summaryrefslogtreecommitdiff
diff options
context:
space:
mode:
Diffstat
-rw-r--r--vp9/common/vp9_idctllm.c25
-rw-r--r--vp9/common/vp9_invtrans.c10
-rw-r--r--vp9/decoder/vp9_dequantize.c6
-rw-r--r--vp9/encoder/vp9_dct.c9
-rw-r--r--vp9/encoder/vp9_encodeintra.c12
-rw-r--r--vp9/encoder/vp9_encodemb.c8
-rw-r--r--vp9/encoder/vp9_rdopt.c106
7 files changed, 107 insertions, 69 deletions
diff --git a/vp9/common/vp9_idctllm.c b/vp9/common/vp9_idctllm.c
index 67cfc9d71..632dae8fd 100644
--- a/vp9/common/vp9_idctllm.c
+++ b/vp9/common/vp9_idctllm.c
@@ -122,7 +122,7 @@ void vp9_dc_only_inv_walsh_add_c(int input_dc, uint8_t *pred_ptr,
}
}
-void idct4_1d(int16_t *input, int16_t *output) {
+static void idct4_1d(int16_t *input, int16_t *output) {
int16_t step[4];
int temp1, temp2;
// stage 1
@@ -200,7 +200,7 @@ void vp9_dc_only_idct_add_c(int input_dc, uint8_t *pred_ptr,
}
}
-void idct8_1d(int16_t *input, int16_t *output) {
+static void idct8_1d(int16_t *input, int16_t *output) {
int16_t step1[8], step2[8];
int temp1, temp2;
// stage 1
@@ -320,10 +320,9 @@ static const transform_2d IHT_4[] = {
void vp9_short_iht4x4_c(int16_t *input, int16_t *output,
int pitch, TX_TYPE tx_type) {
+ int i, j;
int16_t out[4 * 4];
int16_t *outptr = out;
- const int half_pitch = pitch >> 1;
- int i, j;
int16_t temp_in[4], temp_out[4];
const transform_2d ht = IHT_4[tx_type];
@@ -340,7 +339,7 @@ void vp9_short_iht4x4_c(int16_t *input, int16_t *output,
temp_in[j] = out[j * 4 + i];
ht.cols(temp_in, temp_out);
for (j = 0; j < 4; ++j)
- output[j * half_pitch + i] = ROUND_POWER_OF_TWO(temp_out[j], 4);
+ output[j * pitch + i] = ROUND_POWER_OF_TWO(temp_out[j], 4);
}
}
@@ -430,10 +429,9 @@ static const transform_2d IHT_8[] = {
void vp9_short_iht8x8_c(int16_t *input, int16_t *output,
int pitch, TX_TYPE tx_type) {
+ int i, j;
int16_t out[8 * 8];
int16_t *outptr = out;
- const int half_pitch = pitch >> 1;
- int i, j;
int16_t temp_in[8], temp_out[8];
const transform_2d ht = IHT_8[tx_type];
@@ -450,7 +448,7 @@ void vp9_short_iht8x8_c(int16_t *input, int16_t *output,
temp_in[j] = out[j * 8 + i];
ht.cols(temp_in, temp_out);
for (j = 0; j < 8; ++j)
- output[j * half_pitch + i] = ROUND_POWER_OF_TWO(temp_out[j], 5);
+ output[j * pitch + i] = ROUND_POWER_OF_TWO(temp_out[j], 5);
}
}
@@ -486,7 +484,7 @@ void vp9_short_idct1_8x8_c(int16_t *input, int16_t *output) {
output[0] = ROUND_POWER_OF_TWO(out, 5);
}
-void idct16_1d(int16_t *input, int16_t *output) {
