Commit 455f2de2 authored by Jingning Han's avatar Jingning Han
Browse files

Tune tokenization/detokenization flow for speed-up

This commit optimizes the tokenization and detokenization operational
flow for speed-up. It makes the coding process about 0.3% faster at
speed 0.

Change-Id: I28008df7482874e4b5f237f2d418ff82a249dd56
parent b5323ed8
...@@ -118,22 +118,22 @@ static int decode_coefs(VP9_COMMON *cm, const MACROBLOCKD *xd, ...@@ -118,22 +118,22 @@ static int decode_coefs(VP9_COMMON *cm, const MACROBLOCKD *xd,
} }
case TX_8X8: { case TX_8X8: {
scan = get_scan_8x8(get_tx_type_8x8(type, xd)); scan = get_scan_8x8(get_tx_type_8x8(type, xd));
above_ec = (A[0] + A[1]) != 0; above_ec = !!*(uint16_t *)A;
left_ec = (L[0] + L[1]) != 0; left_ec = !!*(uint16_t *)L;
band_translate = vp9_coefband_trans_8x8plus; band_translate = vp9_coefband_trans_8x8plus;
break; break;
} }
case TX_16X16: { case TX_16X16: {
scan = get_scan_16x16(get_tx_type_16x16(type, xd)); scan = get_scan_16x16(get_tx_type_16x16(type, xd));
above_ec = (A[0] + A[1] + A[2] + A[3]) != 0; above_ec = !!*(uint32_t *)A;
left_ec = (L[0] + L[1] + L[2] + L[3]) != 0; left_ec = !!*(uint32_t *)L;
band_translate = vp9_coefband_trans_8x8plus; band_translate = vp9_coefband_trans_8x8plus;
break; break;
} }
case TX_32X32: case TX_32X32:
scan = vp9_default_scan_32x32; scan = vp9_default_scan_32x32;
above_ec = (A[0] + A[1] + A[2] + A[3] + A[4] + A[5] + A[6] + A[7]) != 0; above_ec = !!*(uint64_t *)A;
left_ec = (L[0] + L[1] + L[2] + L[3] + L[4] + L[5] + L[6] + L[7]) != 0; left_ec = !!*(uint64_t *)L;
band_translate = vp9_coefband_trans_8x8plus; band_translate = vp9_coefband_trans_8x8plus;
break; break;
} }
......
...@@ -136,9 +136,7 @@ static void tokenize_b(int plane, int block, BLOCK_SIZE_TYPE bsize, ...@@ -136,9 +136,7 @@ static void tokenize_b(int plane, int block, BLOCK_SIZE_TYPE bsize,
const int eob = xd->plane[plane].eobs[block]; const int eob = xd->plane[plane].eobs[block];
const PLANE_TYPE type = xd->plane[plane].plane_type; const PLANE_TYPE type = xd->plane[plane].plane_type;
const int16_t *qcoeff_ptr = BLOCK_OFFSET(xd->plane[plane].qcoeff, block, 16); const int16_t *qcoeff_ptr = BLOCK_OFFSET(xd->plane[plane].qcoeff, block, 16);
const BLOCK_SIZE_TYPE sb_type = (mbmi->sb_type < BLOCK_SIZE_SB8X8) ? const int bwl = b_width_log2(bsize);
BLOCK_SIZE_SB8X8 : mbmi->sb_type;
const int bwl = b_width_log2(sb_type);
const int off = block >> (2 * tx_size); const int off = block >> (2 * tx_size);
const int mod = bwl - tx_size - xd->plane[plane].subsampling_x; const int mod = bwl - tx_size - xd->plane[plane].subsampling_x;
const int aoff = (off & ((1 << mod) - 1)) << tx_size; const int aoff = (off & ((1 << mod) - 1)) << tx_size;
