_t (*block)[64] = v->block[v->cur_blk_idx];
             ff_update_block_index(s);
             s->dsp.clear_blocks(block[0]);
             mb_pos = s->mb_x + s->mb_y * s->mb_stride;
diff --git a/libavcodec/vc1dsp.c b/libavcodec/vc1dsp.c
index 6963736578..9ddd7c7b96 100644
--- a/libavcodec/vc1dsp.c
+++ b/libavcodec/vc1dsp.c
@@ -80,7 +80,7 @@ static void vc1_h_overlap_c(uint8_t* src, int stride)
     }
 }
 
-static void vc1_v_s_overlap_c(DCTELEM *top,  DCTELEM *bottom)
+static void vc1_v_s_overlap_c(int16_t *top,  int16_t *bottom)
 {
     int i;
     int a, b, c, d;
@@ -106,7 +106,7 @@ static void vc1_v_s_overlap_c(DCTELEM *top,  DCTELEM *bottom)
     }
 }
 
-static void vc1_h_s_overlap_c(DCTELEM *left, DCTELEM *right)
+static void vc1_h_s_overlap_c(int16_t *left, int16_t *right)
 {
     int i;
     int a, b, c, d;
@@ -230,7 +230,7 @@ static void vc1_h_loop_filter16_c(uint8_t *src, int stride, int pq)
 
 /** Do inverse transform on 8x8 block
 */
-static void vc1_inv_trans_8x8_dc_c(uint8_t *dest, int linesize, DCTELEM *block)
+static void vc1_inv_trans_8x8_dc_c(uint8_t *dest, int linesize, int16_t *block)
 {
     int i;
     int dc = block[0];
@@ -249,11 +249,11 @@ static void vc1_inv_trans_8x8_dc_c(uint8_t *dest, int linesize, DCTELEM *block)
     }
 }
 
-static void vc1_inv_trans_8x8_c(DCTELEM block[64])
+static void vc1_inv_trans_8x8_c(int16_t block[64])
 {
     int i;
     register int t1,t2,t3,t4,t5,t6,t7,t8;
-    DCTELEM *src, *dst, temp[64];
+    int16_t *src, *dst, temp[64];
 
     src = block;
     dst = temp;
@@ -320,7 +320,7 @@ static void vc1_inv_trans_8x8_c(DCTELEM block[64])
 
 /** Do inverse transform on 8x4 part of block
 */
-static void vc1_inv_trans_8x4_dc_c(uint8_t *dest, int linesize, DCTELEM *block)
+static void vc1_inv_trans_8x4_dc_c(uint8_t *dest, int linesize, int16_t *block)
 {
     int i;
     int dc = block[0];
@@ -339,11 +339,11 @@ static void vc1_inv_trans_8x4_dc_c(uint8_t *dest, int linesize, DCTELEM *block)
     }
 }
 
-static void vc1_inv_trans_8x4_c(uint8_t *dest, int linesize, DCTELEM *block)
+static void vc1_inv_trans_8x4_c(uint8_t *dest, int linesize, int16_t *block)
 {
     int i;
     register int t1,t2,t3,t4,t5,t6,t7,t8;
-    DCTELEM *src, *dst;
+    int16_t *src, *dst;
 
     src = block;
     dst = block;
@@ -395,7 +395,7 @@ static void vc1_inv_trans_8x4_c(uint8_t *dest, int linesize, DCTELEM *block)
 
 /** Do inverse transform on 4x8 parts of block
 */
-static void vc1_inv_trans_4x8_dc_c(uint8_t *dest, int linesize, DCTELEM *block)
+static void vc1_inv_trans_4x8_dc_c(uint8_t *dest, int linesize, int16_t *block)
 {
     int i;
     int dc = block[0];
@@ -410,11 +410,11 @@ static void vc1_inv_trans_4x8_dc_c(uint8_t *dest, int linesize, DCTELEM *block)
     }
 }
 
-static void vc1_inv_trans_4x8_c(uint8_t *dest, int linesize, DCTELEM *block)
+static void vc1_inv_trans_4x8_c(uint8_t *dest, int linesize, int16_t *block)
 {
     int i;
     register int t1,t2,t3,t4,t5,t6,t7,t8;
-    DCTELEM *src, *dst;
+    int16_t *src, *dst;
 
     src = block;
     dst = block;
@@ -466,7 +466,7 @@ static void vc1_inv_trans_4x8_c(uint8_t *dest, int linesize, DCTELEM *block)
 
 /** Do inverse transform on 4x4 part of block
 */
-static void vc1_inv_trans_4x4_dc_c(uint8_t *dest, int linesize, DCTELEM *block)
+static void vc1_inv_trans_4x4_dc_c(uint8_t *dest, int linesize, int16_t *block)
 {
     int i;
     int dc = block[0];
@@ -481,11 +481,11 @@ static void vc1_inv_trans_4x4_dc_c(uint8_t *dest, int linesize, DCTELEM *block)
     }
 }
 
-static void vc1_inv_trans_4x4_c(uint8_t *dest, int linesize, DCTELEM *block)
+static void vc1_inv_trans_4x4_c(uint8_t *dest, int linesize, int16_t *block)
 {
     int i;
     register int t1,t2,t3,t4;
-    DCTELEM *src, *dst;
+    int16_t *src, *dst;
 
     src = block;
     dst = block;
diff --git a/libavcodec/vc1dsp.h b/libavcodec/vc1dsp.h
index 877581861f..901e46abc9 100644
--- a/libavcodec/vc1dsp.h
+++ b/libavcodec/vc1dsp.h
@@ -32,18 +32,18 @@
 
 typedef struct VC1DSPContext {
     /* vc1 functions */
-    void (*vc1_inv_trans_8x8)(DCTELEM *b);
-    void (*vc1_inv_trans_8x4)(uint8_t *dest, int line_size, DCTELEM *block);
-    void (*vc1_inv_trans_4x8)(uint8_t *dest, int line_size, DCTELEM *block);
-    void (*vc1_inv_trans_4x4)(uint8_t *dest, int line_size, DCTELEM *block);
-    void (*vc1_inv_trans_8x8_dc)(uint8_t *dest, int line_size, DCTELEM *block);
-    void (*vc1_inv_trans_8x4_dc)(uint8_t *dest, int line_size, DCTELEM *block);
-    void (*vc1_inv_trans_4x8_dc)(uint8_t *dest, int line_size, DCTELEM *block);
-    void (*vc1_inv_trans_4x4_dc)(uint8_t *dest, int line_size, DCTELEM *block);
+    void (*vc1_inv_trans_8x8)(int16_t *b);
+    void (*vc1_inv_trans_8x4)(uint8_t *dest, int line_size, int16_t *block);
+    void (*vc1_inv_trans_4x8)(uint8_t *dest, int line_size, int16_t *block);
+    void (*vc1_inv_trans_4x4)(uint8_t *dest, int line_size, int16_t *block);
+    void (*vc1_inv_trans_8x8_dc)(uint8_t *dest, int line_size, int16_t *block);
+    void (*vc1_inv_trans_8x4_dc)(uint8_t *dest, int line_size, int16_t *block);
+    void (*vc1_inv_trans_4x8_dc)(uint8_t *dest, int line_size, int16_t *block);
+    void (*vc1_inv_trans_4x4_dc)(uint8_t *dest, int line_size, int16_t *block);
     void (*vc1_v_overlap)(uint8_t *src, int stride);
     void (*vc1_h_overlap)(uint8_t *src, int stride);
-    void (*vc1_v_s_overlap)(DCTELEM *top,  DCTELEM *bottom);
-    void (*vc1_h_s_overlap)(DCTELEM *left, DCTELEM *right);
+    void (*vc1_v_s_overlap)(int16_t *top,  int16_t *bottom);
+    void (*vc1_h_s_overlap)(int16_t *left, int16_t *right);
     void (*vc1_v_loop_filter4)(uint8_t *src, int stride, int pq);
     void (*vc1_h_loop_filter4)(uint8_t *src, int stride, int pq);
     void (*vc1_v_loop_filter8)(uint8_t *src, int stride, int pq);
diff --git a/libavcodec/vp3.c b/libavcodec/vp3.c
index 11436e5772..792dbb2a61 100644
--- a/libavcodec/vp3.c
+++ b/libavcodec/vp3.c
@@ -142,7 +142,7 @@ typedef struct Vp3DecodeContext {
     DSPContext dsp;
     VideoDSPContext vdsp;
     VP3DSPContext vp3dsp;
-    DECLARE_ALIGNED(16, DCTELEM, block)[64];
+    DECLARE_ALIGNED(16, int16_t, block)[64];
     int flipped_image;
     int last_slice_end;
     int skip_loop_filter;
@@ -926,7 +926,7 @@ static int unpack_vlcs(Vp3DecodeContext *s, GetBitContext *gb,
     int i, j = 0;
     int token;
     int zero_run = 0;
-    DCTELEM coeff = 0;
+    int16_t coeff = 0;
     int bits_to_get;
     int blocks_ended;
     int coeff_i = 0;
@@ -1356,7 +1356,7 @@ static void apply_loop_filter(Vp3DecodeContext *s, int plane, int ystart, int ye
  * for the next block in coding order
  */
 static inline int vp3_dequant(Vp3DecodeContext *s, Vp3Fragment *frag,
-                              int plane, int inter, DCTELEM block[64])
+                              int plane, int inter, int16_t block[64])
 {
     int16_t *dequantizer = s->qmat[frag->qpi][inter][plane];
     uint8_t *perm = s->scantable.permutated;
@@ -1465,7 +1465,7 @@ static void await_reference_row(Vp3DecodeContext *s, Vp3Fragment *fragment, int
 static void render_slice(Vp3DecodeContext *s, int slice)
 {
     int x, y, i, j, fragment;
-    DCTELEM *block = s->block;
+    int16_t *block = s->block;
     int motion_x = 0xdeadbeef, motion_y = 0xdeadbeef;
     int motion_halfpel_index;
     uint8_t *motion_source;
diff --git a/libavcodec/vp3dsp.c b/libavcodec/vp3dsp.c
index df39007649..2c22130642 100644
--- a/libavcodec/vp3dsp.c
+++ b/libavcodec/vp3dsp.c
@@ -213,18 +213,23 @@ static av_always_inline void idct(uint8_t *dst, int stride, int16_t *input, int
     }
 }
 
-static void vp3_idct_put_c(uint8_t *dest/*align 8*/, int line_size, DCTELEM *block/*align 16*/){
+static void vp3_idct_put_c(uint8_t *dest/*align 8*/, int line_size,
+                           int16_t *block/*align 16*/)
+{
     idct(dest, line_size, block, 1);
     memset(block, 0, sizeof(*block) * 64);
 }
 