+static void idct16_1d(int16_t *input, int16_t *output) {
int16_t step1[16], step2[16];
int temp1, temp2;
@@ -853,18 +851,17 @@ static const transform_2d IHT_16[] = {
};
void vp9_short_iht16x16_c(int16_t *input, int16_t *output,
- int pitch, TX_TYPE tx_type) {
+ int input_pitch, TX_TYPE tx_type) {
+ int i, j;
int16_t out[16 * 16];
int16_t *outptr = out;
- const int half_pitch = pitch >> 1;
- int i, j;
int16_t temp_in[16], temp_out[16];
const transform_2d ht = IHT_16[tx_type];
// Rows
for (i = 0; i < 16; ++i) {
ht.rows(input, outptr);
- input += half_pitch;
+ input += input_pitch;
outptr += 16;
}
@@ -912,7 +909,7 @@ void vp9_short_idct1_16x16_c(int16_t *input, int16_t *output) {
output[0] = ROUND_POWER_OF_TWO(out, 6);
}
-void idct32_1d(int16_t *input, int16_t *output) {
+static void idct32_1d(int16_t *input, int16_t *output) {
int16_t step1[32], step2[32];
int temp1, temp2;
diff --git a/vp9/common/vp9_invtrans.c b/vp9/common/vp9_invtrans.c
index c6b961894..d431ea24b 100644
--- a/vp9/common/vp9_invtrans.c
+++ b/vp9/common/vp9_invtrans.c
@@ -25,8 +25,7 @@ void vp9_inverse_transform_mby_4x4(MACROBLOCKD *xd) {
for (i = 0; i < 16; i++) {
TX_TYPE tx_type = get_tx_type_4x4(xd, &xd->block[i]);
if (tx_type != DCT_DCT) {
- vp9_short_iht4x4(xd->block[i].dqcoeff, xd->block[i].diff,
- 32, tx_type);
+ vp9_short_iht4x4(xd->block[i].dqcoeff, xd->block[i].diff, 16, tx_type);
} else {
vp9_inverse_transform_b_4x4(xd, i, 32);
}
@@ -58,8 +57,7 @@ void vp9_inverse_transform_mby_8x8(MACROBLOCKD *xd) {
for (i = 0; i < 9; i += 8) {
TX_TYPE tx_type = get_tx_type_8x8(xd, &xd->block[i]);
if (tx_type != DCT_DCT) {
- vp9_short_iht8x8(xd->block[i].dqcoeff, xd->block[i].diff,
- 32, tx_type);
+ vp9_short_iht8x8(xd->block[i].dqcoeff, xd->block[i].diff, 16, tx_type);
} else {
vp9_inverse_transform_b_8x8(&blockd[i].dqcoeff[0],
&blockd[i].diff[0], 32);
@@ -69,7 +67,7 @@ void vp9_inverse_transform_mby_8x8(MACROBLOCKD *xd) {
TX_TYPE tx_type = get_tx_type_8x8(xd, &xd->block[i]);
if (tx_type != DCT_DCT) {
vp9_short_iht8x8(xd->block[i + 2].dqcoeff, xd->block[i].diff,
- 32, tx_type);
+ 16, tx_type);
} else {
vp9_inverse_transform_b_8x8(&blockd[i + 2].dqcoeff[0],
&blockd[i].diff[0], 32);
@@ -101,7 +99,7 @@ void vp9_inverse_transform_mby_16x16(MACROBLOCKD *xd) {
BLOCKD *bd = &xd->block[0];
TX_TYPE tx_type = get_tx_type_16x16(xd, bd);
if (tx_type != DCT_DCT) {
- vp9_short_iht16x16(bd->dqcoeff, bd->diff, 32, tx_type);
