#define ROW_SHIFT 13
#define COL_SHIFT 18
#define DC_SHIFT 1
+# elif IN_IDCT_DEPTH == 32
+#define ROW_SHIFT 13
+#define COL_SHIFT 21
+#define DC_SHIFT 2
# else
#define ROW_SHIFT 12
#define COL_SHIFT 19
#define DC_SHIFT -1
# endif
-#define MUL(a, b) ((a) * (b))
-#define MAC(a, b, c) ((a) += (b) * (c))
+#define MUL(a, b) ((int)((SUINT)(a) * (b)))
+#define MAC(a, b, c) ((a) += (SUINT)(b) * (c))
#else
#ifdef EXTRA_SHIFT
static inline void FUNC(idctRowCondDC_extrashift)(int16_t *row, int extra_shift)
#else
-static inline void FUNC(idctRowCondDC)(int16_t *row, int extra_shift)
+static inline void FUNC6(idctRowCondDC)(idctin *row, int extra_shift)
#endif
{
SUINT a0, a1, a2, a3, b0, b1, b2, b3;
+// TODO: Add DC-only support for int32_t input
+#if IN_IDCT_DEPTH == 16
#if HAVE_FAST_64BIT
#define ROW0_MASK (0xffffLL << 48 * HAVE_BIGENDIAN)
if (((AV_RN64A(row) & ~ROW0_MASK) | AV_RN64A(row+4)) == 0) {
AV_WN32A(row+6, temp);
return;
}
+#endif
#endif
- a0 = (W4 * row[0]) + (1 << (ROW_SHIFT + extra_shift - 1));
+ a0 = ((SUINT)W4 * row[0]) + (1 << (ROW_SHIFT + extra_shift - 1));
a1 = a0;
a2 = a0;
a3 = a0;
- a0 += W2 * row[2];
- a1 += W6 * row[2];
- a2 -= W6 * row[2];
- a3 -= W2 * row[2];
+ a0 += (SUINT)W2 * row[2];
+ a1 += (SUINT)W6 * row[2];
+ a2 -= (SUINT)W6 * row[2];
+ a3 -= (SUINT)W2 * row[2];
b0 = MUL(W1, row[1]);
MAC(b0, W3, row[3]);
b3 = MUL(W7, row[1]);
MAC(b3, -W5, row[3]);
+#if IN_IDCT_DEPTH == 32
+ if (AV_RN64A(row + 4) | AV_RN64A(row + 6)) {
+#else
if (AV_RN64A(row + 4)) {
- a0 += W4*row[4] + W6*row[6];
- a1 += - W4*row[4] - W2*row[6];
- a2 += - W4*row[4] + W2*row[6];
- a3 += W4*row[4] - W6*row[6];
+#endif
+ a0 += (SUINT) W4*row[4] + (SUINT)W6*row[6];
+ a1 += (SUINT)- W4*row[4] - (SUINT)W2*row[6];
+ a2 += (SUINT)- W4*row[4] + (SUINT)W2*row[6];
+ a3 += (SUINT) W4*row[4] - (SUINT)W6*row[6];
MAC(b0, W5, row[5]);
MAC(b0, W7, row[7]);
}
#define IDCT_COLS do { \
- a0 = W4 * (col[8*0] + ((1<<(COL_SHIFT-1))/W4)); \
+ a0 = (SUINT)W4 * (col[8*0] + ((1<<(COL_SHIFT-1))/W4)); \
a1 = a0; \
a2 = a0; \
a3 = a0; \
\
- a0 += W2*col[8*2]; \
- a1 += W6*col[8*2]; \
- a2 += -W6*col[8*2]; \
- a3 += -W2*col[8*2]; \
+ a0 += (SUINT) W2*col[8*2]; \
+ a1 += (SUINT) W6*col[8*2]; \
+ a2 += (SUINT)-W6*col[8*2]; \
+ a3 += (SUINT)-W2*col[8*2]; \
\
b0 = MUL(W1, col[8*1]); \
b1 = MUL(W3, col[8*1]); \
MAC(b3, -W5, col[8*3]); \
\
if (col[8*4]) { \
- a0 += W4*col[8*4]; \
- a1 += -W4*col[8*4]; \
- a2 += -W4*col[8*4]; \
- a3 += W4*col[8*4]; \
+ a0 += (SUINT) W4*col[8*4]; \
+ a1 += (SUINT)-W4*col[8*4]; \