-static void vp3_idct_add_c(uint8_t *dest/*align 8*/, int line_size, DCTELEM *block/*align 16*/){
+static void vp3_idct_add_c(uint8_t *dest/*align 8*/, int line_size,
+                           int16_t *block/*align 16*/)
+{
     idct(dest, line_size, block, 2);
     memset(block, 0, sizeof(*block) * 64);
 }
 
 static void vp3_idct_dc_add_c(uint8_t *dest/*align 8*/, int line_size,
-                              DCTELEM *block/*align 16*/){
+                              int16_t *block/*align 16*/)
+{
     int i, dc = (block[0] + 15) >> 5;
 
     for(i = 0; i < 8; i++){
diff --git a/libavcodec/vp3dsp.h b/libavcodec/vp3dsp.h
index 35bd95c684..8ffecca780 100644
--- a/libavcodec/vp3dsp.h
+++ b/libavcodec/vp3dsp.h
@@ -21,7 +21,6 @@
 
 #include <stddef.h>
 #include <stdint.h>
-#include "dsputil.h"
 
 typedef struct VP3DSPContext {
     /**
@@ -39,9 +38,9 @@ typedef struct VP3DSPContext {
                                  const uint8_t *b,
                                  ptrdiff_t stride, int h);
 
-    void (*idct_put)(uint8_t *dest, int line_size, DCTELEM *block);
-    void (*idct_add)(uint8_t *dest, int line_size, DCTELEM *block);
-    void (*idct_dc_add)(uint8_t *dest, int line_size, DCTELEM *block);
+    void (*idct_put)(uint8_t *dest, int line_size, int16_t *block);
+    void (*idct_add)(uint8_t *dest, int line_size, int16_t *block);
+    void (*idct_dc_add)(uint8_t *dest, int line_size, int16_t *block);
     void (*v_loop_filter)(uint8_t *src, int stride, int *bounding_values);
     void (*h_loop_filter)(uint8_t *src, int stride, int *bounding_values);
 
diff --git a/libavcodec/vp56.h b/libavcodec/vp56.h
index 3f0c647432..223a202a03 100644
--- a/libavcodec/vp56.h
+++ b/libavcodec/vp56.h
@@ -67,7 +67,7 @@ typedef struct VP56RangeCoder {
 typedef struct VP56RefDc {
     uint8_t not_null_dc;
     VP56Frame ref_frame;
-    DCTELEM dc_coeff;
+    int16_t dc_coeff;
 } VP56RefDc;
 
 typedef struct VP56Macroblock {
@@ -125,12 +125,12 @@ struct vp56_context {
     VP56RefDc *above_blocks;
     VP56RefDc left_block[4];
     int above_block_idx[6];
-    DCTELEM prev_dc[3][3];    /* [plan][ref_frame] */
+    int16_t prev_dc[3][3];    /* [plan][ref_frame] */
 
     /* blocks / macroblock */
     VP56mb mb_type;
     VP56Macroblock *macroblocks;
-    DECLARE_ALIGNED(16, DCTELEM, block_coeff)[6][64];
+    DECLARE_ALIGNED(16, int16_t, block_coeff)[6][64];
 
     /* motion vectors */
     VP56mv mv[6];  /* vectors for each block in MB */
diff --git a/libavcodec/vp8.c b/libavcodec/vp8.c
index a851fbcf07..fc6fd3728f 100644
--- a/libavcodec/vp8.c
+++ b/libavcodec/vp8.c
@@ -761,7 +761,7 @@ void decode_mb_mode(VP8Context *s, VP8Macroblock *mb, int mb_x, int mb_y,
  * @return 0 if no coeffs were decoded
  *         otherwise, the index of the last coeff decoded plus one
  */
-static int decode_block_coeffs_internal(VP56RangeCoder *r, DCTELEM block[16],
+static int decode_block_coeffs_internal(VP56RangeCoder *r, int16_t block[16],
                                         uint8_t probs[16][3][NUM_DCT_TOKENS-1],
                                         int i, uint8_t *token_prob, int16_t qmul[2])
 {
@@ -829,7 +829,7 @@ skip_eob:
  *         otherwise, the index of the last coeff decoded plus one
  */
 static av_always_inline
-int decode_block_coeffs(VP56RangeCoder *c, DCTELEM block[16],
+int decode_block_coeffs(VP56RangeCoder *c, int16_t block[16],
                         uint8_t probs[16][3][NUM_DCT_TOKENS-1],
                         int i, int zero_nhood, int16_t qmul[2])
 {
diff --git a/libavcodec/vp8.h b/libavcodec/vp8.h
index dbbc751e17..19763907d7 100644
--- a/libavcodec/vp8.h
+++ b/libavcodec/vp8.h
@@ -96,8 +96,8 @@ typedef struct VP8Macroblock {
 } VP8Macroblock;
 
 typedef struct VP8ThreadData {
-    DECLARE_ALIGNED(16, DCTELEM, block)[6][4][16];
-    DECLARE_ALIGNED(16, DCTELEM, block_dc)[16];
+    DECLARE_ALIGNED(16, int16_t, block)[6][4][16];
+    DECLARE_ALIGNED(16, int16_t, block_dc)[16];
     /**
      * This is the index plus one of the last non-zero coeff
      * for each of the blocks in the current macroblock.
diff --git a/libavcodec/vp8dsp.c b/libavcodec/vp8dsp.c
index a53643cd26..017278ebea 100644
--- a/libavcodec/vp8dsp.c
+++ b/libavcodec/vp8dsp.c
@@ -29,7 +29,7 @@
 #include "libavutil/common.h"
 
 // TODO: Maybe add dequant
-static void vp8_luma_dc_wht_c(DCTELEM block[4][4][16], DCTELEM dc[16])
+static void vp8_luma_dc_wht_c(int16_t block[4][4][16], int16_t dc[16])
 {
     int i, t0, t1, t2, t3;
 
@@ -62,7 +62,7 @@ static void vp8_luma_dc_wht_c(DCTELEM block[4][4][16], DCTELEM dc[16])
     }
 }
 
-static void vp8_luma_dc_wht_dc_c(DCTELEM block[4][4][16], DCTELEM dc[16])
+static void vp8_luma_dc_wht_dc_c(int16_t block[4][4][16], int16_t dc[16])
 {
     int i, val = (dc[0] + 3) >> 3;
     dc[0] = 0;
@@ -78,10 +78,10 @@ static void vp8_luma_dc_wht_dc_c(DCTELEM block[4][4][16], DCTELEM dc[16])
 #define MUL_20091(a) ((((a)*20091) >> 16) + (a))
 #define MUL_35468(a)  (((a)*35468) >> 16)
 
-static void vp8_idct_add_c(uint8_t *dst, DCTELEM block[16], ptrdiff_t stride)
+static void vp8_idct_add_c(uint8_t *dst, int16_t block[16], ptrdiff_t stride)
 {
     int i, t0, t1, t2, t3;
-    DCTELEM tmp[16];
+    int16_t tmp[16];
 
     for (i = 0; i < 4; i++) {
         t0 = block[0*4+i] + block[2*4+i];
@@ -113,7 +113,7 @@ static void vp8_idct_add_c(uint8_t *dst, DCTELEM block[16], ptrdiff_t stride)
     }
 }
 
-static void vp8_idct_dc_add_c(uint8_t *dst, DCTELEM block[16], ptrdiff_t stride)
+static void vp8_idct_dc_add_c(uint8_t *dst, int16_t block[16], ptrdiff_t stride)
 {
     int i, dc = (block[0] + 4) >> 3;
     block[0] = 0;
@@ -127,7 +127,7 @@ static void vp8_idct_dc_add_c(uint8_t *dst, DCTELEM block[16], ptrdiff_t stride)
     }
 }
 
-static void vp8_idct_dc_add4uv_c(uint8_t *dst, DCTELEM block[4][16], ptrdiff_t stride)
+static void vp8_idct_dc_add4uv_c(uint8_t *dst, int16_t block[4][16], ptrdiff_t stride)
 {
     vp8_idct_dc_add_c(dst+stride*0+0, block[0], stride);
     vp8_idct_dc_add_c(dst+stride*0+4, block[1], stride);
@@ -135,7 +135,7 @@ static void vp8_idct_dc_add4uv_c(uint8_t *dst, DCTELEM block[4][16], ptrdiff_t s
     vp8_idct_dc_add_c(dst+stride*4+4, block[3], stride);
 }
 
-static void vp8_idct_dc_add4y_c(uint8_t *dst, DCTELEM block[4][16], ptrdiff_t stride)
+static void vp8_idct_dc_add4y_c(uint8_t *dst, int16_t block[4][16], ptrdiff_t stride)
 {
     vp8_idct_dc_add_c(dst+ 0, block[0], stride);
     vp8_idct_dc_add_c(dst+ 4, block[1], stride);
diff --git a/libavcodec/vp8dsp.h b/libavcodec/vp8dsp.h
index c4b9aa1a22..6308ff46cf 100644
--- a/libavcodec/vp8dsp.h
+++ b/libavcodec/vp8dsp.h
@@ -27,20 +27,21 @@
 #ifndef AVCODEC_VP8DSP_H
 #define AVCODEC_VP8DSP_H
 
-#include "dsputil.h"
+#include <stddef.h>
+#include <stdint.h>
 
 typedef void (*vp8_mc_func)(uint8_t *dst/*align 8*/, ptrdiff_t dstStride,
                             uint8_t *src/*align 1*/, ptrdiff_t srcStride,
                             int h, int x, int y);
 
 typedef struct VP8DSPContext {
-    void (*vp8_luma_dc_wht)(DCTELEM block[4][4][16], DCTELEM dc[16]);
-    void (*vp8_luma_dc_wht_dc)(DCTELEM block[4][4][16], DCTELEM dc[16]);
-    void (*vp8_idct_add)(uint8_t *dst, DCTELEM block[16], ptrdiff_t stride);
-    void (*vp8_idct_dc_add)(uint8_t *dst, DCTELEM block[16], ptrdiff_t stride);
-    void (*vp8_idct_dc_add4y)(uint8_t *dst, DCTELEM block[4][16],
+    void (*vp8_luma_dc_wht)(int16_t block[4][4][16], int16_t dc[16]);
+    void (*vp8_luma_dc_wht_dc)(int16_t block[4][4][16], int16_t dc[16]);
+    void (*vp8_idct_add)(uint8_t *dst, int16_t block[16], ptrdiff_t stride);
+    void (*vp8_idct_dc_add)(uint8_t *dst, int16_t block[16], ptrdiff_t stride);
+    void (*vp8_idct_dc_add4y)(uint8_t *dst, int16_t block[4][16],
                               ptrdiff_t stride);
-    void (*vp8_idct_dc_add4uv)(uint8_t *dst, DCTELEM block[4][16],
+    void (*vp8_idct_dc_add4uv)(uint8_t *dst, int16_t block[4][16],
                                ptrdiff_t stride);
 
     // loop filter applied to edges between macroblocks
diff --git a/libavcodec/wmv2.c b/libavcodec/wmv2.c
index fc82cb2723..9c94498bf1 100644
--- a/libavcodec/wmv2.c
+++ b/libavcodec/wmv2.c
@@ -48,7 +48,7 @@ av_cold void ff_wmv2_common_init(Wmv2Context * w){
     s->dsp.idct     = NULL;
 }
 
-static void wmv2_add_block(Wmv2Context *w, DCTELEM *block1, uint8_t *dst, int stride, int n){
+static void wmv2_add_block(Wmv2Context *w, int16_t *block1, uint8_t *dst, int stride, int n){
     MpegEncContext * const s= &w->s;
 
   if (s->block_last_index[n] >= 0) {
@@ -72,7 +72,7 @@ static void wmv2_add_block(Wmv2Context *w, DCTELEM *block1, uint8_t *dst, int st
   }
 }
 