...@@ -168,22 +166,22 @@ static void tokenize_b(int plane, int block, BLOCK_SIZE_TYPE bsize, ...@@ -168,22 +166,22 @@ static void tokenize_b(int plane, int block, BLOCK_SIZE_TYPE bsize,
band_translate = vp9_coefband_trans_4x4; band_translate = vp9_coefband_trans_4x4;
break; break;
case TX_8X8: case TX_8X8:
above_ec = (A[0] + A[1]) != 0; above_ec = !!*(uint16_t *)A;
left_ec = (L[0] + L[1]) != 0; left_ec = !!*(uint16_t *)L;
seg_eob = 64; seg_eob = 64;
scan = get_scan_8x8(get_tx_type_8x8(type, xd)); scan = get_scan_8x8(get_tx_type_8x8(type, xd));
band_translate = vp9_coefband_trans_8x8plus; band_translate = vp9_coefband_trans_8x8plus;
break; break;
case TX_16X16: case TX_16X16:
above_ec = (A[0] + A[1] + A[2] + A[3]) != 0; above_ec = !!*(uint32_t *)A;
left_ec = (L[0] + L[1] + L[2] + L[3]) != 0; left_ec = !!*(uint32_t *)L;
seg_eob = 256; seg_eob = 256;
scan = get_scan_16x16(get_tx_type_16x16(type, xd)); scan = get_scan_16x16(get_tx_type_16x16(type, xd));
band_translate = vp9_coefband_trans_8x8plus; band_translate = vp9_coefband_trans_8x8plus;
break; break;
case TX_32X32: case TX_32X32:
above_ec = (A[0] + A[1] + A[2] + A[3] + A[4] + A[5] + A[6] + A[7]) != 0; above_ec = !!*(uint64_t *)A;
left_ec = (L[0] + L[1] + L[2] + L[3] + L[4] + L[5] + L[6] + L[7]) != 0; left_ec = !!*(uint64_t *)L;
seg_eob = 1024; seg_eob = 1024;
scan = vp9_default_scan_32x32; scan = vp9_default_scan_32x32;
band_translate = vp9_coefband_trans_8x8plus; band_translate = vp9_coefband_trans_8x8plus;
...@@ -225,7 +223,7 @@ static void tokenize_b(int plane, int block, BLOCK_SIZE_TYPE bsize, ...@@ -225,7 +223,7 @@ static void tokenize_b(int plane, int block, BLOCK_SIZE_TYPE bsize,
if (!t->skip_eob_node) if (!t->skip_eob_node)
++cpi->common.counts.eob_branch[tx_size][type][ref][band][pt]; ++cpi->common.counts.eob_branch[tx_size][type][ref][band][pt];
} }
token_cache[scan[c]] = vp9_pt_energy_class[token]; token_cache[rc] = vp9_pt_energy_class[token];
++t; ++t;
} while (c < eob && ++c < seg_eob); } while (c < eob && ++c < seg_eob);
...@@ -233,9 +231,8 @@ static void tokenize_b(int plane, int block, BLOCK_SIZE_TYPE bsize, ...@@ -233,9 +231,8 @@ static void tokenize_b(int plane, int block, BLOCK_SIZE_TYPE bsize,
if (xd->mb_to_right_edge < 0 || xd->mb_to_bottom_edge < 0) { if (xd->mb_to_right_edge < 0 || xd->mb_to_bottom_edge < 0) {
set_contexts_on_border(xd, bsize, plane, tx_size, c, aoff, loff, A, L); set_contexts_on_border(xd, bsize, plane, tx_size, c, aoff, loff, A, L);
} else { } else {
for (pt = 0; pt < (1 << tx_size); pt++) { vpx_memset(A, c > 0, sizeof(ENTROPY_CONTEXT) * (1 << tx_size));
A[pt] = L[pt] = c > 0; vpx_memset(L, c > 0, sizeof(ENTROPY_CONTEXT) * (1 << tx_size));
}
} }
} }
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment