summaryrefslogtreecommitdiff
path: root/src/libffmpeg/libavcodec/dsputil.h
diff options
context:
space:
mode:
Diffstat (limited to 'src/libffmpeg/libavcodec/dsputil.h')
-rw-r--r--src/libffmpeg/libavcodec/dsputil.h80
1 files changed, 46 insertions, 34 deletions
diff --git a/src/libffmpeg/libavcodec/dsputil.h b/src/libffmpeg/libavcodec/dsputil.h
index a9b472f86..dc3bc01e8 100644
--- a/src/libffmpeg/libavcodec/dsputil.h
+++ b/src/libffmpeg/libavcodec/dsputil.h
@@ -15,7 +15,7 @@
*
* You should have received a copy of the GNU Lesser General Public
* License along with this library; if not, write to the Free Software
- * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
*/
/**
@@ -31,7 +31,7 @@
#include "common.h"
#include "avcodec.h"
-#if defined(ARCH_X86)
+#if defined(ARCH_X86) || defined(ARCH_X86_64)
#define HAVE_MMX 1
#endif
@@ -102,7 +102,7 @@ typedef void (*tpel_mc_func)(uint8_t *block/*align width (8 or 16)*/, const uint
typedef void (*qpel_mc_func)(uint8_t *dst/*align width (8 or 16)*/, uint8_t *src/*align 1*/, int stride);
typedef void (*h264_chroma_mc_func)(uint8_t *dst/*align 8*/, uint8_t *src/*align 1*/, int srcStride, int h, int x, int y);
typedef void (*h264_weight_func)(uint8_t *block, int stride, int log2_denom, int weight, int offset);
-typedef void (*h264_biweight_func)(uint8_t *dst, uint8_t *src, int stride, int log2_denom, int weightd, int weights, int offsetd, int offsets);
+typedef void (*h264_biweight_func)(uint8_t *dst, uint8_t *src, int stride, int log2_denom, int weightd, int weights, int offset);
#define DEF_OLD_QPEL(name)\
void ff_put_ ## name (uint8_t *dst/*align width (8 or 16)*/, uint8_t *src/*align 1*/, int stride);\
@@ -154,12 +154,12 @@ typedef struct DSPContext {
* global motion compensation.
*/
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);
+ int dxx, int dxy, int dyx, int dyy, int shift, int r, int width, int height);
void (*clear_blocks)(DCTELEM *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
-
+
me_cmp_func sad[5]; /* identical to pix_absAxA except additional void * */
me_cmp_func sse[5];
me_cmp_func hadamard8_diff[5];
@@ -173,6 +173,7 @@ typedef struct DSPContext {
me_cmp_func w53[5];
me_cmp_func w97[5];
me_cmp_func dct_max[5];
+ me_cmp_func dct264_sad[5];
me_cmp_func me_pre_cmp[5];
me_cmp_func me_cmp[5];
@@ -183,7 +184,7 @@ typedef struct DSPContext {
/**
* Halfpel motion compensation with rounding (a+b+1)>>1.
- * this is an array[4][4] of motion compensation funcions for 4
+ * this is an array[4][4] of motion compensation funcions for 4
* horizontal blocksizes (8,16) and the 4 halfpel positions<br>
* *pixels_tab[ 0->16xH 1->8xH ][ xhalfpel + 2*yhalfpel ]
* @param block destination where the result is stored
@@ -195,7 +196,7 @@ typedef struct DSPContext {
/**
* Halfpel motion compensation with rounding (a+b+1)>>1.
- * This is an array[4][4] of motion compensation functions for 4
+ * This is an array[4][4] of motion compensation functions for 4
* horizontal blocksizes (8,16) and the 4 halfpel positions<br>
* *pixels_tab[ 0->16xH 1->8xH ][ xhalfpel + 2*yhalfpel ]
* @param block destination into which the result is averaged (a+b+1)>>1
@@ -207,7 +208,7 @@ typedef struct DSPContext {
/**
* Halfpel motion compensation with no rounding (a+b)>>1.
- * this is an array[2][4] of motion compensation funcions for 2
+ * this is an array[2][4] of motion compensation funcions for 2
* horizontal blocksizes (8,16) and the 4 halfpel positions<br>
* *pixels_tab[ 0->16xH 1->8xH ][ xhalfpel + 2*yhalfpel ]
* @param block destination where the result is stored
@@ -219,7 +220,7 @@ typedef struct DSPContext {
/**
* Halfpel motion compensation with no rounding (a+b)>>1.