-void ff_wmv2_add_mb(MpegEncContext *s, DCTELEM block1[6][64], uint8_t *dest_y, uint8_t *dest_cb, uint8_t *dest_cr){
+void ff_wmv2_add_mb(MpegEncContext *s, int16_t block1[6][64], uint8_t *dest_y, uint8_t *dest_cb, uint8_t *dest_cr){
     Wmv2Context * const w= (Wmv2Context*)s;
 
     wmv2_add_block(w, block1[0], dest_y                    , s->linesize, 0);
diff --git a/libavcodec/wmv2.h b/libavcodec/wmv2.h
index fd7118776d..5ebfae7141 100644
--- a/libavcodec/wmv2.h
+++ b/libavcodec/wmv2.h
@@ -52,7 +52,7 @@ typedef struct Wmv2Context{
     int hshift;
 
     ScanTable abt_scantable[2];
-    DECLARE_ALIGNED(16, DCTELEM, abt_block2)[6][64];
+    DECLARE_ALIGNED(16, int16_t, abt_block2)[6][64];
 }Wmv2Context;
 
 void ff_wmv2_common_init(Wmv2Context * w);
diff --git a/libavcodec/wmv2dec.c b/libavcodec/wmv2dec.c
index dccdd54b6c..5c2df48d14 100644
--- a/libavcodec/wmv2dec.c
+++ b/libavcodec/wmv2dec.c
@@ -291,7 +291,7 @@ static int16_t *wmv2_pred_motion(Wmv2Context *w, int *px, int *py){
     return mot_val;
 }
 
-static inline int wmv2_decode_inter_block(Wmv2Context *w, DCTELEM *block, int n, int cbp){
+static inline int wmv2_decode_inter_block(Wmv2Context *w, int16_t *block, int n, int cbp){
     MpegEncContext * const s= &w->s;
     static const int sub_cbp_table[3]= {2,3,1};
     int sub_cbp;
@@ -331,7 +331,7 @@ static inline int wmv2_decode_inter_block(Wmv2Context *w, DCTELEM *block, int n,
 }
 
 
-int ff_wmv2_decode_mb(MpegEncContext *s, DCTELEM block[6][64])
+int ff_wmv2_decode_mb(MpegEncContext *s, int16_t block[6][64])
 {
     Wmv2Context * const w= (Wmv2Context*)s;
     int cbp, code, i;
diff --git a/libavcodec/wmv2dsp.c b/libavcodec/wmv2dsp.c
index 8e8b006d79..9627442b37 100644
--- a/libavcodec/wmv2dsp.c
+++ b/libavcodec/wmv2dsp.c
@@ -19,6 +19,7 @@
 #include "libavutil/attributes.h"
 #include "libavutil/common.h"
 #include "avcodec.h"
+#include "dsputil.h"
 #include "wmv2dsp.h"
 
 #define W0 2048
@@ -91,7 +92,7 @@ static void wmv2_idct_col(short * b)
     b[8 * 7] = (a0 + a2 - a1 - a5 + (1 << 13)) >> 14;
 }
 
-static void wmv2_idct_add_c(uint8_t *dest, int line_size, DCTELEM *block)
+static void wmv2_idct_add_c(uint8_t *dest, int line_size, int16_t *block)
 {
     int i;
 
@@ -114,7 +115,7 @@ static void wmv2_idct_add_c(uint8_t *dest, int line_size, DCTELEM *block)
     }
 }
 
-static void wmv2_idct_put_c(uint8_t *dest, int line_size, DCTELEM *block)
+static void wmv2_idct_put_c(uint8_t *dest, int line_size, int16_t *block)
 {
     int i;
 
diff --git a/libavcodec/wmv2dsp.h b/libavcodec/wmv2dsp.h
index 242389bdeb..37bee205aa 100644
--- a/libavcodec/wmv2dsp.h
+++ b/libavcodec/wmv2dsp.h
@@ -20,11 +20,10 @@
 #define AVCODEC_WMV2DSP_H
 
 #include <stdint.h>
-#include "dsputil.h"
 
 typedef struct WMV2DSPContext {
-    void (*idct_add)(uint8_t *dest, int line_size, DCTELEM *block);
-    void (*idct_put)(uint8_t *dest, int line_size, DCTELEM *block);
+    void (*idct_add)(uint8_t *dest, int line_size, int16_t *block);
+    void (*idct_put)(uint8_t *dest, int line_size, int16_t *block);
 
     int idct_perm;
 } WMV2DSPContext;
diff --git a/libavcodec/wmv2enc.c b/libavcodec/wmv2enc.c
index e5f3752145..7ec32c60b5 100644
--- a/libavcodec/wmv2enc.c
+++ b/libavcodec/wmv2enc.c
@@ -152,7 +152,7 @@ int ff_wmv2_encode_picture_header(MpegEncContext * s, int picture_number)
  * useless M$ crap features. It is duplicated here in case someone wants
  * to add support for these crap features. */
 void ff_wmv2_encode_mb(MpegEncContext * s,
-                       DCTELEM block[6][64],
+                       int16_t block[6][64],
                        int motion_x, int motion_y)
 {
     Wmv2Context * const w= (Wmv2Context*)s;
diff --git a/libavcodec/x86/dnxhdenc.c b/libavcodec/x86/dnxhdenc.c
index b2ba894bf3..98b8b17e16 100644
--- a/libavcodec/x86/dnxhdenc.c
+++ b/libavcodec/x86/dnxhdenc.c
@@ -26,7 +26,7 @@
 
 #if HAVE_SSE2_INLINE
 
-static void get_pixels_8x4_sym_sse2(DCTELEM *block, const uint8_t *pixels, int line_size)
+static void get_pixels_8x4_sym_sse2(int16_t *block, const uint8_t *pixels, int line_size)
 {
     __asm__ volatile(
         "pxor %%xmm5,      %%xmm5       \n\t"
diff --git a/libavcodec/x86/dsputil_mmx.c b/libavcodec/x86/dsputil_mmx.c
index 858eb90773..2bbe7ff9d3 100644
--- a/libavcodec/x86/dsputil_mmx.c
+++ b/libavcodec/x86/dsputil_mmx.c
@@ -226,10 +226,10 @@ DECLARE_ALIGNED(16, const double, ff_pd_2)[2] = { 2.0, 2.0 };
 /***********************************/
 /* standard MMX */
 
-void ff_put_pixels_clamped_mmx(const DCTELEM *block, uint8_t *pixels,
+void ff_put_pixels_clamped_mmx(const int16_t *block, uint8_t *pixels,
                                int line_size)
 {
-    const DCTELEM *p;
+    const int16_t *p;
     uint8_t *pix;
 
     /* read the pixels */
@@ -301,7 +301,7 @@ void ff_put_pixels_clamped_mmx(const DCTELEM *block, uint8_t *pixels,
     "movq               %%mm3, (%0, %3, 2)  \n\t"           \
     "movq               %%mm4, (%0, %1)     \n\t"
 
-void ff_put_signed_pixels_clamped_mmx(const DCTELEM *block, uint8_t *pixels,
+void ff_put_signed_pixels_clamped_mmx(const int16_t *block, uint8_t *pixels,
                                       int line_size)
 {
     x86_reg line_skip = line_size;
@@ -318,10 +318,10 @@ void ff_put_signed_pixels_clamped_mmx(const DCTELEM *block, uint8_t *pixels,
         : "memory");
 }
 
-void ff_add_pixels_clamped_mmx(const DCTELEM *block, uint8_t *pixels,
+void ff_add_pixels_clamped_mmx(const int16_t *block, uint8_t *pixels,
                                int line_size)
 {
-    const DCTELEM *p;
+    const int16_t *p;
     uint8_t *pix;
     int i;
 
@@ -423,7 +423,7 @@ static void put_pixels16_mmx(uint8_t *block, const uint8_t *pixels,
 }
 
 #define CLEAR_BLOCKS(name, n)                           \
-static void name(DCTELEM *blocks)                       \
+static void name(int16_t *blocks)                       \
 {                                                       \
     __asm__ volatile (                                  \
         "pxor %%mm7, %%mm7              \n\t"           \
@@ -443,7 +443,7 @@ static void name(DCTELEM *blocks)                       \
 CLEAR_BLOCKS(clear_blocks_mmx, 6)
 CLEAR_BLOCKS(clear_block_mmx, 1)
 
-static void clear_block_sse(DCTELEM *block)
+static void clear_block_sse(int16_t *block)
 {
     __asm__ volatile (
         "xorps  %%xmm0, %%xmm0          \n"
@@ -460,7 +460,7 @@ static void clear_block_sse(DCTELEM *block)
     );
 }
 
-static void clear_blocks_sse(DCTELEM *blocks)
+static void clear_blocks_sse(int16_t *blocks)
 {
     __asm__ volatile (
         "xorps  %%xmm0, %%xmm0              \n"
@@ -1882,28 +1882,28 @@ void ff_avg_dirac_pixels32_sse2(uint8_t *dst, const uint8_t *src[5], int stride,
  * converted. */
 #if CONFIG_GPL
 static void ff_libmpeg2mmx_idct_put(uint8_t *dest, int line_size,
-                                    DCTELEM *block)
+                                    int16_t *block)
 {
     ff_mmx_idct(block);
     ff_put_pixels_clamped_mmx(block, dest, line_size);
 }
 
 static void ff_libmpeg2mmx_idct_add(uint8_t *dest, int line_size,
-                                    DCTELEM *block)
+                                    int16_t *block)
 {
     ff_mmx_idct(block);
     ff_add_pixels_clamped_mmx(block, dest, line_size);
 }
 
 static void ff_libmpeg2mmx2_idct_put(uint8_t *dest, int line_size,
-                                     DCTELEM *block)
+                                     int16_t *block)
 {
     ff_mmxext_idct(block);
     ff_put_pixels_clamped_mmx(block, dest, line_size);
 }
 
 static void ff_libmpeg2mmx2_idct_add(uint8_t *dest, int line_size,
-                                     DCTELEM *block)
+                                     int16_t *block)
 {
     ff_mmxext_idct(block);
     ff_add_pixels_clamped_mmx(block, dest, line_size);
diff --git a/libavcodec/x86/dsputil_mmx.h b/libavcodec/x86/dsputil_mmx.h
index a0a44c1275..033a175a15 100644
--- a/libavcodec/x86/dsputil_mmx.h
+++ b/libavcodec/x86/dsputil_mmx.h
@@ -83,9 +83,9 @@ extern const double ff_pd_2[2];
 void ff_dsputilenc_init_mmx(DSPContext* c, AVCodecContext *avctx);
 void ff_dsputil_init_pix_mmx(DSPContext* c, AVCodecContext *avctx);
 