+ a2 += (SUINT)-W4*col[8*4]; \
+ a3 += (SUINT) W4*col[8*4]; \
} \
\
if (col[8*5]) { \
} \
\
if (col[8*6]) { \
- a0 += W6*col[8*6]; \
- a1 += -W2*col[8*6]; \
- a2 += W2*col[8*6]; \
- a3 += -W6*col[8*6]; \
+ a0 += (SUINT) W6*col[8*6]; \
+ a1 += (SUINT)-W2*col[8*6]; \
+ a2 += (SUINT) W2*col[8*6]; \
+ a3 += (SUINT)-W6*col[8*6]; \
} \
\
if (col[8*7]) { \
#ifdef EXTRA_SHIFT
static inline void FUNC(idctSparseCol_extrashift)(int16_t *col)
#else
-static inline void FUNC(idctSparseColPut)(pixel *dest, ptrdiff_t line_size,
- int16_t *col)
+static inline void FUNC6(idctSparseColPut)(pixel *dest, ptrdiff_t line_size,
+ idctin *col)
{
SUINT a0, a1, a2, a3, b0, b1, b2, b3;
dest[0] = av_clip_pixel((int)(a0 - b0) >> COL_SHIFT);
}
-static inline void FUNC(idctSparseColAdd)(pixel *dest, ptrdiff_t line_size,
- int16_t *col)
+static inline void FUNC6(idctSparseColAdd)(pixel *dest, ptrdiff_t line_size,
+ idctin *col)
{
int a0, a1, a2, a3, b0, b1, b2, b3;
dest[0] = av_clip_pixel(dest[0] + ((a0 - b0) >> COL_SHIFT));
}
-static inline void FUNC(idctSparseCol)(int16_t *col)
+static inline void FUNC6(idctSparseCol)(idctin *col)
#endif
{
int a0, a1, a2, a3, b0, b1, b2, b3;
}
#ifndef EXTRA_SHIFT
-void FUNC(ff_simple_idct_put)(uint8_t *dest_, ptrdiff_t line_size, int16_t *block)
+void FUNC6(ff_simple_idct_put)(uint8_t *dest_, ptrdiff_t line_size, int16_t *block_)
{
+ idctin *block = (idctin *)block_;
pixel *dest = (pixel *)dest_;
int i;
line_size /= sizeof(pixel);
for (i = 0; i < 8; i++)
- FUNC(idctRowCondDC)(block + i*8, 0);
+ FUNC6(idctRowCondDC)(block + i*8, 0);
for (i = 0; i < 8; i++)
- FUNC(idctSparseColPut)(dest + i, line_size, block + i);
+ FUNC6(idctSparseColPut)(dest + i, line_size, block + i);
}
-void FUNC(ff_simple_idct_add)(uint8_t *dest_, ptrdiff_t line_size, int16_t *block)
+#if IN_IDCT_DEPTH == 16
+void FUNC6(ff_simple_idct_add)(uint8_t *dest_, ptrdiff_t line_size, int16_t *block)
{
pixel *dest = (pixel *)dest_;
int i;
line_size /= sizeof(pixel);
for (i = 0; i < 8; i++)
- FUNC(idctRowCondDC)(block + i*8, 0);
+ FUNC6(idctRowCondDC)(block + i*8, 0);
for (i = 0; i < 8; i++)
- FUNC(idctSparseColAdd)(dest + i, line_size, block + i);
+ FUNC6(idctSparseColAdd)(dest + i, line_size, block + i);
}
-void FUNC(ff_simple_idct)(int16_t *block)
+void FUNC6(ff_simple_idct)(int16_t *block)
{
int i;
for (i = 0; i < 8; i++)
- FUNC(idctRowCondDC)(block + i*8, 0);
+ FUNC6(idctRowCondDC)(block + i*8, 0);
for (i = 0; i < 8; i++)
- FUNC(idctSparseCol)(block + i);
+ FUNC6(idctSparseCol)(block + i);
}
#endif
+#endif