- * this is an array[2][4] of motion compensation funcions for 2
+ * this is an array[2][4] of motion compensation funcions for 2
* horizontal blocksizes (8,16) and the 4 halfpel positions<br>
* *pixels_tab[ 0->16xH 1->8xH ][ xhalfpel + 2*yhalfpel ]
* @param block destination into which the result is averaged (a+b)>>1
@@ -228,9 +229,9 @@ typedef struct DSPContext {
* @param h height
*/
op_pixels_func avg_no_rnd_pixels_tab[4][4];
-
+
void (*put_no_rnd_pixels_l2[2])(uint8_t *block/*align width (8 or 16)*/, const uint8_t *a/*align 1*/, const uint8_t *b/*align 1*/, int line_size, int h);
-
+
/**
* Thirdpel motion compensation with rounding (a+b+1)>>1.
* this is an array[12] of motion compensation funcions for the 9 thirdpel positions<br>
@@ -248,21 +249,21 @@ typedef struct DSPContext {
qpel_mc_func put_no_rnd_qpel_pixels_tab[2][16];
qpel_mc_func avg_no_rnd_qpel_pixels_tab[2][16];
qpel_mc_func put_mspel_pixels_tab[8];
-
+
/**
* h264 Chram MC
*/
h264_chroma_mc_func put_h264_chroma_pixels_tab[3];
h264_chroma_mc_func avg_h264_chroma_pixels_tab[3];
- qpel_mc_func put_h264_qpel_pixels_tab[3][16];
- qpel_mc_func avg_h264_qpel_pixels_tab[3][16];
-
+ qpel_mc_func put_h264_qpel_pixels_tab[4][16];
+ qpel_mc_func avg_h264_qpel_pixels_tab[4][16];
+
h264_weight_func weight_h264_pixels_tab[10];
h264_biweight_func biweight_h264_pixels_tab[10];
-
+
me_cmp_func pix_abs[2][4];
-
+
/* huffyuv specific */
void (*add_bytes)(uint8_t *dst/*align 16*/, uint8_t *src/*align 16*/, int w);
void (*diff_bytes)(uint8_t *dst/*align 16*/, uint8_t *src1/*align 16*/, uint8_t *src2/*align 1*/,int w);
@@ -279,7 +280,7 @@ typedef struct DSPContext {
void (*h264_h_loop_filter_chroma)(uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0);
void (*h264_v_loop_filter_chroma_intra)(uint8_t *pix, int stride, int alpha, int beta);
void (*h264_h_loop_filter_chroma_intra)(uint8_t *pix, int stride, int alpha, int beta);
-
+
void (*h263_v_loop_filter)(uint8_t *src, int stride, int qscale);
void (*h263_h_loop_filter)(uint8_t *src, int stride, int qscale);
@@ -288,23 +289,23 @@ typedef struct DSPContext {
/* (I)DCT */
void (*fdct)(DCTELEM *block/* align 16*/);
void (*fdct248)(DCTELEM *block/* align 16*/);
-
+
/* IDCT really*/
void (*idct)(DCTELEM *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 horizotal line of dest
*/
void (*idct_put)(uint8_t *dest/*align 8*/, int line_size, DCTELEM *block/*align 16*/);
-
+
/**
* block -> idct -> add dest -> clip to unsigned 8 bit -> dest.
* @param line_size size in bytes of a horizotal line of dest
*/
void (*idct_add)(uint8_t *dest/*align 8*/, int line_size, DCTELEM *block/*align 16*/);
-
+
/**
* idct input permutation.
