Prefix three global arrays with av1_ BUG=aomedia:1540 Change-Id: If34f10fa5405f75b63f9946651e9cff0349686a7
diff --git a/av1/common/arm/av1_inv_txfm_neon.c b/av1/common/arm/av1_inv_txfm_neon.c index 7a23174..3f3833f 100644 --- a/av1/common/arm/av1_inv_txfm_neon.c +++ b/av1/common/arm/av1_inv_txfm_neon.c
@@ -3686,11 +3686,11 @@ int16x8_t b[32 * 4]; int eobx, eoby; get_eobx_eoby_scan_default(&eobx, &eoby, tx_size, eob); - const int8_t *shift = inv_txfm_shift_ls[tx_size]; + const int8_t *shift = av1_inv_txfm_shift_ls[tx_size]; const int txw_idx = get_txw_idx(tx_size); const int txh_idx = get_txh_idx(tx_size); - const int cos_bit_col = inv_cos_bit_col[txw_idx][txh_idx]; - const int cos_bit_row = inv_cos_bit_row[txw_idx][txh_idx]; + const int cos_bit_col = av1_inv_cos_bit_col[txw_idx][txh_idx]; + const int cos_bit_row = av1_inv_cos_bit_row[txw_idx][txh_idx]; const int txfm_size_col = tx_size_wide[tx_size]; const int txfm_size_row = tx_size_high[tx_size]; lowbd_inv_txfm2d_memset_neon(&a[0], (txfm_size_col * (txfm_size_row) >> 3), @@ -3757,11 +3757,11 @@ int16x8_t b[16 * 2]; int eobx, eoby, ud_flip, lr_flip; get_eobx_eoby_scan_v_identity(&eobx, &eoby, tx_size, eob); - const int8_t *shift = inv_txfm_shift_ls[tx_size]; + const int8_t *shift = av1_inv_txfm_shift_ls[tx_size]; const int txw_idx = get_txw_idx(tx_size); const int txh_idx = get_txh_idx(tx_size); - const int cos_bit_col = inv_cos_bit_col[txw_idx][txh_idx]; - const int cos_bit_row = inv_cos_bit_row[txw_idx][txh_idx]; + const int cos_bit_col = av1_inv_cos_bit_col[txw_idx][txh_idx]; + const int cos_bit_row = av1_inv_cos_bit_row[txw_idx][txh_idx]; const int txfm_size_col = tx_size_wide[tx_size]; const int txfm_size_row = tx_size_high[tx_size]; lowbd_inv_txfm2d_memset_neon(&b[0], (txfm_size_col * (txfm_size_row) >> 3), @@ -3838,11 +3838,11 @@ int16x8_t b[16 * 2]; int eobx, eoby, ud_flip, lr_flip; get_eobx_eoby_scan_h_identity(&eobx, &eoby, tx_size, eob); - const int8_t *shift = inv_txfm_shift_ls[tx_size]; + const int8_t *shift = av1_inv_txfm_shift_ls[tx_size]; const int txw_idx = get_txw_idx(tx_size); const int txh_idx = get_txh_idx(tx_size); - const int cos_bit_col = inv_cos_bit_col[txw_idx][txh_idx]; - const int cos_bit_row = inv_cos_bit_row[txw_idx][txh_idx]; + const int cos_bit_col = av1_inv_cos_bit_col[txw_idx][txh_idx]; + const int cos_bit_row = av1_inv_cos_bit_row[txw_idx][txh_idx]; const int txfm_size_col = tx_size_wide[tx_size]; const int txfm_size_row = tx_size_high[tx_size]; lowbd_inv_txfm2d_memset_neon(&a[0], (txfm_size_col * (txfm_size_row) >> 3), @@ -3911,11 +3911,11 @@ DECLARE_ALIGNED(32, int, txfm_buf[4 * 4 + 8 + 8]); int32_t *temp_in = txfm_buf; - const int8_t *shift = inv_txfm_shift_ls[tx_size]; + const int8_t *shift = av1_inv_txfm_shift_ls[tx_size]; const int txw_idx = get_txw_idx(tx_size); const int txh_idx = get_txh_idx(tx_size); - const int cos_bit_row = inv_cos_bit_row[txw_idx][txh_idx]; - const int cos_bit_col = inv_cos_bit_col[txw_idx][txh_idx]; + const int cos_bit_row = av1_inv_cos_bit_row[txw_idx][txh_idx]; + const int cos_bit_col = av1_inv_cos_bit_col[txw_idx][txh_idx]; const int txfm_size_col = tx_size_wide[tx_size]; const int txfm_size_row = tx_size_high[tx_size]; const int buf_offset = AOMMAX(txfm_size_row, txfm_size_col); @@ -3973,11 +3973,11 @@ DECLARE_ALIGNED(32, int, txfm_buf[4 * 8 + 8 + 8]); int32_t *temp_in = txfm_buf; - const int8_t *shift = inv_txfm_shift_ls[tx_size]; + const int8_t *shift = av1_inv_txfm_shift_ls[tx_size]; const int txw_idx = get_txw_idx(tx_size); const int txh_idx = get_txh_idx(tx_size); - const int cos_bit_row = inv_cos_bit_row[txw_idx][txh_idx]; - const int cos_bit_col = inv_cos_bit_col[txw_idx][txh_idx]; + const int cos_bit_row = av1_inv_cos_bit_row[txw_idx][txh_idx]; + const int cos_bit_col = av1_inv_cos_bit_col[txw_idx][txh_idx]; const int txfm_size_col = tx_size_wide[tx_size]; const int txfm_size_row = tx_size_high[tx_size]; const int buf_offset = AOMMAX(txfm_size_row, txfm_size_col); @@ -4037,11 +4037,11 @@ DECLARE_ALIGNED(32, int, txfm_buf[8 * 4 + 8 + 8]); int32_t *temp_in = txfm_buf; - const int8_t *shift = inv_txfm_shift_ls[tx_size]; + const int8_t *shift = av1_inv_txfm_shift_ls[tx_size]; const int txw_idx = get_txw_idx(tx_size); const int txh_idx = get_txh_idx(tx_size); - const int cos_bit_row = inv_cos_bit_row[txw_idx][txh_idx]; - const int cos_bit_col = inv_cos_bit_col[txw_idx][txh_idx]; + const int cos_bit_row = av1_inv_cos_bit_row[txw_idx][txh_idx]; + const int cos_bit_col = av1_inv_cos_bit_col[txw_idx][txh_idx]; const int txfm_size_col = tx_size_wide[tx_size]; const int txfm_size_row = tx_size_high[tx_size]; const int buf_offset = AOMMAX(txfm_size_row, txfm_size_col); @@ -4101,11 +4101,11 @@ DECLARE_ALIGNED(32, int, txfm_buf[4 * 16 + 16 + 16]); int32_t *temp_in = txfm_buf; - const int8_t *shift = inv_txfm_shift_ls[tx_size]; + const int8_t *shift = av1_inv_txfm_shift_ls[tx_size]; const int txw_idx = get_txw_idx(tx_size); const int txh_idx = get_txh_idx(tx_size); - const int cos_bit_row = inv_cos_bit_row[txw_idx][txh_idx]; - const int cos_bit_col = inv_cos_bit_col[txw_idx][txh_idx]; + const int cos_bit_row = av1_inv_cos_bit_row[txw_idx][txh_idx]; + const int cos_bit_col = av1_inv_cos_bit_col[txw_idx][txh_idx]; const int txfm_size_col = tx_size_wide[tx_size]; const int txfm_size_row = tx_size_high[tx_size]; const int buf_offset = AOMMAX(txfm_size_row, txfm_size_col); @@ -4163,11 +4163,11 @@ DECLARE_ALIGNED(32, int, txfm_buf[16 * 4 + 16 + 16]); int32_t *temp_in = txfm_buf; - const int8_t *shift = inv_txfm_shift_ls[tx_size]; + const int8_t *shift = av1_inv_txfm_shift_ls[tx_size]; const int txw_idx = get_txw_idx(tx_size); const int txh_idx = get_txh_idx(tx_size); - const int cos_bit_row = inv_cos_bit_row[txw_idx][txh_idx]; - const int cos_bit_col = inv_cos_bit_col[txw_idx][txh_idx]; + const int cos_bit_row = av1_inv_cos_bit_row[txw_idx][txh_idx]; + const int cos_bit_col = av1_inv_cos_bit_col[txw_idx][txh_idx]; const int txfm_size_col = tx_size_wide[tx_size]; const int txfm_size_row = tx_size_high[tx_size]; const int buf_offset = AOMMAX(txfm_size_row, txfm_size_col); @@ -4225,11 +4225,11 @@ int16x8_t b[64 * 8]; int eobx, eoby, ud_flip, lr_flip; get_eobx_eoby_scan_default(&eobx, &eoby, tx_size, eob); - const int8_t *shift = inv_txfm_shift_ls[tx_size]; + const int8_t *shift = av1_inv_txfm_shift_ls[tx_size]; const int txw_idx = get_txw_idx(tx_size); const int txh_idx = get_txh_idx(tx_size); - const int cos_bit_col = inv_cos_bit_col[txw_idx][txh_idx]; - const int cos_bit_row = inv_cos_bit_row[txw_idx][txh_idx]; + const int cos_bit_col = av1_inv_cos_bit_col[txw_idx][txh_idx]; + const int cos_bit_row = av1_inv_cos_bit_row[txw_idx][txh_idx]; const int txfm_size_col = tx_size_wide[tx_size]; const int txfm_size_row = tx_size_high[tx_size]; const int rect_type = get_rect_tx_log_ratio(txfm_size_col, txfm_size_row);
diff --git a/av1/common/av1_inv_txfm1d_cfg.h b/av1/common/av1_inv_txfm1d_cfg.h index 7d80a00..47fedbd 100644 --- a/av1/common/av1_inv_txfm1d_cfg.h +++ b/av1/common/av1_inv_txfm1d_cfg.h
@@ -36,12 +36,12 @@ 7, // 64x16 transform }; -extern const int8_t *inv_txfm_shift_ls[TX_SIZES_ALL]; +extern const int8_t *av1_inv_txfm_shift_ls[TX_SIZES_ALL]; -// Values in both inv_cos_bit_col and inv_cos_bit_row are always 12 +// Values in both av1_inv_cos_bit_col and av1_inv_cos_bit_row are always 12 // for each valid row and col combination #define INV_COS_BIT 12 -extern const int8_t inv_cos_bit_col[5 /*row*/][5 /*col*/]; -extern const int8_t inv_cos_bit_row[5 /*row*/][5 /*col*/]; +extern const int8_t av1_inv_cos_bit_col[5 /*row*/][5 /*col*/]; +extern const int8_t av1_inv_cos_bit_row[5 /*row*/][5 /*col*/]; #endif // AOM_AV1_COMMON_AV1_INV_TXFM1D_CFG_H_
diff --git a/av1/common/av1_inv_txfm2d.c b/av1/common/av1_inv_txfm2d.c index 40200d2..92486aa 100644 --- a/av1/common/av1_inv_txfm2d.c +++ b/av1/common/av1_inv_txfm2d.c
@@ -149,7 +149,7 @@ static const int8_t inv_shift_16x64[2] = { -2, -4 }; static const int8_t inv_shift_64x16[2] = { -2, -4 }; -const int8_t *inv_txfm_shift_ls[TX_SIZES_ALL] = { +const int8_t *av1_inv_txfm_shift_ls[TX_SIZES_ALL] = { inv_shift_4x4, inv_shift_8x8, inv_shift_16x16, inv_shift_32x32, inv_shift_64x64, inv_shift_4x8, inv_shift_8x4, inv_shift_8x16, inv_shift_16x8, inv_shift_16x32, inv_shift_32x16, inv_shift_32x64, @@ -158,7 +158,7 @@ }; /* clang-format off */ -const int8_t inv_cos_bit_col[MAX_TXWH_IDX] // txw_idx +const int8_t av1_inv_cos_bit_col[MAX_TXWH_IDX] // txw_idx [MAX_TXWH_IDX] = { // txh_idx { INV_COS_BIT, INV_COS_BIT, INV_COS_BIT, 0, 0 }, { INV_COS_BIT, INV_COS_BIT, INV_COS_BIT, INV_COS_BIT, 0 }, @@ -167,7 +167,7 @@ { 0, 0, INV_COS_BIT, INV_COS_BIT, INV_COS_BIT } }; -const int8_t inv_cos_bit_row[MAX_TXWH_IDX] // txw_idx +const int8_t av1_inv_cos_bit_row[MAX_TXWH_IDX] // txw_idx [MAX_TXWH_IDX] = { // txh_idx { INV_COS_BIT, INV_COS_BIT, INV_COS_BIT, 0, 0 }, { INV_COS_BIT, INV_COS_BIT, INV_COS_BIT, INV_COS_BIT, 0 }, @@ -188,11 +188,11 @@ set_flip_cfg(tx_type, cfg); const TX_TYPE_1D tx_type_1d_col = vtx_tab[tx_type]; const TX_TYPE_1D tx_type_1d_row = htx_tab[tx_type]; - cfg->shift = inv_txfm_shift_ls[tx_size]; + cfg->shift = av1_inv_txfm_shift_ls[tx_size]; const int txw_idx = get_txw_idx(tx_size); const int txh_idx = get_txh_idx(tx_size); - cfg->cos_bit_col = inv_cos_bit_col[txw_idx][txh_idx]; - cfg->cos_bit_row = inv_cos_bit_row[txw_idx][txh_idx]; + cfg->cos_bit_col = av1_inv_cos_bit_col[txw_idx][txh_idx]; + cfg->cos_bit_row = av1_inv_cos_bit_row[txw_idx][txh_idx]; cfg->txfm_type_col = av1_txfm_type_ls[txh_idx][tx_type_1d_col]; if (cfg->txfm_type_col == TXFM_TYPE_ADST4) { memcpy(cfg->stage_range_col, iadst4_range, sizeof(iadst4_range));
diff --git a/av1/common/x86/av1_inv_txfm_avx2.c b/av1/common/x86/av1_inv_txfm_avx2.c index 3f5ad89..cae7504 100644 --- a/av1/common/x86/av1_inv_txfm_avx2.c +++ b/av1/common/x86/av1_inv_txfm_avx2.c
@@ -1614,11 +1614,11 @@ __m256i buf1[64 * 16]; int eobx, eoby; get_eobx_eoby_scan_default(&eobx, &eoby, tx_size, eob); - const int8_t *shift = inv_txfm_shift_ls[tx_size]; + const int8_t *shift = av1_inv_txfm_shift_ls[tx_size]; const int txw_idx = get_txw_idx(tx_size); const int txh_idx = get_txh_idx(tx_size); - const int cos_bit_col = inv_cos_bit_col[txw_idx][txh_idx]; - const int cos_bit_row = inv_cos_bit_row[txw_idx][txh_idx]; + const int cos_bit_col = av1_inv_cos_bit_col[txw_idx][txh_idx]; + const int cos_bit_row = av1_inv_cos_bit_row[txw_idx][txh_idx]; const int txfm_size_col = tx_size_wide[tx_size]; const int txfm_size_row = tx_size_high[tx_size]; const int buf_size_w_div16 = txfm_size_col >> 4; @@ -1754,7 +1754,7 @@ TX_SIZE tx_size, int32_t eob) { (void)eob; - const int8_t *shift = inv_txfm_shift_ls[tx_size]; + const int8_t *shift = av1_inv_txfm_shift_ls[tx_size]; const int txw_idx = get_txw_idx(tx_size); const int txh_idx = get_txh_idx(tx_size); const int txfm_size_col = tx_size_wide[tx_size]; @@ -1776,10 +1776,10 @@ TX_SIZE tx_size, int eob) { int eobx, eoby; get_eobx_eoby_scan_h_identity(&eobx, &eoby, tx_size, eob); - const int8_t *shift = inv_txfm_shift_ls[tx_size]; + const int8_t *shift = av1_inv_txfm_shift_ls[tx_size]; const int txw_idx = get_txw_idx(tx_size); const int txh_idx = get_txh_idx(tx_size); - const int cos_bit_col = inv_cos_bit_col[txw_idx][txh_idx]; + const int cos_bit_col = av1_inv_cos_bit_col[txw_idx][txh_idx]; const int txfm_size_col = tx_size_wide[tx_size]; const int txfm_size_row = tx_size_high[tx_size]; const int txfm_size_col_notzero = AOMMIN(32, txfm_size_col); @@ -1816,10 +1816,10 @@ __m256i buf1[64]; int eobx, eoby; get_eobx_eoby_scan_v_identity(&eobx, &eoby, tx_size, eob); - const int8_t *shift = inv_txfm_shift_ls[tx_size]; + const int8_t *shift = av1_inv_txfm_shift_ls[tx_size]; const int txw_idx = get_txw_idx(tx_size); const int txh_idx = get_txh_idx(tx_size); - const int cos_bit_row = inv_cos_bit_row[txw_idx][txh_idx]; + const int cos_bit_row = av1_inv_cos_bit_row[txw_idx][txh_idx]; const int txfm_size_col = tx_size_wide[tx_size]; const int txfm_size_row = tx_size_high[tx_size]; const int buf_size_w_div16 = txfm_size_col >> 4;
diff --git a/av1/common/x86/av1_inv_txfm_ssse3.c b/av1/common/x86/av1_inv_txfm_ssse3.c index 2208a91..d5f0527 100644 --- a/av1/common/x86/av1_inv_txfm_ssse3.c +++ b/av1/common/x86/av1_inv_txfm_ssse3.c
@@ -2419,7 +2419,7 @@ static INLINE void lowbd_inv_txfm2d_add_idtx_ssse3(const int32_t *input, uint8_t *output, int stride, TX_SIZE tx_size) { - const int8_t *shift = inv_txfm_shift_ls[tx_size]; + const int8_t *shift = av1_inv_txfm_shift_ls[tx_size]; const int txw_idx = get_txw_idx(tx_size); const int txh_idx = get_txh_idx(tx_size); const int txfm_size_col = tx_size_wide[tx_size]; @@ -2445,11 +2445,11 @@ (void)eob; __m128i buf[4]; const TX_SIZE tx_size = TX_4X4; - const int8_t *shift = inv_txfm_shift_ls[tx_size]; + const int8_t *shift = av1_inv_txfm_shift_ls[tx_size]; const int txw_idx = get_txw_idx(tx_size); const int txh_idx = get_txh_idx(tx_size); - const int cos_bit_row = inv_cos_bit_row[txw_idx][txh_idx]; - const int cos_bit_col = inv_cos_bit_col[txw_idx][txh_idx]; + const int cos_bit_row = av1_inv_cos_bit_row[txw_idx][txh_idx]; + const int cos_bit_col = av1_inv_cos_bit_col[txw_idx][txh_idx]; const int txfm_size_col = tx_size_wide[tx_size]; const int txfm_size_row = tx_size_high[tx_size]; @@ -2511,11 +2511,11 @@ __m128i buf1[64 * 8]; int eobx, eoby; get_eobx_eoby_scan_default(&eobx, &eoby, tx_size, eob); - const int8_t *shift = inv_txfm_shift_ls[tx_size]; + const int8_t *shift = av1_inv_txfm_shift_ls[tx_size]; const int txw_idx = get_txw_idx(tx_size); const int txh_idx = get_txh_idx(tx_size); - const int cos_bit_col = inv_cos_bit_col[txw_idx][txh_idx]; - const int cos_bit_row = inv_cos_bit_row[txw_idx][txh_idx]; + const int cos_bit_col = av1_inv_cos_bit_col[txw_idx][txh_idx]; + const int cos_bit_row = av1_inv_cos_bit_row[txw_idx][txh_idx]; const int txfm_size_col = tx_size_wide[tx_size]; const int txfm_size_row = tx_size_high[tx_size]; const int buf_size_w_div8 = txfm_size_col >> 3; @@ -2581,12 +2581,12 @@ static INLINE void lowbd_inv_txfm2d_add_h_identity_ssse3( const int32_t *input, uint8_t *output, int stride, TX_TYPE tx_type, TX_SIZE tx_size, int eob) { - const int8_t *shift = inv_txfm_shift_ls[tx_size]; + const int8_t *shift = av1_inv_txfm_shift_ls[tx_size]; int eobx, eoby; get_eobx_eoby_scan_h_identity(&eobx, &eoby, tx_size, eob); const int txw_idx = get_txw_idx(tx_size); const int txh_idx = get_txh_idx(tx_size); - const int cos_bit_col = inv_cos_bit_col[txw_idx][txh_idx]; + const int cos_bit_col = av1_inv_cos_bit_col[txw_idx][txh_idx]; const int txfm_size_col = tx_size_wide[tx_size]; const int txfm_size_row = tx_size_high[tx_size]; const int buf_size_w_div8 = (eobx + 8) >> 3; @@ -2627,10 +2627,10 @@ __m128i buf1[64]; int eobx, eoby; get_eobx_eoby_scan_v_identity(&eobx, &eoby, tx_size, eob); - const int8_t *shift = inv_txfm_shift_ls[tx_size]; + const int8_t *shift = av1_inv_txfm_shift_ls[tx_size]; const int txw_idx = get_txw_idx(tx_size); const int txh_idx = get_txh_idx(tx_size); - const int cos_bit_row = inv_cos_bit_row[txw_idx][txh_idx]; + const int cos_bit_row = av1_inv_cos_bit_row[txw_idx][txh_idx]; const int txfm_size_col = tx_size_wide[tx_size]; const int txfm_size_row = tx_size_high[tx_size]; const int buf_size_w_div8 = txfm_size_col >> 3; @@ -2717,11 +2717,11 @@ (void)eob; __m128i buf[8]; const TX_SIZE tx_size = TX_4X8; - const int8_t *shift = inv_txfm_shift_ls[tx_size]; + const int8_t *shift = av1_inv_txfm_shift_ls[tx_size]; const int txw_idx = get_txw_idx(tx_size); const int txh_idx = get_txh_idx(tx_size); - const int cos_bit_row = inv_cos_bit_row[txw_idx][txh_idx]; - const int cos_bit_col = inv_cos_bit_col[txw_idx][txh_idx]; + const int cos_bit_row = av1_inv_cos_bit_row[txw_idx][txh_idx]; + const int cos_bit_col = av1_inv_cos_bit_col[txw_idx][txh_idx]; const int txfm_size_col = tx_size_wide[tx_size]; const int txfm_size_row = tx_size_high[tx_size]; @@ -2757,11 +2757,11 @@ (void)eob; __m128i buf[8]; const TX_SIZE tx_size = TX_8X4; - const int8_t *shift = inv_txfm_shift_ls[tx_size]; + const int8_t *shift = av1_inv_txfm_shift_ls[tx_size]; const int txw_idx = get_txw_idx(tx_size); const int txh_idx = get_txh_idx(tx_size); - const int cos_bit_row = inv_cos_bit_row[txw_idx][txh_idx]; - const int cos_bit_col = inv_cos_bit_col[txw_idx][txh_idx]; + const int cos_bit_row = av1_inv_cos_bit_row[txw_idx][txh_idx]; + const int cos_bit_col = av1_inv_cos_bit_col[txw_idx][txh_idx]; const int txfm_size_col = tx_size_wide[tx_size]; const int txfm_size_row = tx_size_high[tx_size]; @@ -2797,11 +2797,11 @@ (void)eob; __m128i buf[16]; const TX_SIZE tx_size = TX_4X16; - const int8_t *shift = inv_txfm_shift_ls[tx_size]; + const int8_t *shift = av1_inv_txfm_shift_ls[tx_size]; const int txw_idx = get_txw_idx(tx_size); const int txh_idx = get_txh_idx(tx_size); - const int cos_bit_row = inv_cos_bit_row[txw_idx][txh_idx]; - const int cos_bit_col = inv_cos_bit_col[txw_idx][txh_idx]; + const int cos_bit_row = av1_inv_cos_bit_row[txw_idx][txh_idx]; + const int cos_bit_col = av1_inv_cos_bit_col[txw_idx][txh_idx]; const int txfm_size_col = tx_size_wide[tx_size]; const int txfm_size_row = tx_size_high[tx_size]; @@ -2857,11 +2857,11 @@ (void)eob; __m128i buf[16]; const TX_SIZE tx_size = TX_16X4; - const int8_t *shift = inv_txfm_shift_ls[tx_size]; + const int8_t *shift = av1_inv_txfm_shift_ls[tx_size]; const int txw_idx = get_txw_idx(tx_size); const int txh_idx = get_txh_idx(tx_size); - const int cos_bit_row = inv_cos_bit_row[txw_idx][txh_idx]; - const int cos_bit_col = inv_cos_bit_col[txw_idx][txh_idx]; + const int cos_bit_row = av1_inv_cos_bit_row[txw_idx][txh_idx]; + const int cos_bit_col = av1_inv_cos_bit_col[txw_idx][txh_idx]; const int txfm_size_col = tx_size_wide[tx_size]; const int txfm_size_row = tx_size_high[tx_size]; const int buf_size_w_div8 = txfm_size_col >> 3;