+ vp9_short_iht16x16(bd->dqcoeff, bd->diff, 16, tx_type);
} else {
vp9_inverse_transform_b_16x16(&xd->block[0].dqcoeff[0],
&xd->block[0].diff[0], 32);
diff --git a/vp9/decoder/vp9_dequantize.c b/vp9/decoder/vp9_dequantize.c
index 0e205bdda..1d6c66afd 100644
--- a/vp9/decoder/vp9_dequantize.c
+++ b/vp9/decoder/vp9_dequantize.c
@@ -65,7 +65,7 @@ void vp9_ht_dequant_idct_add_c(TX_TYPE tx_type, int16_t *input,
for (i = 0; i < 16; i++)
input[i] = dq[i] * input[i];
- vp9_short_iht4x4(input, output, 8, tx_type);
+ vp9_short_iht4x4(input, output, 4, tx_type);
vpx_memset(input, 0, 32);
add_residual(diff_ptr, pred, pitch, dest, stride, 4, 4);
@@ -86,7 +86,7 @@ void vp9_ht_dequant_idct_add_8x8_c(TX_TYPE tx_type, int16_t *input,
for (i = 1; i < 64; i++)
input[i] *= dq[1];
- vp9_short_iht8x8(input, output, 16, tx_type);
+ vp9_short_iht8x8(input, output, 8, tx_type);
vpx_memset(input, 0, 128);
add_residual(diff_ptr, pred, pitch, dest, stride, 8, 8);
@@ -247,7 +247,7 @@ void vp9_ht_dequant_idct_add_16x16_c(TX_TYPE tx_type, int16_t *input,
input[i] *= dq[1];
// inverse hybrid transform
- vp9_short_iht16x16(input, output, 32, tx_type);
+ vp9_short_iht16x16(input, output, 16, tx_type);
// the idct halves ( >> 1) the pitch
// vp9_short_idct16x16_c(input, output, 32);
diff --git a/vp9/encoder/vp9_dct.c b/vp9/encoder/vp9_dct.c
index e2f3e2677..9c2203dea 100644
--- a/vp9/encoder/vp9_dct.c
+++ b/vp9/encoder/vp9_dct.c
@@ -105,7 +105,6 @@ void vp9_short_fht4x4_c(int16_t *input, int16_t *output,
int pitch, TX_TYPE tx_type) {
int16_t out[4 * 4];
int16_t *outptr = &out[0];
- const int short_pitch = pitch >> 1;
int i, j;
int16_t temp_in[4], temp_out[4];
@@ -137,7 +136,7 @@ void vp9_short_fht4x4_c(int16_t *input, int16_t *output,
// column transform
for (i = 0; i < 4; ++i) {
for (j = 0; j < 4; ++j)
- temp_in[j] = input[j * short_pitch + i] << 4;
+ temp_in[j] = input[j * pitch + i] << 4;
if (i == 0 && temp_in[0])
temp_in[0] += 1;
fwdc(temp_in, temp_out);
@@ -308,7 +307,6 @@ void vp9_short_fht8x8_c(int16_t *input, int16_t *output,
int pitch, TX_TYPE tx_type) {
int16_t out[64];
int16_t *outptr = &out[0];
- const int short_pitch = pitch >> 1;
int i, j;
int16_t temp_in[8], temp_out[8];
@@ -339,7 +337,7 @@ void vp9_short_fht8x8_c(int16_t *input, int16_t *output,
// column transform
for (i = 0; i < 8; ++i) {
for (j = 0; j < 8; ++j)
- temp_in[j] = input[j * short_pitch + i] << 2;
+ temp_in[j] = input[j * pitch + i] << 2;
fwdc(temp_in, temp_out);
for (j = 0; j < 8; ++j)
outptr[j * 8 + i] = temp_out[j];
@@ -697,7 +695,6 @@ void vp9_short_fht16x16_c(int16_t *input, int16_t *output,
int pitch, TX_TYPE tx_type) {
int16_t out[256];
int16_t *outptr = &out[0];
- const int short_pitch = pitch >> 1;
int i, j;
int16_t temp_in[16], temp_out[16];
@@ -728,7 +725,7 @@ void vp9_short_fht16x16_c(int16_t *input, int16_t *output,
// column transform
for (i = 0; i < 16; ++i) {
for (j = 0; j < 16; ++j)
- temp_in[j] = input[j * short_pitch + i] << 2;
+ temp_in[j] = input[j * pitch + i] << 2;
fwdc(temp_in, temp_out);
for (j = 0; j < 16; ++j)
outptr[j * 16 + i] = (temp_out[j] + 1 + (temp_out[j] > 0)) >> 2;
diff --git a/vp9/encoder/vp9_encodeintra.c b/vp9/encoder/vp9_encodeintra.c
index ef64db1db..43bb4640c 100644
--- a/vp9/encoder/vp9_encodeintra.c
+++ b/vp9/encoder/vp9_encodeintra.c
@@ -54,9 +54,9 @@ void vp9_encode_intra4x4block(MACROBLOCK *x, int ib) {
tx_type = get_tx_type_4x4(&x->e_mbd, b);
if (tx_type != DCT_DCT) {
- vp9_short_fht4x4(be->src_diff, be->coeff, 32, tx_type);
+ vp9_short_fht4x4(be->src_diff, be->coeff, 16, tx_type);
vp9_ht_quantize_b_4x4(be, b, tx_type);
- vp9_short_iht4x4(b->dqcoeff, b->diff, 32, tx_type);
+ vp9_short_iht4x4(b->dqcoeff, b->diff, 16, tx_type);
} else {
x->fwd_txm4x4(be->src_diff, be->coeff, 32);
x->quantize_b_4x4(be, b) ;
@@ -149,10 +149,10 @@ void vp9_encode_intra8x8(MACROBLOCK *x, int ib) {
tx_type = get_tx_type_8x8(xd, &xd->block[ib]);
if (tx_type != DCT_DCT) {
- vp9_short_fht8x8(be->src_diff, (x->block + idx)->coeff, 32, tx_type);
+ vp9_short_fht8x8(be->src_diff, (x->block + idx)->coeff, 16, tx_type);
x->quantize_b_8x8(x->block + idx, xd->block + idx);
vp9_short_iht8x8(xd->block[idx].dqcoeff, xd->block[ib].diff,
- 32, tx_type);
+ 16, tx_type);
} else {
x->fwd_txm8x8(be->src_diff, (x->block + idx)->coeff, 32);
x->quantize_b_8x8(x->block + idx, xd->block + idx);
@@ -164,9 +164,9 @@ void vp9_encode_intra8x8(MACROBLOCK *x, int ib) {
be = &x->block[ib + iblock[i]];
tx_type = get_tx_type_4x4(xd, b);
if (tx_type != DCT_DCT) {
- vp9_short_fht4x4(be->src_diff, be->coeff, 32, tx_type);
+ vp9_short_fht4x4(be->src_diff, be->coeff, 16, tx_type);
vp9_ht_quantize_b_4x4(be, b, tx_type);
- vp9_short_iht4x4(b->dqcoeff, b->diff, 32, tx_type);
+ vp9_short_iht4x4(b->dqcoeff, b->diff, 16, tx_type);
} else if (!(i & 1) && get_tx_type_4x4(xd, b + 1) == DCT_DCT) {