-void ff_add_pixels_clamped_mmx(const DCTELEM *block, uint8_t *pixels, int line_size);
-void ff_put_pixels_clamped_mmx(const DCTELEM *block, uint8_t *pixels, int line_size);
-void ff_put_signed_pixels_clamped_mmx(const DCTELEM *block, uint8_t *pixels, int line_size);
+void ff_add_pixels_clamped_mmx(const int16_t *block, uint8_t *pixels, int line_size);
+void ff_put_pixels_clamped_mmx(const int16_t *block, uint8_t *pixels, int line_size);
+void ff_put_signed_pixels_clamped_mmx(const int16_t *block, uint8_t *pixels, int line_size);
 
 void ff_put_cavs_qpel8_mc00_mmxext(uint8_t *dst, uint8_t *src, int stride);
 void ff_avg_cavs_qpel8_mc00_mmxext(uint8_t *dst, uint8_t *src, int stride);
@@ -100,8 +100,8 @@ void ff_put_rv40_qpel16_mc33_mmx(uint8_t *block, uint8_t *pixels, int line_size)
 void ff_avg_rv40_qpel8_mc33_mmx(uint8_t *block, uint8_t *pixels, int line_size);
 void ff_avg_rv40_qpel16_mc33_mmx(uint8_t *block, uint8_t *pixels, int line_size);
 
-void ff_mmx_idct(DCTELEM *block);
-void ff_mmxext_idct(DCTELEM *block);
+void ff_mmx_idct(int16_t *block);
+void ff_mmxext_idct(int16_t *block);
 
 
 void ff_deinterlace_line_mmx(uint8_t *dst,
diff --git a/libavcodec/x86/dsputilenc.asm b/libavcodec/x86/dsputilenc.asm
index 55eacf5b96..1839bee24a 100644
--- a/libavcodec/x86/dsputilenc.asm
+++ b/libavcodec/x86/dsputilenc.asm
@@ -335,7 +335,7 @@ cglobal sse16, 5, 5, 8
     RET
 
 INIT_MMX mmx
-; get_pixels_mmx(DCTELEM *block, const uint8_t *pixels, int line_size)
+; get_pixels_mmx(int16_t *block, const uint8_t *pixels, int line_size)
 cglobal get_pixels, 3,4
     movsxdifnidn r2, r2d
     add          r0, 128
@@ -392,7 +392,7 @@ cglobal get_pixels, 3, 4
     RET
 
 INIT_MMX mmx
-; diff_pixels_mmx(DCTELEM *block, const uint8_t *s1, const unint8_t *s2, stride)
+; diff_pixels_mmx(int16_t *block, const uint8_t *s1, const unint8_t *s2, stride)
 cglobal diff_pixels, 4,5
     movsxdifnidn r3, r3d
     pxor         m7, m7
diff --git a/libavcodec/x86/dsputilenc_mmx.c b/libavcodec/x86/dsputilenc_mmx.c
index 8a50a26ce2..5ac3c1c10a 100644
--- a/libavcodec/x86/dsputilenc_mmx.c
+++ b/libavcodec/x86/dsputilenc_mmx.c
@@ -30,9 +30,9 @@
 #include "libavcodec/mathops.h"
 #include "dsputil_mmx.h"
 
-void ff_get_pixels_mmx(DCTELEM *block, const uint8_t *pixels, int line_size);
-void ff_get_pixels_sse2(DCTELEM *block, const uint8_t *pixels, int line_size);
-void ff_diff_pixels_mmx(DCTELEM *block, const uint8_t *s1, const uint8_t *s2, int stride);
+void ff_get_pixels_mmx(int16_t *block, const uint8_t *pixels, int line_size);
+void ff_get_pixels_sse2(int16_t *block, const uint8_t *pixels, int line_size);
+void ff_diff_pixels_mmx(int16_t *block, const uint8_t *s1, const uint8_t *s2, int stride);
 int ff_pix_sum16_mmx(uint8_t * pix, int line_size);
 int ff_pix_norm1_mmx(uint8_t *pix, int line_size);
 
@@ -798,7 +798,7 @@ static void sub_hfyu_median_prediction_mmxext(uint8_t *dst, const uint8_t *src1,
     HSUM(%%xmm0, %%xmm1, %0)
 
 #define DCT_SAD_FUNC(cpu) \
-static int sum_abs_dctelem_##cpu(DCTELEM *block){\
+static int sum_abs_dctelem_##cpu(int16_t *block){\
     int sum;\
     __asm__ volatile(\
         DCT_SAD\
diff --git a/libavcodec/x86/h264_idct.asm b/libavcodec/x86/h264_idct.asm
index d0ac0fe16d..8fef7b8199 100644
--- a/libavcodec/x86/h264_idct.asm
+++ b/libavcodec/x86/h264_idct.asm
@@ -302,7 +302,7 @@ cglobal h264_idct8_dc_add_8, 3, 3, 0
 
 INIT_MMX mmx
 ; ff_h264_idct_add16_mmx(uint8_t *dst, const int *block_offset,
-;             DCTELEM *block, int stride, const uint8_t nnzc[6*8])
+;             int16_t *block, int stride, const uint8_t nnzc[6*8])
 cglobal h264_idct_add16_8, 5, 7 + npicregs, 0, dst, block_offset, block, stride, nnzc, cntr, coeff, picreg
     xor          r5, r5
 %ifdef PIC
@@ -324,7 +324,7 @@ cglobal h264_idct_add16_8, 5, 7 + npicregs, 0, dst, block_offset, block, stride,
     REP_RET
 
 ; ff_h264_idct8_add4_mmx(uint8_t *dst, const int *block_offset,
-;                        DCTELEM *block, int stride, const uint8_t nnzc[6*8])
+;                        int16_t *block, int stride, const uint8_t nnzc[6*8])
 cglobal h264_idct8_add4_8, 5, 7 + npicregs, 0, dst, block_offset, block, stride, nnzc, cntr, coeff, picreg
     %assign pad 128+4-(stack_offset&7)
     SUB         rsp, pad
@@ -357,7 +357,7 @@ cglobal h264_idct8_add4_8, 5, 7 + npicregs, 0, dst, block_offset, block, stride,
 
 INIT_MMX mmxext
 ; ff_h264_idct_add16_mmxext(uint8_t *dst, const int *block_offset,
-;                           DCTELEM *block, int stride, const uint8_t nnzc[6*8])
+;                           int16_t *block, int stride, const uint8_t nnzc[6*8])
 cglobal h264_idct_add16_8, 5, 8 + npicregs, 0, dst1, block_offset, block, stride, nnzc, cntr, coeff, dst2, picreg
     xor          r5, r5
 %ifdef PIC
@@ -402,7 +402,7 @@ cglobal h264_idct_add16_8, 5, 8 + npicregs, 0, dst1, block_offset, block, stride
 
 INIT_MMX mmx
 ; ff_h264_idct_add16intra_mmx(uint8_t *dst, const int *block_offset,
-;                             DCTELEM *block, int stride, const uint8_t nnzc[6*8])
+;                             int16_t *block, int stride, const uint8_t nnzc[6*8])
 cglobal h264_idct_add16intra_8, 5, 7 + npicregs, 0, dst, block_offset, block, stride, nnzc, cntr, coeff, picreg
     xor          r5, r5
 %ifdef PIC
@@ -426,7 +426,7 @@ cglobal h264_idct_add16intra_8, 5, 7 + npicregs, 0, dst, block_offset, block, st
 
 INIT_MMX mmxext
 ; ff_h264_idct_add16intra_mmxext(uint8_t *dst, const int *block_offset,
-;                                DCTELEM *block, int stride,
+;                                int16_t *block, int stride,
 ;                                const uint8_t nnzc[6*8])
 cglobal h264_idct_add16intra_8, 5, 8 + npicregs, 0, dst1, block_offset, block, stride, nnzc, cntr, coeff, dst2, picreg
     xor          r5, r5
@@ -469,7 +469,7 @@ cglobal h264_idct_add16intra_8, 5, 8 + npicregs, 0, dst1, block_offset, block, s
     REP_RET
 
 ; ff_h264_idct8_add4_mmxext(uint8_t *dst, const int *block_offset,
-;                           DCTELEM *block, int stride,
+;                           int16_t *block, int stride,
 ;                           const uint8_t nnzc[6*8])
 cglobal h264_idct8_add4_8, 5, 8 + npicregs, 0, dst1, block_offset, block, stride, nnzc, cntr, coeff, dst2, picreg
     %assign pad 128+4-(stack_offset&7)
@@ -530,7 +530,7 @@ cglobal h264_idct8_add4_8, 5, 8 + npicregs, 0, dst1, block_offset, block, stride
 
 INIT_XMM sse2
 ; ff_h264_idct8_add4_sse2(uint8_t *dst, const int *block_offset,
-;                         DCTELEM *block, int stride, const uint8_t nnzc[6*8])
+;                         int16_t *block, int stride, const uint8_t nnzc[6*8])
 cglobal h264_idct8_add4_8, 5, 8 + npicregs, 10, dst1, block_offset, block, stride, nnzc, cntr, coeff, dst2, picreg
     xor          r5, r5
 %ifdef PIC
@@ -605,7 +605,7 @@ h264_idct_add8_mmx_plane:
     rep ret
 
 ; ff_h264_idct_add8_mmx(uint8_t **dest, const int *block_offset,
-;                       DCTELEM *block, int stride, const uint8_t nnzc[6*8])
+;                       int16_t *block, int stride, const uint8_t nnzc[6*8])
 cglobal h264_idct_add8_8, 5, 8 + npicregs, 0, dst1, block_offset, block, stride, nnzc, cntr, coeff, dst2, picreg
     mov          r5, 16
     add          r2, 512
@@ -669,7 +669,7 @@ h264_idct_add8_mmxext_plane:
 
 INIT_MMX mmxext
 ; ff_h264_idct_add8_mmxext(uint8_t **dest, const int *block_offset,
-;                          DCTELEM *block, int stride, const uint8_t nnzc[6*8])
+;                          int16_t *block, int stride, const uint8_t nnzc[6*8])
 cglobal h264_idct_add8_8, 5, 8 + npicregs, 0, dst1, block_offset, block, stride, nnzc, cntr, coeff, dst2, picreg
     mov          r5, 16
     add          r2, 512
@@ -746,7 +746,7 @@ h264_add8x4_idct_sse2:
 %endmacro
 
 ; ff_h264_idct_add16_sse2(uint8_t *dst, const int *block_offset,
-;                         DCTELEM *block, int stride, const uint8_t nnzc[6*8])
+;                         int16_t *block, int stride, const uint8_t nnzc[6*8])
 cglobal h264_idct_add16_8, 5, 5 + ARCH_X86_64, 8
 %if ARCH_X86_64
     mov         r5, r0
@@ -793,7 +793,7 @@ cglobal h264_idct_add16_8, 5, 5 + ARCH_X86_64, 8
 %endmacro
 
 ; ff_h264_idct_add16intra_sse2(uint8_t *dst, const int *block_offset,
-;                              DCTELEM *block, int stride, const uint8_t nnzc[6*8])
+;                              int16_t *block, int stride, const uint8_t nnzc[6*8])
 cglobal h264_idct_add16intra_8, 5, 7 + ARCH_X86_64, 8
 %if ARCH_X86_64
     mov         r7, r0
@@ -844,7 +844,7 @@ cglobal h264_idct_add16intra_8, 5, 7 + ARCH_X86_64, 8
 %endmacro
 