diff --git a/av1/common/x86/highbd_inv_txfm_avx2.c b/av1/common/x86/highbd_inv_txfm_avx2.c index fe22465..10fc721 100644 --- a/av1/common/x86/highbd_inv_txfm_avx2.c +++ b/av1/common/x86/highbd_inv_txfm_avx2.c
@@ -4219,7 +4219,7 @@ __m256i buf1[64 * 8]; int eobx, eoby; get_eobx_eoby_scan_default(&eobx, &eoby, tx_size, eob); - const int8_t *shift = inv_txfm_shift_ls[tx_size]; + const int8_t *shift = av1_inv_txfm_shift_ls[tx_size]; const int txw_idx = get_txw_idx(tx_size); const int txh_idx = get_txh_idx(tx_size); const int txfm_size_col = tx_size_wide[tx_size]; @@ -4255,7 +4255,8 @@ av1_round_shift_rect_array_32_avx2( buf0, buf0, buf_size_nonzero_w_div8 << 3, 0, NewInvSqrt2); } - row_txfm(buf0, buf0, inv_cos_bit_row[txw_idx][txh_idx], 0, bd, -shift[0]); + row_txfm(buf0, buf0, av1_inv_cos_bit_row[txw_idx][txh_idx], 0, bd, + -shift[0]); __m256i *_buf1 = buf1 + i * 8; if (lr_flip) { @@ -4272,7 +4273,7 @@ // 2nd stage: column transform for (int i = 0; i < buf_size_w_div8; i++) { col_txfm(buf1 + i * txfm_size_row, buf1 + i * txfm_size_row, - inv_cos_bit_col[txw_idx][txh_idx], 1, bd, 0); + av1_inv_cos_bit_col[txw_idx][txh_idx], 1, bd, 0); av1_round_shift_array_32_avx2(buf1 + i * txfm_size_row, buf1 + i * txfm_size_row, txfm_size_row,
diff --git a/av1/common/x86/highbd_inv_txfm_sse4.c b/av1/common/x86/highbd_inv_txfm_sse4.c index 8a8641d..2119e31 100644 --- a/av1/common/x86/highbd_inv_txfm_sse4.c +++ b/av1/common/x86/highbd_inv_txfm_sse4.c
@@ -646,105 +646,113 @@ void av1_inv_txfm2d_add_4x4_sse4_1(const int32_t *coeff, uint16_t *output, int stride, TX_TYPE tx_type, int bd) { __m128i in[4]; - const int8_t *shift = inv_txfm_shift_ls[TX_4X4]; + const int8_t *shift = av1_inv_txfm_shift_ls[TX_4X4]; const int txw_idx = get_txw_idx(TX_4X4); const int txh_idx = get_txh_idx(TX_4X4); switch (tx_type) { case DCT_DCT: load_buffer_4x4(coeff, in); - idct4x4_sse4_1(in, in, inv_cos_bit_row[txw_idx][txh_idx], 0, bd, 0); - idct4x4_sse4_1(in, in, inv_cos_bit_col[txw_idx][txh_idx], 1, bd, 0); + idct4x4_sse4_1(in, in, av1_inv_cos_bit_row[txw_idx][txh_idx], 0, bd, 0); + idct4x4_sse4_1(in, in, av1_inv_cos_bit_col[txw_idx][txh_idx], 1, bd, 0); write_buffer_4x4(in, output, stride, 0, 0, -shift[1], bd); break; case ADST_DCT: load_buffer_4x4(coeff, in); - idct4x4_sse4_1(in, in, inv_cos_bit_row[txw_idx][txh_idx], 0, bd, 0); - iadst4x4_sse4_1(in, in, inv_cos_bit_col[txw_idx][txh_idx], 1, bd, 0); + idct4x4_sse4_1(in, in, av1_inv_cos_bit_row[txw_idx][txh_idx], 0, bd, 0); + iadst4x4_sse4_1(in, in, av1_inv_cos_bit_col[txw_idx][txh_idx], 1, bd, 0); write_buffer_4x4(in, output, stride, 0, 0, -shift[1], bd); break; case DCT_ADST: load_buffer_4x4(coeff, in); - iadst4x4_sse4_1(in, in, inv_cos_bit_row[txw_idx][txh_idx], 0, bd, 0); - idct4x4_sse4_1(in, in, inv_cos_bit_col[txw_idx][txh_idx], 1, bd, 0); + iadst4x4_sse4_1(in, in, av1_inv_cos_bit_row[txw_idx][txh_idx], 0, bd, 0); + idct4x4_sse4_1(in, in, av1_inv_cos_bit_col[txw_idx][txh_idx], 1, bd, 0); write_buffer_4x4(in, output, stride, 0, 0, -shift[1], bd); break; case ADST_ADST: load_buffer_4x4(coeff, in); - iadst4x4_sse4_1(in, in, inv_cos_bit_row[txw_idx][txh_idx], 0, bd, 0); - iadst4x4_sse4_1(in, in, inv_cos_bit_col[txw_idx][txh_idx], 1, bd, 0); + iadst4x4_sse4_1(in, in, av1_inv_cos_bit_row[txw_idx][txh_idx], 0, bd, 0); + iadst4x4_sse4_1(in, in, av1_inv_cos_bit_col[txw_idx][txh_idx], 1, bd, 0); write_buffer_4x4(in, output, stride, 0, 0, -shift[1], bd); break; case FLIPADST_DCT: load_buffer_4x4(coeff, in); - idct4x4_sse4_1(in, in, inv_cos_bit_row[txw_idx][txh_idx], 0, bd, 0); - iadst4x4_sse4_1(in, in, inv_cos_bit_col[txw_idx][txh_idx], 1, bd, 0); + idct4x4_sse4_1(in, in, av1_inv_cos_bit_row[txw_idx][txh_idx], 0, bd, 0); + iadst4x4_sse4_1(in, in, av1_inv_cos_bit_col[txw_idx][txh_idx], 1, bd, 0); write_buffer_4x4(in, output, stride, 0, 1, -shift[1], bd); break; case DCT_FLIPADST: load_buffer_4x4(coeff, in); - iadst4x4_sse4_1(in, in, inv_cos_bit_row[txw_idx][txh_idx], 0, bd, 0); - idct4x4_sse4_1(in, in, inv_cos_bit_col[txw_idx][txh_idx], 1, bd, 0); + iadst4x4_sse4_1(in, in, av1_inv_cos_bit_row[txw_idx][txh_idx], 0, bd, 0); + idct4x4_sse4_1(in, in, av1_inv_cos_bit_col[txw_idx][txh_idx], 1, bd, 0); write_buffer_4x4(in, output, stride, 1, 0, -shift[1], bd); break; case