x->fwd_txm8x4(be->src_diff, be->coeff, 32);
x->quantize_b_4x4_pair(be, be + 1, b, b + 1);
diff --git a/vp9/encoder/vp9_encodemb.c b/vp9/encoder/vp9_encodemb.c
index 61516ddec..a753bf40f 100644
--- a/vp9/encoder/vp9_encodemb.c
+++ b/vp9/encoder/vp9_encodemb.c
@@ -174,7 +174,7 @@ void vp9_transform_mby_4x4(MACROBLOCK *x) {
BLOCK *b = &x->block[i];
TX_TYPE tx_type = get_tx_type_4x4(xd, &xd->block[i]);
if (tx_type != DCT_DCT) {
- vp9_short_fht4x4(b->src_diff, b->coeff, 32, tx_type);
+ vp9_short_fht4x4(b->src_diff, b->coeff, 16, tx_type);
} else if (!(i & 1) && get_tx_type_4x4(xd, &xd->block[i + 1]) == DCT_DCT) {
x->fwd_txm8x4(&x->block[i].src_diff[0],
&x->block[i].coeff[0], 32);
@@ -209,7 +209,7 @@ void vp9_transform_mby_8x8(MACROBLOCK *x) {
BLOCK *b = &x->block[i];
tx_type = get_tx_type_8x8(xd, &xd->block[i]);
if (tx_type != DCT_DCT) {
- vp9_short_fht8x8(b->src_diff, b->coeff, 32, tx_type);
+ vp9_short_fht8x8(b->src_diff, b->coeff, 16, tx_type);
} else {
x->fwd_txm8x8(&x->block[i].src_diff[0],
&x->block[i].coeff[0], 32);
@@ -219,7 +219,7 @@ void vp9_transform_mby_8x8(MACROBLOCK *x) {
BLOCK *b = &x->block[i];
tx_type = get_tx_type_8x8(xd, &xd->block[i]);
if (tx_type != DCT_DCT) {
- vp9_short_fht8x8(b->src_diff, (b + 2)->coeff, 32, tx_type);
+ vp9_short_fht8x8(b->src_diff, (b + 2)->coeff, 16, tx_type);
} else {
x->fwd_txm8x8(&x->block[i].src_diff[0],
&x->block[i + 2].coeff[0], 32);
@@ -247,7 +247,7 @@ void vp9_transform_mby_16x16(MACROBLOCK *x) {
TX_TYPE tx_type = get_tx_type_16x16(xd, &xd->block[0]);
vp9_clear_system_state();
if (tx_type != DCT_DCT) {
- vp9_short_fht16x16(b->src_diff, b->coeff, 32, tx_type);
+ vp9_short_fht16x16(b->src_diff, b->coeff, 16, tx_type);
} else {
x->fwd_txm16x16(&x->block[0].src_diff[0],
&x->block[0].coeff[0], 32);
diff --git a/vp9/encoder/vp9_rdopt.c b/vp9/encoder/vp9_rdopt.c
index d26b5ae7b..6e1122f3e 100644
--- a/vp9/encoder/vp9_rdopt.c
+++ b/vp9/encoder/vp9_rdopt.c
@@ -402,6 +402,12 @@ static INLINE int cost_coeffs(MACROBLOCK *mb,
unsigned int (*token_costs)[PREV_COEF_CONTEXTS][MAX_ENTROPY_TOKENS] =
mb->token_costs[tx_size][type][ref];
ENTROPY_CONTEXT a_ec = *a, l_ec = *l;
+#if CONFIG_CNVCONTEXT
+ ENTROPY_CONTEXT *const a1 = a +
+ sizeof(ENTROPY_CONTEXT_PLANES)/sizeof(ENTROPY_CONTEXT);
+ ENTROPY_CONTEXT *const l1 = l +
+ sizeof(ENTROPY_CONTEXT_PLANES)/sizeof(ENTROPY_CONTEXT);
+#endif
switch (tx_size) {
case TX_4X4:
@@ -416,6 +422,10 @@ static INLINE int cost_coeffs(MACROBLOCK *mb,