 ; ff_h264_idct_add8_sse2(uint8_t **dest, const int *block_offset,
-;                        DCTELEM *block, int stride, const uint8_t nnzc[6*8])
+;                        int16_t *block, int stride, const uint8_t nnzc[6*8])
 cglobal h264_idct_add8_8, 5, 7 + ARCH_X86_64, 8
     add          r2, 512
 %if ARCH_X86_64
@@ -861,7 +861,7 @@ cglobal h264_idct_add8_8, 5, 7 + ARCH_X86_64, 8
     add8_sse2_cycle 3, 0x64
     RET
 
-;void ff_h264_luma_dc_dequant_idct_mmx(DCTELEM *output, DCTELEM *input, int qmul)
+;void ff_h264_luma_dc_dequant_idct_mmx(int16_t *output, int16_t *input, int qmul)
 
 %macro WALSH4_1D 5
     SUMSUB_BADC w, %4, %3, %2, %1, %5
diff --git a/libavcodec/x86/h264_intrapred_init.c b/libavcodec/x86/h264_intrapred_init.c
index deededaddb..015cc6ff19 100644
--- a/libavcodec/x86/h264_intrapred_init.c
+++ b/libavcodec/x86/h264_intrapred_init.c
@@ -20,6 +20,7 @@
 
 #include "libavutil/cpu.h"
 #include "libavutil/x86/cpu.h"
+#include "libavcodec/avcodec.h"
 #include "libavcodec/h264pred.h"
 
 #define PRED4x4(TYPE, DEPTH, OPT) \
diff --git a/libavcodec/x86/h264dsp_init.c b/libavcodec/x86/h264dsp_init.c
index 65d300fe4d..a0c22eb186 100644
--- a/libavcodec/x86/h264dsp_init.c
+++ b/libavcodec/x86/h264dsp_init.c
@@ -48,7 +48,7 @@ IDCT_ADD_FUNC(8, 10, avx)
 #define IDCT_ADD_REP_FUNC(NUM, REP, DEPTH, OPT)                         \
 void ff_h264_idct ## NUM ## _add ## REP ## _ ## DEPTH ## _ ## OPT       \
     (uint8_t *dst, const int *block_offset,                             \
-     DCTELEM *block, int stride, const uint8_t nnzc[6 * 8]);
+     int16_t *block, int stride, const uint8_t nnzc[6 * 8]);
 
 IDCT_ADD_REP_FUNC(8, 4, 8, mmx)
 IDCT_ADD_REP_FUNC(8, 4, 8, mmxext)
@@ -70,7 +70,7 @@ IDCT_ADD_REP_FUNC(, 16intra, 10, avx)
 #define IDCT_ADD_REP_FUNC2(NUM, REP, DEPTH, OPT)                      \
 void ff_h264_idct ## NUM ## _add ## REP ## _ ## DEPTH ## _ ## OPT     \
     (uint8_t **dst, const int *block_offset,                          \
-     DCTELEM *block, int stride, const uint8_t nnzc[6 * 8]);
+     int16_t *block, int stride, const uint8_t nnzc[6 * 8]);
 
 IDCT_ADD_REP_FUNC2(, 8, 8, mmx)
 IDCT_ADD_REP_FUNC2(, 8, 8, mmxext)
@@ -78,8 +78,8 @@ IDCT_ADD_REP_FUNC2(, 8, 8, sse2)
 IDCT_ADD_REP_FUNC2(, 8, 10, sse2)
 IDCT_ADD_REP_FUNC2(, 8, 10, avx)
 
-void ff_h264_luma_dc_dequant_idct_mmx(DCTELEM *output, DCTELEM *input, int qmul);
-void ff_h264_luma_dc_dequant_idct_sse2(DCTELEM *output, DCTELEM *input, int qmul);
+void ff_h264_luma_dc_dequant_idct_mmx(int16_t *output, int16_t *input, int qmul);
+void ff_h264_luma_dc_dequant_idct_sse2(int16_t *output, int16_t *input, int qmul);
 
 /***********************************/
 /* deblocking */
diff --git a/libavcodec/x86/idct_mmx_xvid.c b/libavcodec/x86/idct_mmx_xvid.c
index b62865e679..5e9f405452 100644
--- a/libavcodec/x86/idct_mmx_xvid.c
+++ b/libavcodec/x86/idct_mmx_xvid.c
@@ -531,25 +531,25 @@ __asm__ volatile(
     :: "r"(block), "r"(rounder_0), "r"(tab_i_04_xmm), "r"(tg_1_16));
 }
 
-void ff_idct_xvid_mmx_put(uint8_t *dest, int line_size, DCTELEM *block)
+void ff_idct_xvid_mmx_put(uint8_t *dest, int line_size, int16_t *block)
 {
     ff_idct_xvid_mmx(block);
     ff_put_pixels_clamped_mmx(block, dest, line_size);
 }
 
-void ff_idct_xvid_mmx_add(uint8_t *dest, int line_size, DCTELEM *block)
+void ff_idct_xvid_mmx_add(uint8_t *dest, int line_size, int16_t *block)
 {
     ff_idct_xvid_mmx(block);
     ff_add_pixels_clamped_mmx(block, dest, line_size);
 }
 
-void ff_idct_xvid_mmxext_put(uint8_t *dest, int line_size, DCTELEM *block)
+void ff_idct_xvid_mmxext_put(uint8_t *dest, int line_size, int16_t *block)
 {
     ff_idct_xvid_mmxext(block);
     ff_put_pixels_clamped_mmx(block, dest, line_size);
 }
 
-void ff_idct_xvid_mmxext_add(uint8_t *dest, int line_size, DCTELEM *block)
+void ff_idct_xvid_mmxext_add(uint8_t *dest, int line_size, int16_t *block)
 {
     ff_idct_xvid_mmxext(block);
     ff_add_pixels_clamped_mmx(block, dest, line_size);
diff --git a/libavcodec/x86/idct_xvid.h b/libavcodec/x86/idct_xvid.h
index a584fba377..7a2847b864 100644
--- a/libavcodec/x86/idct_xvid.h
+++ b/libavcodec/x86/idct_xvid.h
@@ -28,15 +28,13 @@
 
 #include <stdint.h>
 
-#include "libavcodec/dsputil.h"
-
 void ff_idct_xvid_mmx(short *block);
-void ff_idct_xvid_mmx_put(uint8_t *dest, int line_size, DCTELEM *block);
-void ff_idct_xvid_mmx_add(uint8_t *dest, int line_size, DCTELEM *block);
+void ff_idct_xvid_mmx_put(uint8_t *dest, int line_size, int16_t *block);
+void ff_idct_xvid_mmx_add(uint8_t *dest, int line_size, int16_t *block);
 
 void ff_idct_xvid_mmxext(short *block);
-void ff_idct_xvid_mmxext_put(uint8_t *dest, int line_size, DCTELEM *block);
-void ff_idct_xvid_mmxext_add(uint8_t *dest, int line_size, DCTELEM *block);
+void ff_idct_xvid_mmxext_put(uint8_t *dest, int line_size, int16_t *block);
+void ff_idct_xvid_mmxext_add(uint8_t *dest, int line_size, int16_t *block);
 
 void ff_idct_xvid_sse2(short *block);
 void ff_idct_xvid_sse2_put(uint8_t *dest, int line_size, short *block);
diff --git a/libavcodec/x86/mpegvideo.c b/libavcodec/x86/mpegvideo.c
index 9870ed7d51..f4bf05ffdd 100644
--- a/libavcodec/x86/mpegvideo.c
+++ b/libavcodec/x86/mpegvideo.c
@@ -29,7 +29,7 @@
 #if HAVE_INLINE_ASM
 
 static void dct_unquantize_h263_intra_mmx(MpegEncContext *s,
-                                  DCTELEM *block, int n, int qscale)
+                                  int16_t *block, int n, int qscale)
 {
     x86_reg level, qmul, qadd, nCoeffs;
 
@@ -104,7 +104,7 @@ __asm__ volatile(
 
 
 static void dct_unquantize_h263_inter_mmx(MpegEncContext *s,
-                                  DCTELEM *block, int n, int qscale)
+                                  int16_t *block, int n, int qscale)
 {
     x86_reg qmul, qadd, nCoeffs;
 
@@ -187,7 +187,7 @@ __asm__ volatile(
  high3 += tlow1
 */
 static void dct_unquantize_mpeg1_intra_mmx(MpegEncContext *s,
-                                     DCTELEM *block, int n, int qscale)
+                                     int16_t *block, int n, int qscale)
 {
     x86_reg nCoeffs;
     const uint16_t *quant_matrix;
@@ -256,7 +256,7 @@ __asm__ volatile(
 }
 
 static void dct_unquantize_mpeg1_inter_mmx(MpegEncContext *s,
-                                     DCTELEM *block, int n, int qscale)
+                                     int16_t *block, int n, int qscale)
 {
     x86_reg nCoeffs;
     const uint16_t *quant_matrix;
@@ -322,7 +322,7 @@ __asm__ volatile(
 }
 
 static void dct_unquantize_mpeg2_intra_mmx(MpegEncContext *s,
-                                     DCTELEM *block, int n, int qscale)
+                                     int16_t *block, int n, int qscale)
 {
     x86_reg nCoeffs;
     const uint16_t *quant_matrix;
@@ -388,7 +388,7 @@ __asm__ volatile(
 }
 
 static void dct_unquantize_mpeg2_inter_mmx(MpegEncContext *s,
-                                     DCTELEM *block, int n, int qscale)
+                                     int16_t *block, int n, int qscale)
 {
     x86_reg nCoeffs;
     const uint16_t *quant_matrix;
@@ -464,7 +464,7 @@ __asm__ volatile(
         );
 }
 
-static void  denoise_dct_mmx(MpegEncContext *s, DCTELEM *block){
+static void  denoise_dct_mmx(MpegEncContext *s, int16_t *block){
     const int intra= s->mb_intra;
     int *sum= s->dct_error_sum[intra];
     uint16_t *offset= s->dct_offset[intra];
@@ -518,7 +518,7 @@ static void  denoise_dct_mmx(MpegEncContext *s, DCTELEM *block){
     );
 }
 
-static void  denoise_dct_sse2(MpegEncContext *s, DCTELEM *block){
+static void  denoise_dct_sse2(MpegEncContext *s, int16_t *block){
     const int intra= s->mb_intra;
     int *sum= s->dct_error_sum[intra];
     uint16_t *offset= s->dct_offset[intra];
diff --git a/libavcodec/x86/mpegvideoenc_template.c b/libavcodec/x86/mpegvideoenc_template.c
index 47c3b43f94..1e0505ea3c 100644
--- a/libavcodec/x86/mpegvideoenc_template.c
+++ b/libavcodec/x86/mpegvideoenc_template.c
@@ -92,7 +92,7 @@
 #endif
 
 static int RENAME(dct_quantize)(MpegEncContext *s,
-                            DCTELEM *block, int n,
+                            int16_t *block, int n,
                             int qscale, int *overflow)
 {
     x86_reg last_non_zero_p1;
diff --git a/libavcodec/x86/proresdsp.asm b/libavcodec/x86/proresdsp.asm
index 094f872bc8..aedacc2cdc 100644
--- a/libavcodec/x86/proresdsp.asm
+++ b/libavcodec/x86/proresdsp.asm
@@ -232,7 +232,7 @@ section .text align=16
 %endmacro
 
 ; void prores_idct_put_10_<opt>(uint8_t *pixels, int stride,
-;                               DCTELEM *block, const int16_t *qmat);
+;                               int16_t *block, const int16_t *qmat);
 %macro idct_put_fn 1
 cglobal prores_idct_put_10, 4, 4, %1
     movsxd      r1,  r1d
diff --git a/libavcodec/x86/proresdsp_init.c b/libavcodec/x86/proresdsp_init.c
index 21ce0986d2..91ff257564 100644
--- a/libavcodec/x86/proresdsp_init.c
+++ b/libavcodec/x86/proresdsp_init.c
@@ -21,14 +21,15 @@
  */
 
 #include "libavutil/x86/cpu.h"
+#include "libavcodec/dsputil.h"
 #include "libavcodec/proresdsp.h"
 
 void ff_prores_idct_put_10_sse2(uint16_t *dst, int linesize,
-                                DCTELEM *block, const int16_t *qmat);
+                                int16_t *block, const int16_t *qmat);
 void ff_prores_idct_put_10_sse4(uint16_t *dst, int linesize,
-                                DCTELEM *block, const int16_t *qmat);
+                                int16_t *block, const int16_t *qmat);
 void ff_prores_idct_put_10_avx (uint16_t *dst, int linesize,
-                                DCTELEM *block, const int16_t *qmat);
+                                int16_t *block, const int16_t *qmat);
 
 void ff_proresdsp_x86_init(ProresDSPContext *dsp, AVCodecContext *avctx)
 {
diff --git a/libavcodec/x86/rv34dsp.asm b/libavcodec/x86/rv34dsp.asm
index c099ac5b1f..4d9c35b600 100644
--- a/libavcodec/x86/rv34dsp.asm
+++ b/libavcodec/x86/rv34dsp.asm
@@ -133,7 +133,7 @@ cglobal rv34_idct_dc_add, 3, 3
     mova        mm5, [pd_512]           ; 0x200
 %endmacro
 
-; ff_rv34_idct_add_mmxext(uint8_t *dst, ptrdiff_t stride, DCTELEM *block);
+; ff_rv34_idct_add_mmxext(uint8_t *dst, ptrdiff_t stride, int16_t *block);
 %macro COL_TRANSFORM  4
     pshufw      mm3, %2, 0xDD        ; col. 1,3,1,3
     pshufw       %2, %2, 0x88        ; col. 0,2,0,2
diff --git a/libavcodec/x86/rv34dsp_init.c b/libavcodec/x86/rv34dsp_init.c
index 6b6cf914ea..a93096791b 100644
--- a/libavcodec/x86/rv34dsp_init.c
+++ b/libavcodec/x86/rv34dsp_init.c
@@ -25,11 +25,11 @@
 #include "libavcodec/dsputil.h"
 #include "libavcodec/rv34dsp.h"
 
-void ff_rv34_idct_dc_mmxext(DCTELEM *block);
-void ff_rv34_idct_dc_noround_mmxext(DCTELEM *block);
+void ff_rv34_idct_dc_mmxext(int16_t *block);
+void ff_rv34_idct_dc_noround_mmxext(int16_t *block);
 void ff_rv34_idct_dc_add_mmx(uint8_t *dst, ptrdiff_t stride, int dc);
 void ff_rv34_idct_dc_add_sse4(uint8_t *dst, ptrdiff_t stride, int dc);
-void ff_rv34_idct_add_mmxext(uint8_t *dst, ptrdiff_t stride, DCTELEM *block);
+void ff_rv34_idct_add_mmxext(uint8_t *dst, ptrdiff_t stride, int16_t *block);
 
 av_cold void ff_rv34dsp_init_x86(RV34DSPContext* c, DSPContext *dsp)
 {
diff --git a/libavcodec/x86/simple_idct.c b/libavcodec/x86/simple_idct.c
index c514d755ee..f9fd7dc3c9 100644
--- a/libavcodec/x86/simple_idct.c
+++ b/libavcodec/x86/simple_idct.c
@@ -1154,12 +1154,12 @@ void ff_simple_idct_mmx(int16_t *block)
 
 //FIXME merge add/put into the idct
 
-void ff_simple_idct_put_mmx(uint8_t *dest, int line_size, DCTELEM *block)
+void ff_simple_idct_put_mmx(uint8_t *dest, int line_size, int16_t *block)
 {
     idct(block);
     ff_put_pixels_clamped_mmx(block, dest, line_size);
 }
-void ff_simple_idct_add_mmx(uint8_t *dest, int line_size, DCTELEM *block)
+void ff_simple_idct_add_mmx(uint8_t *dest, int line_size, int16_t *block)
 {
     idct(block);
     ff_add_pixels_clamped_mmx(block, dest, line_size);
diff --git a/libavcodec/x86/vc1dsp_mmx.c b/libavcodec/x86/vc1dsp_mmx.c
index 6ce65e4e16..a7695357d9 100644
--- a/libavcodec/x86/vc1dsp_mmx.c
+++ b/libavcodec/x86/vc1dsp_mmx.c
@@ -493,7 +493,7 @@ DECLARE_FUNCTION(3, 2)
 DECLARE_FUNCTION(3, 3)
 
 static void vc1_inv_trans_4x4_dc_mmxext(uint8_t *dest, int linesize,
-                                        DCTELEM *block)
+                                        int16_t *block)
 {
     int dc = block[0];
     dc = (17 * dc +  4) >> 3;
@@ -532,7 +532,7 @@ static void vc1_inv_trans_4x4_dc_mmxext(uint8_t *dest, int linesize,
 }
 
 static void vc1_inv_trans_4x8_dc_mmxext(uint8_t *dest, int linesize,
-                                        DCTELEM *block)
+                                        int16_t *block)
 {
     int dc = block[0];
     dc = (17 * dc +  4) >> 3;
@@ -594,7 +594,7 @@ static void vc1_inv_trans_4x8_dc_mmxext(uint8_t *dest, int linesize,
 }
 
 static void vc1_inv_trans_8x4_dc_mmxext(uint8_t *dest, int linesize,
-                                        DCTELEM *block)
+                                        int16_t *block)
 {
     int dc = block[0];
     dc = ( 3 * dc +  1) >> 1;
@@ -633,7 +633,7 @@ static void vc1_inv_trans_8x4_dc_mmxext(uint8_t *dest, int linesize,
 }
 
 static void vc1_inv_trans_8x8_dc_mmxext(uint8_t *dest, int linesize,
-                                        DCTELEM *block)
+                                        int16_t *block)
 {
     int dc = block[0];
     dc = (3 * dc +  1) >> 1;
diff --git a/libavcodec/x86/vp3dsp_init.c b/libavcodec/x86/vp3dsp_init.c
index e425f503ac..1df570ec0b 100644
--- a/libavcodec/x86/vp3dsp_init.c
+++ b/libavcodec/x86/vp3dsp_init.c
@@ -25,17 +25,18 @@
 #include "libavutil/x86/cpu.h"
 #include "libavutil/x86/asm.h"
 #include "libavcodec/avcodec.h"
+#include "libavcodec/dsputil.h"
 #include "libavcodec/vp3dsp.h"
 #include "config.h"
 
-void ff_vp3_idct_put_mmx(uint8_t *dest, int line_size, DCTELEM *block);
-void ff_vp3_idct_add_mmx(uint8_t *dest, int line_size, DCTELEM *block);
+void ff_vp3_idct_put_mmx(uint8_t *dest, int line_size, int16_t *block);
+void ff_vp3_idct_add_mmx(uint8_t *dest, int line_size, int16_t *block);
 
-void ff_vp3_idct_put_sse2(uint8_t *dest, int line_size, DCTELEM *block);
-void ff_vp3_idct_add_sse2(uint8_t *dest, int line_size, DCTELEM *block);
+void ff_vp3_idct_put_sse2(uint8_t *dest, int line_size, int16_t *block);
+void ff_vp3_idct_add_sse2(uint8_t *dest, int line_size, int16_t *block);
 
 void ff_vp3_idct_dc_add_mmxext(uint8_t *dest, int line_size,
-                               DCTELEM *block);
+                               int16_t *block);
 
 void ff_vp3_v_loop_filter_mmxext(uint8_t *src, int stride,
                                  int *bounding_values);
diff --git a/libavcodec/x86/vp8dsp.asm b/libavcodec/x86/vp8dsp.asm
index 84805fdc66..c5fb4555d3 100644
--- a/libavcodec/x86/vp8dsp.asm
+++ b/libavcodec/x86/vp8dsp.asm
@@ -899,7 +899,7 @@ cglobal put_vp8_pixels16, 5, 5, 2, dst, dststride, src, srcstride, height
     REP_RET
 
 ;-----------------------------------------------------------------------------
-; void vp8_idct_dc_add_<opt>(uint8_t *dst, DCTELEM block[16], int stride);
+; void vp8_idct_dc_add_<opt>(uint8_t *dst, int16_t block[16], int stride);
 ;-----------------------------------------------------------------------------
 
 %macro ADD_DC 4
@@ -977,7 +977,7 @@ cglobal vp8_idct_dc_add, 3, 3, 6, dst, block, stride
     RET
 
 ;-----------------------------------------------------------------------------
-; void vp8_idct_dc_add4y_<opt>(uint8_t *dst, DCTELEM block[4][16], int stride);
+; void vp8_idct_dc_add4y_<opt>(uint8_t *dst, int16_t block[4][16], int stride);
 ;-----------------------------------------------------------------------------
 
 %if ARCH_X86_32
@@ -1050,7 +1050,7 @@ cglobal vp8_idct_dc_add4y, 3, 3, 6, dst, block, stride
     RET
 
 ;-----------------------------------------------------------------------------
-; void vp8_idct_dc_add4uv_<opt>(uint8_t *dst, DCTELEM block[4][16], int stride);
+; void vp8_idct_dc_add4uv_<opt>(uint8_t *dst, int16_t block[4][16], int stride);
 ;-----------------------------------------------------------------------------
 
 INIT_MMX mmx
@@ -1092,7 +1092,7 @@ cglobal vp8_idct_dc_add4uv, 3, 3, 0, dst, block, stride
     RET
 
 ;-----------------------------------------------------------------------------
-; void vp8_idct_add_<opt>(uint8_t *dst, DCTELEM block[16], int stride);
+; void vp8_idct_add_<opt>(uint8_t *dst, int16_t block[16], int stride);
 ;-----------------------------------------------------------------------------
 
 ; calculate %1=mul_35468(%1)-mul_20091(%2); %2=mul_20091(%1)+mul_35468(%2)
@@ -1172,7 +1172,7 @@ INIT_MMX sse
 VP8_IDCT_ADD
 
 ;-----------------------------------------------------------------------------
-; void vp8_luma_dc_wht_mmxext(DCTELEM block[4][4][16], DCTELEM dc[16])
+; void vp8_luma_dc_wht_mmxext(int16_t block[4][4][16], int16_t dc[16])
 ;-----------------------------------------------------------------------------
 