* several optimized IDCTs need a permutated input (relative to the normal order of the reference
@@ -329,7 +330,7 @@ typedef struct DSPContext {
void (*add_8x8basis)(int16_t rem[64], int16_t basis[64], int scale);
#define BASIS_SHIFT 16
#define RECON_SHIFT 6
-
+
void (*h264_idct_add)(uint8_t *dst, DCTELEM *block, int stride);
void (*h264_idct8_add)(uint8_t *dst, DCTELEM *block, int stride);
} DSPContext;
@@ -345,7 +346,7 @@ void ff_block_permute(DCTELEM *block, uint8_t *permutation, const uint8_t *scant
void ff_set_cmp(DSPContext* c, me_cmp_func *cmp, int type);
-#define BYTE_VEC32(c) ((c)*0x01010101UL)
+#define BYTE_VEC32(c) ((c)*0x01010101UL)
static inline uint32_t rnd_avg32(uint32_t a, uint32_t b)
{
@@ -369,6 +370,7 @@ static inline int get_penalty_factor(int lambda, int lambda2, int type){
case FF_CMP_W97:
return (2*lambda)>>(FF_LAMBDA_SHIFT);
case FF_CMP_SATD:
+ case FF_CMP_DCT264:
return (2*lambda)>>FF_LAMBDA_SHIFT;
case FF_CMP_RD:
case FF_CMP_PSNR:
@@ -391,7 +393,11 @@ static inline int get_penalty_factor(int lambda, int lambda2, int type){
one or more MultiMedia extension */
int mm_support(void);
-#define __align16 __attribute__ ((aligned (16)))
+#ifdef __GNUC__
+ #define DECLARE_ALIGNED_16(t,v) t v __attribute__ ((aligned (16)))
+#else
+ #define DECLARE_ALIGNED_16(t,v) __declspec(align(16)) t v
+#endif
#if defined(HAVE_MMX)
@@ -422,7 +428,12 @@ static inline void emms(void)
emms();\
}
-#define __align8 __attribute__ ((aligned (8)))
+#ifdef __GNUC__
+ #define DECLARE_ALIGNED_8(t,v) t v __attribute__ ((aligned (8)))
+#else
+ #define DECLARE_ALIGNED_8(t,v) __declspec(align(8)) t v
+#endif
+
#define STRIDE_ALIGN 8
void dsputil_init_mmx(DSPContext* c, AVCodecContext *avctx);
@@ -432,7 +443,7 @@ void dsputil_init_pix_mmx(DSPContext* c, AVCodecContext *avctx);
/* This is to use 4 bytes read to the IDCT pointers for some 'zero'
line optimizations */
-#define __align8 __attribute__ ((aligned (4)))
+#define DECLARE_ALIGNED_8(t,v) t v __attribute__ ((aligned (4)))
#define STRIDE_ALIGN 4
#define MM_IWMMXT 0x0100 /* XScale IWMMXT */
@@ -444,7 +455,7 @@ void dsputil_init_armv4l(DSPContext* c, AVCodecContext *avctx);
#elif defined(HAVE_MLIB)
/* SPARC/VIS IDCT needs 8-byte aligned DCT blocks */
-#define __align8 __attribute__ ((aligned (8)))
+#define DECLARE_ALIGNED_8(t,v) t v __attribute__ ((aligned (8)))
#define STRIDE_ALIGN 8
void dsputil_init_mlib(DSPContext* c, AVCodecContext *avctx);
@@ -452,13 +463,13 @@ void dsputil_init_mlib(DSPContext* c, AVCodecContext *avctx);
#elif defined(ARCH_SPARC)
/* SPARC/VIS IDCT needs 8-byte aligned DCT blocks */
-#define __align8 __attribute__ ((aligned (8)))
+#define DECLARE_ALIGNED_8(t,v) t v __attribute__ ((aligned (8)))
#define STRIDE_ALIGN 8
void dsputil_init_vis(DSPContext* c, AVCodecContext *avctx);
#elif defined(ARCH_ALPHA)
-#define __align8 __attribute__ ((aligned (8)))
+#define DECLARE_ALIGNED_8(t,v) t v __attribute__ ((aligned (8)))
#define STRIDE_ALIGN 8
void dsputil_init_alpha(DSPContext* c, AVCodecContext *avctx);
@@ -475,28 +486,28 @@ extern int mm_flags;
#undef pixel
#endif
-#define __align8 __attribute__ ((aligned (16)))
+#define DECLARE_ALIGNED_8(t,v) t v __attribute__ ((aligned (16)))
#define STRIDE_ALIGN 16
void dsputil_init_ppc(DSPContext* c, AVCodecContext *avctx);
#elif defined(HAVE_MMI)
-#define __align8 __attribute__ ((aligned (16)))
+#define DECLARE_ALIGNED_8(t,v) t v __attribute__ ((aligned (16)))
#define STRIDE_ALIGN 16
void dsputil_init_mmi(DSPContext* c, AVCodecContext *avctx);
#elif defined(ARCH_SH4)
-#define __align8 __attribute__ ((aligned (8)))
+#define DECLARE_ALIGNED_8(t,v) t v __attribute__ ((aligned (8)))
#define STRIDE_ALIGN 8
void dsputil_init_sh4(DSPContext* c, AVCodecContext *avctx);
#else
-#define __align8 __attribute__ ((aligned (8)))
+#define DECLARE_ALIGNED_8(t,v) t v __attribute__ ((aligned (8)))
#define STRIDE_ALIGN 8
#endif
@@ -511,6 +522,7 @@ struct unaligned_16 { uint16_t l; } __attribute__((packed));
#define LD32(a) (((const struct unaligned_32 *) (a))->l)
#define LD64(a) (((const struct unaligned_64 *) (a))->l)
+#define ST16(a, b) (((struct unaligned_16 *) (a))->l) = (b)
#define ST32(a, b) (((struct unaligned_32 *) (a))->l) = (b)
#else /* __GNUC__ */
generated by cgit v1.2.3 (git 2.25.1) at 2025-04-26 21:22:24 +0000