}
break;
case TX_8X8:
+#if CONFIG_CNVCONTEXT
+ a_ec = (a[0] + a[1]) != 0;
+ l_ec = (l[0] + l[1]) != 0;
+#endif
scan = vp9_default_zig_zag1d_8x8;
seg_eob = 64;
break;
@@ -425,12 +435,27 @@ static INLINE int cost_coeffs(MACROBLOCK *mb,
if (type == PLANE_TYPE_UV) {
const int uv_idx = ib - 16;
qcoeff_ptr = xd->sb_coeff_data.qcoeff + 1024 + 64 * uv_idx;
+#if CONFIG_CNVCONTEXT
+ a_ec = (a[0] + a[1] + a1[0] + a1[1]) != 0;
+ l_ec = (l[0] + l[1] + l1[0] + l1[1]) != 0;
+ } else {
+ a_ec = (a[0] + a[1] + a[2] + a[3]) != 0;
+ l_ec = (l[0] + l[1] + l[2] + l[3]) != 0;
+#endif
}
break;
case TX_32X32:
scan = vp9_default_zig_zag1d_32x32;
seg_eob = 1024;
qcoeff_ptr = xd->sb_coeff_data.qcoeff;
+#if CONFIG_CNVCONTEXT
+ a_ec = a[0] + a[1] + a[2] + a[3] +
+ a1[0] + a1[1] + a1[2] + a1[3];
+ l_ec = l[0] + l[1] + l[2] + l[3] +
+ l1[0] + l1[1] + l1[2] + l1[3];
+ a_ec = a_ec != 0;
+ l_ec = l_ec != 0;
+#endif
break;
default:
abort();
@@ -459,6 +484,22 @@ static INLINE int cost_coeffs(MACROBLOCK *mb,
// is eob first coefficient;
pt = (c > 0);
*a = *l = pt;
+#if CONFIG_CNVCONTEXT
+ if (tx_size >= TX_8X8) {
+ a[1] = l[1] = pt;
+ if (tx_size >= TX_16X16) {
+ if (type == PLANE_TYPE_UV) {
+ a1[0] = a1[1] = l1[0] = l1[1] = pt;
+ } else {
+ a[2] = a[3] = l[2] = l[3] = pt;
+ if (tx_size >= TX_32X32) {
+ a1[0] = a1[1] = a1[2] = a1[3] = pt;
+ l1[0] = l1[1] = l1[2] = l1[3] = pt;
+ }
+ }
+ }
+ }
+#endif
return cost;
}
@@ -701,15 +742,15 @@ static void copy_predictor(uint8_t *dst, const uint8_t *predictor) {
static int rdcost_sby_32x32(MACROBLOCK *x, int backup) {
MACROBLOCKD * const xd = &x->e_mbd;
- ENTROPY_CONTEXT_PLANES t_above, t_left;
+ ENTROPY_CONTEXT_PLANES t_above[2], t_left[2];
ENTROPY_CONTEXT *ta, *tl;
if (backup) {
ta = (ENTROPY_CONTEXT *) &t_above,
tl = (ENTROPY_CONTEXT *) &t_left;
- vpx_memcpy(&t_above, xd->above_context, sizeof(ENTROPY_CONTEXT_PLANES));
- vpx_memcpy(&t_left, xd->left_context, sizeof(ENTROPY_CONTEXT_PLANES));
+ vpx_memcpy(&t_above, xd->above_context, sizeof(ENTROPY_CONTEXT_PLANES) * 2);
+ vpx_memcpy(&t_left, xd->left_context, sizeof(ENTROPY_CONTEXT_PLANES) * 2);
} else {
ta = (ENTROPY_CONTEXT *) xd->above_context;
tl = (ENTROPY_CONTEXT *) xd->left_context;
@@ -1013,7 +1054,7 @@ static int64_t rd_pick_intra4x4block(VP9_COMP *cpi, MACROBLOCK *x, BLOCK *be,
b->bmi.as_mode.first = mode;
tx_type = get_tx_type_4x4(xd, b);
if (tx_type != DCT_DCT) {
- vp9_short_fht4x4(be->src_diff, be->coeff, 32, tx_type);
+ vp9_short_fht4x4(be->src_diff, be->coeff, 16, tx_type);
vp9_ht_quantize_b_4x4(be, b, tx_type);
} else {
x->fwd_txm4x4(be->src_diff, be->coeff, 32);
@@ -1046,7 +1087,7 @@ static int64_t rd_pick_intra4x4block(VP9_COMP *cpi, MACROBLOCK *x, BLOCK *be,
// inverse transform
if (best_tx_type != DCT_DCT)
- vp9_short_iht4x4(best_dqcoeff, b->diff, 32, best_tx_type);
+ vp9_short_iht4x4(best_dqcoeff, b->diff, 16, best_tx_type);
else
xd->inv_txm4x4(best_dqcoeff, b->diff, 32);
@@ -1279,8 +1320,9 @@ static int64_t rd_pick_intra8x8block(VP9_COMP *cpi, MACROBLOCK *x, int ib,
int distortion = 0, rate = 0;
BLOCK *be = x->block + ib;
BLOCKD *b = xd->block + ib;
- ENTROPY_CONTEXT ta0, ta1, besta0 = 0, besta1 = 0;
- ENTROPY_CONTEXT tl0, tl1, bestl0 = 0, bestl1 = 0;
+ ENTROPY_CONTEXT_PLANES ta, tl;
+ ENTROPY_CONTEXT *ta0, *ta1, besta0 = 0, besta1 = 0;
+ ENTROPY_CONTEXT *tl0, *tl1, bestl0 = 0, bestl1 = 0;
/*
* The predictor buffer is a 2d buffer with a stride of 16. Create
@@ -1309,7 +1351,7 @@ static int64_t rd_pick_intra8x8block(VP9_COMP *cpi, MACROBLOCK *x, int ib,
if (xd->mode_info_context->mbmi.txfm_size == TX_8X8) {
TX_TYPE tx_type = get_tx_type_8x8(xd, b);
if (tx_type != DCT_DCT)
- vp9_short_fht8x8(be->src_diff, (x->block + idx)->coeff, 32, tx_type);
+ vp9_short_fht8x8(be->src_diff, (x->block + idx)->coeff, 16, tx_type);
else
x->fwd_txm8x8(be->src_diff, (x->block + idx)->coeff, 32);
x->quantize_b_8x8(x->block + idx, xd->block + idx);
@@ -1317,23 +1359,29 @@ static int64_t rd_pick_intra8x8block(VP9_COMP *cpi, MACROBLOCK *x, int ib,
// compute quantization mse of 8x8 block
distortion = vp9_block_error_c((x->block + idx)->coeff,
(xd->block + idx)->dqcoeff, 64);
- ta0 = a[vp9_block2above[TX_8X8][idx]];
- tl0 = l[vp9_block2left[TX_8X8][idx]];
+
+ vpx_memcpy(&ta, a, sizeof(ENTROPY_CONTEXT_PLANES));
+ vpx_memcpy(&tl, l, sizeof(ENTROPY_CONTEXT_PLANES));
+
+ ta0 = ((ENTROPY_CONTEXT*)&ta) + vp9_block2above[TX_8X8][idx];
+ tl0 = ((ENTROPY_CONTEXT*)&tl) + vp9_block2left[TX_8X8][idx];
+ ta1 = ta0 + 1;
+ tl1 = tl0 + 1;
rate_t = cost_coeffs(x, xd->block + idx, PLANE_TYPE_Y_WITH_DC,
- &ta0, &tl0, TX_8X8);
+ ta0, tl0, TX_8X8);
rate += rate_t;
- ta1 = ta0;
- tl1 = tl0;
} else {
static const int iblock[4] = {0, 1, 4, 5};
TX_TYPE tx_type;
int i;
- ta0 = a[vp9_block2above[TX_4X4][ib]];
- ta1 = a[vp9_block2above[TX_4X4][ib + 1]];
- tl0 = l[vp9_block2left[TX_4X4][ib]];
- tl1 = l[vp9_block2left[TX_4X4][ib + 4]];