 %macro SCATTER_WHT 3
diff --git a/libavcodec/x86/vp8dsp_init.c b/libavcodec/x86/vp8dsp_init.c
index 735619c32b..09e2d9109e 100644
--- a/libavcodec/x86/vp8dsp_init.c
+++ b/libavcodec/x86/vp8dsp_init.c
@@ -230,21 +230,21 @@ HVBILIN(ssse3, 8,  4,  8)
 HVBILIN(ssse3, 8,  8, 16)
 HVBILIN(ssse3, 8, 16, 16)
 
-extern void ff_vp8_idct_dc_add_mmx(uint8_t *dst, DCTELEM block[16],
+extern void ff_vp8_idct_dc_add_mmx(uint8_t *dst, int16_t block[16],
                                    ptrdiff_t stride);
-extern void ff_vp8_idct_dc_add_sse4(uint8_t *dst, DCTELEM block[16],
+extern void ff_vp8_idct_dc_add_sse4(uint8_t *dst, int16_t block[16],
                                     ptrdiff_t stride);
-extern void ff_vp8_idct_dc_add4y_mmx(uint8_t *dst, DCTELEM block[4][16],
+extern void ff_vp8_idct_dc_add4y_mmx(uint8_t *dst, int16_t block[4][16],
                                       ptrdiff_t stride);
-extern void ff_vp8_idct_dc_add4y_sse2(uint8_t *dst, DCTELEM block[4][16],
+extern void ff_vp8_idct_dc_add4y_sse2(uint8_t *dst, int16_t block[4][16],
                                       ptrdiff_t stride);
-extern void ff_vp8_idct_dc_add4uv_mmx(uint8_t *dst, DCTELEM block[2][16],
+extern void ff_vp8_idct_dc_add4uv_mmx(uint8_t *dst, int16_t block[2][16],
                                       ptrdiff_t stride);
-extern void ff_vp8_luma_dc_wht_mmx(DCTELEM block[4][4][16], DCTELEM dc[16]);
-extern void ff_vp8_luma_dc_wht_sse(DCTELEM block[4][4][16], DCTELEM dc[16]);
-extern void ff_vp8_idct_add_mmx(uint8_t *dst, DCTELEM block[16],
+extern void ff_vp8_luma_dc_wht_mmx(int16_t block[4][4][16], int16_t dc[16]);
+extern void ff_vp8_luma_dc_wht_sse(int16_t block[4][4][16], int16_t dc[16]);
+extern void ff_vp8_idct_add_mmx(uint8_t *dst, int16_t block[16],
                                 ptrdiff_t stride);
-extern void ff_vp8_idct_add_sse(uint8_t *dst, DCTELEM block[16],
+extern void ff_vp8_idct_add_sse(uint8_t *dst, int16_t block[16],
                                 ptrdiff_t stride);
 
 #define DECLARE_LOOP_FILTER(NAME)\
diff --git a/libavfilter/libmpcodecs/vf_fspp.c b/libavfilter/libmpcodecs/vf_fspp.c
index 6f806b21c7..6090c9aeab 100644
--- a/libavfilter/libmpcodecs/vf_fspp.c
+++ b/libavfilter/libmpcodecs/vf_fspp.c
@@ -164,10 +164,10 @@ static void mul_thrmat_c(struct vf_priv_s *p,int q)
         ((short*)p->threshold_mtx)[a]=q * ((short*)p->threshold_mtx_noq)[a];//ints faster in C
 }
 
-static void column_fidct_c(int16_t* thr_adr, DCTELEM *data, DCTELEM *output, int cnt);
-static void row_idct_c(DCTELEM* workspace,
+static void column_fidct_c(int16_t* thr_adr, int16_t *data, int16_t *output, int cnt);
+static void row_idct_c(int16_t* workspace,
                        int16_t* output_adr, int output_stride, int cnt);
-static void row_fdct_c(DCTELEM *data, const uint8_t *pixels, int line_size, int cnt);
+static void row_fdct_c(int16_t *data, const uint8_t *pixels, int line_size, int cnt);
 
 //this is rather ugly, but there is no need for function pointers
 #define store_slice_s store_slice_c
@@ -393,10 +393,10 @@ static void mul_thrmat_mmx(struct vf_priv_s *p, int q)
         );
 }
 
-static void column_fidct_mmx(int16_t* thr_adr,  DCTELEM *data,  DCTELEM *output,  int cnt);
-static void row_idct_mmx(DCTELEM* workspace,
+static void column_fidct_mmx(int16_t* thr_adr,  int16_t *data,  int16_t *output,  int cnt);
+static void row_idct_mmx(int16_t* workspace,
                          int16_t* output_adr,  int output_stride,  int cnt);
-static void row_fdct_mmx(DCTELEM *data,  const uint8_t *pixels,  int line_size,  int cnt);
+static void row_fdct_mmx(int16_t *data,  const uint8_t *pixels,  int line_size,  int cnt);
 
 #define store_slice_s store_slice_mmx
 #define store_slice2_s store_slice2_mmx
@@ -416,8 +416,8 @@ static void filter(struct vf_priv_s *p, uint8_t *dst, uint8_t *src,
     const int step=6-p->log2_count;
     const int qps= 3 + is_luma;
     int32_t __attribute__((aligned(32))) block_align[4*8*BLOCKSZ+ 4*8*BLOCKSZ];
-    DCTELEM *block= (DCTELEM *)block_align;
-    DCTELEM *block3=(DCTELEM *)(block_align+4*8*BLOCKSZ);
+    int16_t *block= (int16_t *)block_align;
+    int16_t *block3=(int16_t *)(block_align+4*8*BLOCKSZ);
 
     memset(block3, 0, 4*8*BLOCKSZ);
 
@@ -460,8 +460,8 @@ static void filter(struct vf_priv_s *p, uint8_t *dst, uint8_t *src,
                     column_fidct_s((int16_t*)(&p->threshold_mtx[0]), block+x*8, block3+x*8, 8); //yes, this is a HOTSPOT
                 }
             row_idct_s(block3+0*8, p->temp + (y&15)*stride+x0+2-(y&1), stride, 2*(BLOCKSZ-1));
-            memmove(block, block+(BLOCKSZ-1)*64, 8*8*sizeof(DCTELEM)); //cycling
-            memmove(block3, block3+(BLOCKSZ-1)*64, 6*8*sizeof(DCTELEM));
+            memmove(block, block+(BLOCKSZ-1)*64, 8*8*sizeof(int16_t)); //cycling
+            memmove(block3, block3+(BLOCKSZ-1)*64, 6*8*sizeof(int16_t));
         }
         //
         es=width+8-x0; //  8, ...
@@ -694,7 +694,7 @@ const vf_info_t ff_vf_info_fspp = {
 
 //#define MANGLE(a) #a
 
-//typedef int16_t DCTELEM; //! only int16_t
+//typedef int16_t int16_t; //! only int16_t
 
 #define DCTSIZE 8
 #define DCTSIZE_S "8"
@@ -745,15 +745,15 @@ static const int16_t FIX_1_082392200=FIX(1.082392200, 13);
 
 #if !HAVE_MMX
 
-static void column_fidct_c(int16_t* thr_adr, DCTELEM *data, DCTELEM *output, int cnt)
+static void column_fidct_c(int16_t* thr_adr, int16_t *data, int16_t *output, int cnt)
 {
     int_simd16_t tmp0, tmp1, tmp2, tmp3, tmp4, tmp5, tmp6, tmp7;
     int_simd16_t tmp10, tmp11, tmp12, tmp13;
     int_simd16_t z1,z2,z3,z4,z5, z10, z11, z12, z13;
     int_simd16_t d0, d1, d2, d3, d4, d5, d6, d7;
 
-    DCTELEM* dataptr;
-    DCTELEM* wsptr;
+    int16_t* dataptr;
+    int16_t* wsptr;
     int16_t *threshold;
     int ctr;
 
@@ -870,7 +870,7 @@ static void column_fidct_c(int16_t* thr_adr, DCTELEM *data, DCTELEM *output, int
 
 #else /* HAVE_MMX */
 
-static void column_fidct_mmx(int16_t* thr_adr,  DCTELEM *data,  DCTELEM *output,  int cnt)
+static void column_fidct_mmx(int16_t* thr_adr,  int16_t *data,  int16_t *output,  int cnt)
 {
     uint64_t __attribute__((aligned(8))) temps[4];
     __asm__ volatile(
@@ -1605,14 +1605,14 @@ static void column_fidct_mmx(int16_t* thr_adr,  DCTELEM *data,  DCTELEM *output,
 
 #if !HAVE_MMX
 
-static void row_idct_c(DCTELEM* workspace,
+static void row_idct_c(int16_t* workspace,
                        int16_t* output_adr, int output_stride, int cnt)
 {
     int_simd16_t tmp0, tmp1, tmp2, tmp3, tmp4, tmp5, tmp6, tmp7;
     int_simd16_t tmp10, tmp11, tmp12, tmp13;
     int_simd16_t z5, z10, z11, z12, z13;
     int16_t* outptr;
-    DCTELEM* wsptr;
+    int16_t* wsptr;
 
     cnt*=4;
     wsptr = workspace;
@@ -1670,7 +1670,7 @@ static void row_idct_c(DCTELEM* workspace,
 
 #else /* HAVE_MMX */
 
-static void row_idct_mmx (DCTELEM* workspace,
+static void row_idct_mmx (int16_t* workspace,
                           int16_t* output_adr,  int output_stride,  int cnt)
 {
     uint64_t __attribute__((aligned(8))) temps[4];
@@ -1874,12 +1874,12 @@ static void row_idct_mmx (DCTELEM* workspace,
 
 #if !HAVE_MMX
 
-static void row_fdct_c(DCTELEM *data, const uint8_t *pixels, int line_size, int cnt)
+static void row_fdct_c(int16_t *data, const uint8_t *pixels, int line_size, int cnt)
 {
     int_simd16_t tmp0, tmp1, tmp2, tmp3, tmp4, tmp5, tmp6, tmp7;
     int_simd16_t tmp10, tmp11, tmp12, tmp13;
     int_simd16_t z1, z2, z3, z4, z5, z11, z13;
-    DCTELEM *dataptr;
+    int16_t *dataptr;
 
     cnt*=4;
     // Pass 1: process rows.
@@ -1937,7 +1937,7 @@ static void row_fdct_c(DCTELEM *data, const uint8_t *pixels, int line_size, int
 
 #else /* HAVE_MMX */
 
-static void row_fdct_mmx(DCTELEM *data,  const uint8_t *pixels,  int line_size,  int cnt)
+static void row_fdct_mmx(int16_t *data,  const uint8_t *pixels,  int line_size,  int cnt)
 {
     uint64_t __attribute__((aligned(8))) temps[4];
     __asm__ volatile(
diff --git a/libavfilter/libmpcodecs/vf_pp7.c b/libavfilter/libmpcodecs/vf_pp7.c
index 7f0e39d6ff..14bd863525 100644
--- a/libavfilter/libmpcodecs/vf_pp7.c
+++ b/libavfilter/libmpcodecs/vf_pp7.c
@@ -44,7 +44,7 @@
 #define XMIN(a,b) ((a) < (b) ? (a) : (b))
 #define XMAX(a,b) ((a) > (b) ? (a) : (b))
 