+ vpx_memcpy(&ta, a, sizeof(ENTROPY_CONTEXT_PLANES));
+ vpx_memcpy(&tl, l, sizeof(ENTROPY_CONTEXT_PLANES));
+ ta0 = ((ENTROPY_CONTEXT*)&ta) + vp9_block2above[TX_4X4][ib];
+ tl0 = ((ENTROPY_CONTEXT*)&tl) + vp9_block2left[TX_4X4][ib];
+ ta1 = ta0 + 1;
+ tl1 = tl0 + 1;
distortion = 0;
rate_t = 0;
for (i = 0; i < 4; ++i) {
@@ -1342,7 +1390,7 @@ static int64_t rd_pick_intra8x8block(VP9_COMP *cpi, MACROBLOCK *x, int ib,
be = &x->block[ib + iblock[i]];
tx_type = get_tx_type_4x4(xd, b);
if (tx_type != DCT_DCT) {
- vp9_short_fht4x4(be->src_diff, be->coeff, 32, tx_type);
+ vp9_short_fht4x4(be->src_diff, be->coeff, 16, tx_type);
vp9_ht_quantize_b_4x4(be, b, tx_type);
} else if (!(i & 1) && get_tx_type_4x4(xd, b + 1) == DCT_DCT) {
x->fwd_txm8x4(be->src_diff, be->coeff, 32);
@@ -1354,15 +1402,13 @@ static int64_t rd_pick_intra8x8block(VP9_COMP *cpi, MACROBLOCK *x, int ib,
}
distortion += vp9_block_error_c(be->coeff, b->dqcoeff, 16 << do_two);
rate_t += cost_coeffs(x, b, PLANE_TYPE_Y_WITH_DC,
- // i&1 ? &ta1 : &ta0, i&2 ? &tl1 : &tl0,
- &ta0, &tl0,
+ i&1 ? ta1 : ta0, i&2 ? tl1 : tl0,
TX_4X4);
if (do_two) {
+ i++;
rate_t += cost_coeffs(x, b + 1, PLANE_TYPE_Y_WITH_DC,
- // i&1 ? &ta1 : &ta0, i&2 ? &tl1 : &tl0,
- &ta0, &tl0,
+ i&1 ? ta1 : ta0, i&2 ? tl1 : tl0,
TX_4X4);
- i++;
}
}
b = &xd->block[ib];
@@ -1376,10 +1422,10 @@ static int64_t rd_pick_intra8x8block(VP9_COMP *cpi, MACROBLOCK *x, int ib,
*bestrate = rate;
*bestratey = rate_t;
*bestdistortion = distortion;
- besta0 = ta0;
- besta1 = ta1;
- bestl0 = tl0;
- bestl1 = tl1;
+ besta0 = *ta0;
+ besta1 = *ta1;
+ bestl0 = *tl0;
+ bestl1 = *tl1;
best_rd = this_rd;
*best_mode = mode;
copy_predictor_8x8(best_predictor, b->predictor);
@@ -1532,12 +1578,12 @@ static int rd_cost_sbuv_16x16(MACROBLOCK *x, int backup) {
int b;
int cost = 0;
MACROBLOCKD *const xd = &x->e_mbd;
- ENTROPY_CONTEXT_PLANES t_above, t_left;
+ ENTROPY_CONTEXT_PLANES t_above[2], t_left[2];
ENTROPY_CONTEXT *ta, *tl;
if (backup) {
- vpx_memcpy(&t_above, xd->above_context, sizeof(ENTROPY_CONTEXT_PLANES));
- vpx_memcpy(&t_left, xd->left_context, sizeof(ENTROPY_CONTEXT_PLANES));
+ vpx_memcpy(&t_above, xd->above_context, sizeof(ENTROPY_CONTEXT_PLANES) * 2);
+ vpx_memcpy(&t_left, xd->left_context, sizeof(ENTROPY_CONTEXT_PLANES) * 2);
ta = (ENTROPY_CONTEXT *) &t_above;
tl = (ENTROPY_CONTEXT *) &t_left;
generated by cgit v1.2.3 (git 2.25.1) at 2024-10-10 21:22:23 +0000