-typedef short DCTELEM;
+typedef short int16_t;
 
 //===========================================================================//
 static const uint8_t  __attribute__((aligned(8))) dither[8][8]={
@@ -66,7 +66,7 @@ struct vf_priv_s {
     uint8_t *src;
 };
 #if 0
-static inline void dct7_c(DCTELEM *dst, int s0, int s1, int s2, int s3, int step){
+static inline void dct7_c(int16_t *dst, int s0, int s1, int s2, int s3, int step){
     int s, d;
     int dst2[64];
 //#define S0 (1024/0.37796447300922719759)
@@ -113,7 +113,7 @@ static inline void dct7_c(DCTELEM *dst, int s0, int s1, int s2, int s3, int step
 }
 #endif
 
-static inline void dctA_c(DCTELEM *dst, uint8_t *src, int stride){
+static inline void dctA_c(int16_t *dst, uint8_t *src, int stride){
     int i;
 
     for(i=0; i<4; i++){
@@ -135,7 +135,7 @@ static inline void dctA_c(DCTELEM *dst, uint8_t *src, int stride){
     }
 }
 
-static void dctB_c(DCTELEM *dst, DCTELEM *src){
+static void dctB_c(int16_t *dst, int16_t *src){
     int i;
 
     for(i=0; i<4; i++){
@@ -158,7 +158,7 @@ static void dctB_c(DCTELEM *dst, DCTELEM *src){
 }
 
 #if HAVE_MMX
-static void dctB_mmx(DCTELEM *dst, DCTELEM *src){
+static void dctB_mmx(int16_t *dst, int16_t *src){
     __asm__ volatile (
         "movq  (%0), %%mm0      \n\t"
         "movq  1*4*2(%0), %%mm1 \n\t"
@@ -191,7 +191,7 @@ static void dctB_mmx(DCTELEM *dst, DCTELEM *src){
 }
 #endif
 
-static void (*dctB)(DCTELEM *dst, DCTELEM *src)= dctB_c;
+static void (*dctB)(int16_t *dst, int16_t *src)= dctB_c;
 
 #define N0 4
 #define N1 5
@@ -228,7 +228,7 @@ static void init_thres2(void){
     }
 }
 
-static int hardthresh_c(DCTELEM *src, int qp){
+static int hardthresh_c(int16_t *src, int qp){
     int i;
     int a;
 
@@ -244,7 +244,7 @@ static int hardthresh_c(DCTELEM *src, int qp){
     return (a + (1<<11))>>12;
 }
 
-static int mediumthresh_c(DCTELEM *src, int qp){
+static int mediumthresh_c(int16_t *src, int qp){
     int i;
     int a;
 
@@ -265,7 +265,7 @@ static int mediumthresh_c(DCTELEM *src, int qp){
     return (a + (1<<11))>>12;
 }
 
-static int softthresh_c(DCTELEM *src, int qp){
+static int softthresh_c(int16_t *src, int qp){
     int i;
     int a;
 
@@ -282,14 +282,14 @@ static int softthresh_c(DCTELEM *src, int qp){
     return (a + (1<<11))>>12;
 }
 
-static int (*requantize)(DCTELEM *src, int qp)= hardthresh_c;
+static int (*requantize)(int16_t *src, int qp)= hardthresh_c;
 
 static void filter(struct vf_priv_s *p, uint8_t *dst, uint8_t *src, int dst_stride, int src_stride, int width, int height, uint8_t *qp_store, int qp_stride, int is_luma){
     int x, y;
     const int stride= is_luma ? p->temp_stride : ((width+16+15)&(~15));
     uint8_t  *p_src= p->src + 8*stride;
-    DCTELEM *block= (DCTELEM *)p->src;
-    DCTELEM *temp= (DCTELEM *)(p->src + 32);
+    int16_t *block= (int16_t *)p->src;
+    int16_t *temp= (int16_t *)(p->src + 32);
 
     if (!src || !dst) return; // HACK avoid crash for Y8 colourspace
     for(y=0; y<height; y++){
@@ -310,7 +310,7 @@ static void filter(struct vf_priv_s *p, uint8_t *dst, uint8_t *src, int dst_stri
         for(x=-8; x<0; x+=4){
             const int index= x + y*stride + (8-3)*(1+stride) + 8; //FIXME silly offset
             uint8_t *src  = p_src + index;
-            DCTELEM *tp= temp+4*x;
+            int16_t *tp= temp+4*x;
 
             dctA_c(tp+4*8, src, stride);
         }
@@ -328,7 +328,7 @@ static void filter(struct vf_priv_s *p, uint8_t *dst, uint8_t *src, int dst_stri
             for(; x<end; x++){
                 const int index= x + y*stride + (8-3)*(1+stride) + 8; //FIXME silly offset
                 uint8_t *src  = p_src + index;
-                DCTELEM *tp= temp+4*x;
+                int16_t *tp= temp+4*x;
                 int v;
 
                 if((x&3)==0)
diff --git a/libavfilter/libmpcodecs/vf_spp.c b/libavfilter/libmpcodecs/vf_spp.c
index b53b57edc9..f821374e06 100644
--- a/libavfilter/libmpcodecs/vf_spp.c
+++ b/libavfilter/libmpcodecs/vf_spp.c
@@ -105,7 +105,7 @@ struct vf_priv_s {
 
 #define SHIFT 22
 
-static void hardthresh_c(DCTELEM dst[64], DCTELEM src[64], int qp, uint8_t *permutation){
+static void hardthresh_c(int16_t dst[64], int16_t src[64], int qp, uint8_t *permutation){
         int i;
         int bias= 0; //FIXME
         unsigned int threshold1, threshold2;
@@ -113,7 +113,7 @@ static void hardthresh_c(DCTELEM dst[64], DCTELEM src[64], int qp, uint8_t *perm
         threshold1= qp*((1<<4) - bias) - 1;
         threshold2= (threshold1<<1);
 
-        memset(dst, 0, 64*sizeof(DCTELEM));
+        memset(dst, 0, 64*sizeof(int16_t));
         dst[0]= (src[0] + 4)>>3;
 
         for(i=1; i<64; i++){
@@ -125,7 +125,7 @@ static void hardthresh_c(DCTELEM dst[64], DCTELEM src[64], int qp, uint8_t *perm
         }
 }
 
-static void softthresh_c(DCTELEM dst[64], DCTELEM src[64], int qp, uint8_t *permutation){
+static void softthresh_c(int16_t dst[64], int16_t src[64], int qp, uint8_t *permutation){
         int i;
         int bias= 0; //FIXME
         unsigned int threshold1, threshold2;
@@ -133,7 +133,7 @@ static void softthresh_c(DCTELEM dst[64], DCTELEM src[64], int qp, uint8_t *perm
         threshold1= qp*((1<<4) - bias) - 1;
         threshold2= (threshold1<<1);
 
-        memset(dst, 0, 64*sizeof(DCTELEM));
+        memset(dst, 0, 64*sizeof(int16_t));
         dst[0]= (src[0] + 4)>>3;
 
         for(i=1; i<64; i++){
@@ -149,7 +149,7 @@ static void softthresh_c(DCTELEM dst[64], DCTELEM src[64], int qp, uint8_t *perm
 }
 
 #if HAVE_MMX
-static void hardthresh_mmx(DCTELEM dst[64], DCTELEM src[64], int qp, uint8_t *permutation){
+static void hardthresh_mmx(int16_t dst[64], int16_t src[64], int qp, uint8_t *permutation){
         int bias= 0; //FIXME
         unsigned int threshold1;
 
@@ -217,7 +217,7 @@ static void hardthresh_mmx(DCTELEM dst[64], DCTELEM src[64], int qp, uint8_t *pe
         dst[0]= (src[0] + 4)>>3;
 }
 
-static void softthresh_mmx(DCTELEM dst[64], DCTELEM src[64], int qp, uint8_t *permutation){
+static void softthresh_mmx(int16_t dst[64], int16_t src[64], int qp, uint8_t *permutation){
         int bias= 0; //FIXME
         unsigned int threshold1;
 
@@ -294,7 +294,7 @@ static void softthresh_mmx(DCTELEM dst[64], DCTELEM src[64], int qp, uint8_t *pe
 }
 #endif
 
-static inline void add_block(int16_t *dst, int stride, DCTELEM block[64]){
+static inline void add_block(int16_t *dst, int stride, int16_t block[64]){
         int y;
 
         for(y=0; y<8; y++){
@@ -372,15 +372,15 @@ static void store_slice_mmx(uint8_t *dst, int16_t *src, int dst_stride, int src_
 
 static void (*store_slice)(uint8_t *dst, int16_t *src, int dst_stride, int src_stride, int width, int height, int log2_scale)= store_slice_c;
 
-static void (*requantize)(DCTELEM dst[64], DCTELEM src[64], int qp, uint8_t *permutation)= hardthresh_c;
+static void (*requantize)(int16_t dst[64], int16_t src[64], int qp, uint8_t *permutation)= hardthresh_c;
 
 static void filter(struct vf_priv_s *p, uint8_t *dst, uint8_t *src, int dst_stride, int src_stride, int width, int height, uint8_t *qp_store, int qp_stride, int is_luma){
         int x, y, i;
         const int count= 1<<p->log2_count;
         const int stride= is_luma ? p->temp_stride : ((width+16+15)&(~15));
         uint64_t __attribute__((aligned(16))) block_align[32];
-        DCTELEM *block = (DCTELEM *)block_align;
-        DCTELEM *block2= (DCTELEM *)(block_align+16);
+        int16_t *block = (int16_t *)block_align;
+        int16_t *block2= (int16_t *)(block_align+16);
 
         if (!src || !dst) return; // HACK avoid crash for Y8 colourspace
         for(y=0; y<height; y++){
diff --git a/libavfilter/vf_decimate.c b/libavfilter/vf_decimate.c
index b81c3ee8f2..0d89b8153f 100644
--- a/libavfilter/vf_decimate.c
+++ b/libavfilter/vf_decimate.c
@@ -63,7 +63,7 @@ static int diff_planes(AVFilterContext *ctx,
     int x, y;
     int d, c = 0;
     int t = (w/16)*(h/16)*decimate->frac;
-    DCTELEM block[8*8];
+    int16_t block[8*8];
 
     /* compute difference for blocks of 8x8 bytes */
     for (y = 0; y < h-7; y += 4) {
diff --git a/tools/patcheck b/tools/patcheck
index a1d4c06e80..83db4c0438 100755
--- a/tools/patcheck
+++ b/tools/patcheck
@@ -173,7 +173,7 @@ vertical align =
 /* and * align
 arrays fitting in smaller types
 variables written to twice with no interspaced read
-memset(block, 0, 6*64*sizeof(DCTELEM)); -> clear_blocks
+memset(block, 0, 6*64*sizeof(int16_t)); -> clear_blocks
 check existence of long_name in AVCodec
 check that the patch does not touch codec & (de)muxer layer at the same time ->split