FLIPADST_FLIPADST: load_buffer_4x4(coeff, in); - iadst4x4_sse4_1(in, in, inv_cos_bit_row[txw_idx][txh_idx], 0, bd, 0); - iadst4x4_sse4_1(in, in, inv_cos_bit_col[txw_idx][txh_idx], 1, bd, 0); + iadst4x4_sse4_1(in, in, av1_inv_cos_bit_row[txw_idx][txh_idx], 0, bd, 0); + iadst4x4_sse4_1(in, in, av1_inv_cos_bit_col[txw_idx][txh_idx], 1, bd, 0); write_buffer_4x4(in, output, stride, 1, 1, -shift[1], bd); break; case ADST_FLIPADST: load_buffer_4x4(coeff, in); - iadst4x4_sse4_1(in, in, inv_cos_bit_row[txw_idx][txh_idx], 0, bd, 0); - iadst4x4_sse4_1(in, in, inv_cos_bit_col[txw_idx][txh_idx], 1, bd, 0); + iadst4x4_sse4_1(in, in, av1_inv_cos_bit_row[txw_idx][txh_idx], 0, bd, 0); + iadst4x4_sse4_1(in, in, av1_inv_cos_bit_col[txw_idx][txh_idx], 1, bd, 0); write_buffer_4x4(in, output, stride, 1, 0, -shift[1], bd); break; case FLIPADST_ADST: load_buffer_4x4(coeff, in); - iadst4x4_sse4_1(in, in, inv_cos_bit_row[txw_idx][txh_idx], 0, bd, 0); - iadst4x4_sse4_1(in, in, inv_cos_bit_col[txw_idx][txh_idx], 1, bd, 0); + iadst4x4_sse4_1(in, in, av1_inv_cos_bit_row[txw_idx][txh_idx], 0, bd, 0); + iadst4x4_sse4_1(in, in, av1_inv_cos_bit_col[txw_idx][txh_idx], 1, bd, 0); write_buffer_4x4(in, output, stride, 0, 1, -shift[1], bd); break; case IDTX: load_buffer_4x4(coeff, in); - iidentity4_sse4_1(in, in, inv_cos_bit_row[txw_idx][txh_idx], 0, bd, 0); - iidentity4_sse4_1(in, in, inv_cos_bit_col[txw_idx][txh_idx], 1, bd, 0); + iidentity4_sse4_1(in, in, av1_inv_cos_bit_row[txw_idx][txh_idx], 0, bd, + 0); + iidentity4_sse4_1(in, in, av1_inv_cos_bit_col[txw_idx][txh_idx], 1, bd, + 0); write_buffer_4x4(in, output, stride, 0, 0, -shift[1], bd); break; case V_DCT: load_buffer_4x4(coeff, in); - iidentity4_sse4_1(in, in, inv_cos_bit_row[txw_idx][txh_idx], 0, bd, 0); - idct4x4_sse4_1(in, in, inv_cos_bit_col[txw_idx][txh_idx], 1, bd, 0); + iidentity4_sse4_1(in, in, av1_inv_cos_bit_row[txw_idx][txh_idx], 0, bd, + 0); + idct4x4_sse4_1(in, in, av1_inv_cos_bit_col[txw_idx][txh_idx], 1, bd, 0); write_buffer_4x4(in, output, stride, 0, 0, -shift[1], bd); break; case H_DCT: load_buffer_4x4(coeff, in); - idct4x4_sse4_1(in, in, inv_cos_bit_row[txw_idx][txh_idx], 0, bd, 0); - iidentity4_sse4_1(in, in, inv_cos_bit_col[txw_idx][txh_idx], 1, bd, 0); + idct4x4_sse4_1(in, in, av1_inv_cos_bit_row[txw_idx][txh_idx], 0, bd, 0); + iidentity4_sse4_1(in, in, av1_inv_cos_bit_col[txw_idx][txh_idx], 1, bd, + 0); write_buffer_4x4(in, output, stride, 0, 0, -shift[1], bd); break; case V_ADST: load_buffer_4x4(coeff, in); - iidentity4_sse4_1(in, in, inv_cos_bit_row[txw_idx][txh_idx], 0, bd, 0); - iadst4x4_sse4_1(in, in, inv_cos_bit_col[txw_idx][txh_idx], 1, bd, 0); + iidentity4_sse4_1(in, in, av1_inv_cos_bit_row[txw_idx][txh_idx], 0, bd, + 0); + iadst4x4_sse4_1(in, in, av1_inv_cos_bit_col[txw_idx][txh_idx], 1, bd, 0); write_buffer_4x4(in, output, stride, 0, 0, -shift[1], bd); break; case H_ADST: load_buffer_4x4(coeff, in); - iadst4x4_sse4_1(in, in, inv_cos_bit_row[txw_idx][txh_idx], 0, bd, 0); - iidentity4_sse4_1(in, in, inv_cos_bit_col[txw_idx][txh_idx], 1, bd, 0); + iadst4x4_sse4_1(in, in, av1_inv_cos_bit_row[txw_idx][txh_idx], 0, bd, 0); + iidentity4_sse4_1(in, in, av1_inv_cos_bit_col[txw_idx][txh_idx], 1, bd, + 0); write_buffer_4x4(in, output, stride, 0, 0, -shift[1], bd); break; case V_FLIPADST: load_buffer_4x4(coeff, in); - iidentity4_sse4_1(in, in, inv_cos_bit_row[txw_idx][txh_idx], 0, bd, 0); - iadst4x4_sse4_1(in, in, inv_cos_bit_col[txw_idx][txh_idx], 1, bd, 0); + iidentity4_sse4_1(in, in, av1_inv_cos_bit_row[txw_idx][txh_idx], 0, bd, + 0); + iadst4x4_sse4_1(in, in, av1_inv_cos_bit_col[txw_idx][txh_idx], 1, bd, 0); write_buffer_4x4(in, output, stride, 0, 1, -shift[1], bd); break; case H_FLIPADST: load_buffer_4x4(coeff, in); - iadst4x4_sse4_1(in, in, inv_cos_bit_row[txw_idx][txh_idx], 0, bd, 0); - iidentity4_sse4_1(in, in, inv_cos_bit_col[txw_idx][txh_idx], 1, bd, 0); + iadst4x4_sse4_1(in, in, av1_inv_cos_bit_row[txw_idx][txh_idx], 0, bd, 0); + iidentity4_sse4_1(in, in, av1_inv_cos_bit_col[txw_idx][txh_idx], 1, bd, + 0); write_buffer_4x4(in, output, stride, 1, 0, -shift[1], bd); break; default: assert(0); @@ -1352,7 +1360,7 @@ void av1_inv_txfm2d_add_8x8_sse4_1(const int32_t *coeff, uint16_t *output, int stride, TX_TYPE tx_type, int bd) { __m128i in[16], out[16]; - const int8_t *shift = inv_txfm_shift_ls[TX_8X8]; + const int8_t *shift = av1_inv_txfm_shift_ls[TX_8X8]; const int txw_idx = get_txw_idx(TX_8X8); const int txh_idx = get_txh_idx(TX_8X8); @@ -1360,82 +1368,82 @@ case DCT_DCT: load_buffer_8x8(coeff, in); transpose_8x8(in, out); - idct8x8_sse4_1(out, in, inv_cos_bit_row[txw_idx][txh_idx], 0, bd, + idct8x8_sse4_1(out, in, av1_inv_cos_bit_row[txw_idx][txh_idx], 0, bd, -shift[0]); transpose_8x8(in, out); - idct8x8_sse4_1(out, in, inv_cos_bit_col[txw_idx][txh_idx], 1, bd, 0); + idct8x8_sse4_1(out, in, av1_inv_cos_bit_col[txw_idx][txh_idx], 1, bd, 0); write_buffer_8x8(in, output, stride, 0, 0, -shift[1], bd); break; case DCT_ADST: load_buffer_8x8(coeff, in); transpose_8x8(in, out); - iadst8x8_sse4_1(out, in, inv_cos_bit_row[txw_idx][txh_idx], 0, bd, + iadst8x8_sse4_1(out, in, av1_inv_cos_bit_row[txw_idx][txh_idx], 0, bd, -shift[0]); transpose_8x8(in, out); - idct8x8_sse4_1(out, in, inv_cos_bit_col[txw_idx][txh_idx], 1, bd, 0); + idct8x8_sse4_1(out, in, av1_inv_cos_bit_col[txw_idx][txh_idx], 1, bd, 0); write_buffer_8x8(in, output, stride, 0, 0, -shift[1], bd); break; case ADST_DCT: load_buffer_8x8(coeff, in); transpose_8x8(in, out); - idct8x8_sse4_1(out, in, inv_cos_bit_row[txw_idx][txh_idx], 0, bd, + idct8x8_sse4_1(out, in, av1_inv_cos_bit_row[txw_idx][txh_idx], 0, bd, -shift[0]); transpose_8x8(in, out); - iadst8x8_sse4_1(out, in, inv_cos_bit_col[txw_idx][txh_idx], 1, bd, 0); + iadst8x8_sse4_1(out, in, av1_inv_cos_bit_col[txw_idx][txh_idx], 1, bd, 0); write_buffer_8x8(in, output, stride, 0, 0, -shift[1], bd); break; case ADST_ADST: load_buffer_8x8(coeff, in); transpose_8x8(in, out); - iadst8x8_sse4_1(out, in, inv_cos_bit_row[txw_idx][txh_idx], 0, bd, + iadst8x8_sse4_1(out, in, av1_inv_cos_bit_row[txw_idx][txh_idx], 0, bd, -shift[0]); transpose_8x8(in, out); - iadst8x8_sse4_1(out, in, inv_cos_bit_col[txw_idx][txh_idx], 1, bd, 0); + iadst8x8_sse4_1(out, in, av1_inv_cos_bit_col[txw_idx][txh_idx], 1, bd, 0); write_buffer_8x8(in, output, stride, 0, 0, -shift[1], bd); break; case FLIPADST_DCT: load_buffer_8x8(coeff, in); transpose_8x8(in, out); - idct8x8_sse4_1(out, in, inv_cos_bit_row[txw_idx][txh_idx], 0, bd, + idct8x8_sse4_1(out, in, av1_inv_cos_bit_row[txw_idx][txh_idx], 0, bd, -shift[0]); transpose_8x8(in, out); - iadst8x8_sse4_1(out, in, inv_cos_bit_col[txw_idx][txh_idx], 1, bd, 0); + iadst8x8_sse4_1(out, in, av1_inv_cos_bit_col[txw_idx][txh_idx], 1, bd, 0); write_buffer_8x8(in, output, stride, 0, 1, -shift[1], bd); break; case DCT_FLIPADST: load_buffer_8x8(coeff, in); transpose_8x8(in, out); - iadst8x8_sse4_1(out, in, inv_cos_bit_row[txw_idx][txh_idx], 0, bd, + iadst8x8_sse4_1(out, in, av1_inv_cos_bit_row[txw_idx][txh_idx], 0, bd, -shift[0]); transpose_8x8(in, out); - idct8x8_sse4_1(out, in, inv_cos_bit_col[txw_idx][txh_idx], 1, bd, 0); + idct8x8_sse4_1(out, in, av1_inv_cos_bit_col[txw_idx][txh_idx], 1, bd, 0); write_buffer_8x8(in, output, stride, 1, 0, -shift[1], bd); break; case ADST_FLIPADST: load_buffer_8x8(coeff, in); transpose_8x8(in, out); - iadst8x8_sse4_1(out, in, inv_cos_bit_row[txw_idx][txh_idx], 0, bd, + iadst8x8_sse4_1(out, in, av1_inv_cos_bit_row[txw_idx][txh_idx], 0, bd, -shift[0]); transpose_8x8(in, out); - iadst8x8_sse4_1(out, in, inv_cos_bit_col[txw_idx][txh_idx], 1, bd, 0); + iadst8x8_sse4_1(out, in, av1_inv_cos_bit_col[txw_idx][txh_idx], 1, bd, 0); write_buffer_8x8(in, output, stride, 1, 0, -shift[1], bd); break; case FLIPADST_FLIPADST: load_buffer_8x8(coeff, in); transpose_8x8(in, out); - iadst8x8_sse4_1(out, in, inv_cos_bit_row[txw_idx][txh_idx], 0, bd, + iadst8x8_sse4_1(out, in, av1_inv_cos_bit_row[txw_idx][txh_idx], 0, bd, -shift[0]); transpose_8x8(in, out); - iadst8x8_sse4_1(out, in, inv_cos_bit_col[txw_idx][txh_idx], 1, bd, 0); + iadst8x8_sse4_1(out, in, av1_inv_cos_bit_col[txw_idx][txh_idx], 1, bd, 0); write_buffer_8x8(in, output, stride, 1, 1, -shift[1], bd); break; case FLIPADST_ADST: load_buffer_8x8(coeff, in); transpose_8x8(in, out); - iadst8x8_sse4_1(out, in, inv_cos_bit_row[txw_idx][txh_idx], 0, bd, + iadst8x8_sse4_1(out, in, av1_inv_cos_bit_row[txw_idx][txh_idx], 0, bd, -shift[0]); transpose_8x8(in, out); - iadst8x8_sse4_1(out, in, inv_cos_bit_col[txw_idx][txh_idx], 1, bd, 0); + iadst8x8_sse4_1(out, in, av1_inv_cos_bit_col[txw_idx][txh_idx], 1, bd, 0); write_buffer_8x8(in, output, stride, 0, 1, -shift[1], bd); break; default: assert(0); @@ -5333,7 +5341,7 @@ __m128i buf1[64]; int eobx, eoby; get_eobx_eoby_scan_v_identity(&eobx, &eoby, tx_size, eob); - const int8_t *shift = inv_txfm_shift_ls[tx_size]; + const int8_t *shift = av1_inv_txfm_shift_ls[tx_size]; const int txw_idx = get_txw_idx(tx_size); const int txh_idx = get_txh_idx(tx_size); const int txfm_size_col = tx_size_wide[tx_size]; @@ -5361,7 +5369,8 @@ av1_round_shift_rect_array_32_sse4_1(buf0, buf0, input_stride, 0, NewInvSqrt2); } - row_txfm(buf0, buf0, inv_cos_bit_row[txw_idx][txh_idx], 0, bd, -shift[0]); + row_txfm(buf0, buf0, av1_inv_cos_bit_row[txw_idx][txh_idx], 0, bd, + -shift[0]); __m128i *_buf1 = buf1 + i * 4; @@ -5374,7 +5383,7 @@ } for (int i = 0; i < buf_size_w_div4; i++) { col_txfm(buf1 + i * txfm_size_row, buf1 + i * txfm_size_row, - inv_cos_bit_col[txw_idx][txh_idx], 1, bd, 0); + av1_inv_cos_bit_col[txw_idx][txh_idx], 1, bd, 0); av1_round_shift_array_32_sse4_1(buf1 + i * txfm_size_row, buf1 + i * txfm_size_row, txfm_size_row, @@ -5395,7 +5404,7 @@ __m128i buf1[64]; int eobx, eoby; get_eobx_eoby_scan_h_identity(&eobx, &eoby, tx_size, eob); - const int8_t *shift = inv_txfm_shift_ls[tx_size]; + const int8_t *shift = av1_inv_txfm_shift_ls[tx_size]; const int txw_idx = get_txw_idx(tx_size); const int txh_idx = get_txh_idx(tx_size); const int txfm_size_col = tx_size_wide[tx_size]; @@ -5427,7 +5436,8 @@ av1_round_shift_rect_array_32_sse4_1( buf0, buf0, (buf_size_nonzero_w_div8 << 3), 0, NewInvSqrt2); } - row_txfm(buf0, buf0, inv_cos_bit_row[txw_idx][txh_idx], 0, bd, -shift[0]); + row_txfm(buf0, buf0, av1_inv_cos_bit_row[txw_idx][txh_idx], 0, bd, + -shift[0]); __m128i *_buf1 = buf1 + i * 4; if (lr_flip) { @@ -5450,7 +5460,7 @@ } for (int i = 0; i < buf_size_w_div8; i++) { col_txfm(buf1 + i * txfm_size_row, buf1 + i * txfm_size_row, - inv_cos_bit_col[txw_idx][txh_idx], 1, bd, 0); + av1_inv_cos_bit_col[txw_idx][txh_idx], 1, bd, 0); av1_round_shift_array_32_sse4_1(buf1 + i * txfm_size_row, buf1 + i * txfm_size_row, txfm_size_row, @@ -5472,7 +5482,7 @@ int eob, const int bd) { (void)eob; __m128i buf1[64 * 4]; - const int8_t *shift = inv_txfm_shift_ls[tx_size]; + const int8_t *shift = av1_inv_txfm_shift_ls[tx_size]; const int txw_idx = get_txw_idx(tx_size); const int txh_idx = get_txh_idx(tx_size); const int txfm_size_col = tx_size_wide[tx_size]; @@ -5496,7 +5506,8 @@ av1_round_shift_rect_array_32_sse4_1(buf0, buf0, input_stride, 0, NewInvSqrt2); } - row_txfm(buf0, buf0, inv_cos_bit_row[txw_idx][txh_idx], 0, bd, -shift[0]); + row_txfm(buf0, buf0, av1_inv_cos_bit_row[txw_idx][txh_idx], 0, bd, + -shift[0]); __m128i *_buf1 = buf1 + i * 4; for (int j = 0; j < (input_stride >> 2); ++j) { @@ -5508,7 +5519,7 @@ } for (int i = 0; i < (input_stride >> 2); i++) { col_txfm(buf1 + i * txfm_size_row, buf1 + i * txfm_size_row, - inv_cos_bit_col[txw_idx][txh_idx], 1, bd, 0); + av1_inv_cos_bit_col[txw_idx][txh_idx], 1, bd, 0); av1_round_shift_array_32_sse4_1(buf1 + i * txfm_size_row, buf1 + i * txfm_size_row, txfm_size_row, @@ -5532,7 +5543,7 @@ __m128i buf1[64 * 16]; int eobx, eoby; get_eobx_eoby_scan_default(&eobx, &eoby, tx_size, eob); - const int8_t *shift = inv_txfm_shift_ls[tx_size]; + const int8_t *shift = av1_inv_txfm_shift_ls[tx_size]; const int txw_idx = get_txw_idx(tx_size); const int txh_idx = get_txh_idx(tx_size); const int txfm_size_col = tx_size_wide[tx_size]; @@ -5570,7 +5581,8 @@ av1_round_shift_rect_array_32_sse4_1( buf0, buf0, buf_size_nonzero_w_div8 << 3, 0, NewInvSqrt2); } - row_txfm(buf0, buf0, inv_cos_bit_row[txw_idx][txh_idx], 0, bd, -shift[0]); + row_txfm(buf0, buf0, av1_inv_cos_bit_row[txw_idx][txh_idx], 0, bd, + -shift[0]); __m128i *_buf1 = buf1 + i * 4; if (lr_flip) { @@ -5594,7 +5606,7 @@ // 2nd stage: column transform for (int i = 0; i < buf_size_w_div8; i++) { col_txfm(buf1 + i * txfm_size_row, buf1 + i * txfm_size_row, - inv_cos_bit_col[txw_idx][txh_idx], 1, bd, 0); + av1_inv_cos_bit_col[txw_idx][txh_idx], 1, bd, 0); av1_round_shift_array_32_sse4_1(buf1 + i * txfm_size_row, buf1 + i * txfm_size_row, txfm_size_row, @@ -5617,7 +5629,7 @@ int eob, const int bd) { (void)eob; __m128i buf1[8]; - const int8_t *shift = inv_txfm_shift_ls[tx_size]; + const int8_t *shift = av1_inv_txfm_shift_ls[tx_size]; const int txw_idx = get_txw_idx(tx_size); const int txh_idx = get_txh_idx(tx_size); const int txfm_size_col = tx_size_wide[tx_size]; @@ -5640,8 +5652,8 @@ load_buffer_32bit_input(input_row, input_stride, buf0_cur, txfm_size_row); av1_round_shift_rect_array_32_sse4_1(buf0, buf0, txfm_size_row, 0, NewInvSqrt2); - row_txfm(buf0, buf0, inv_cos_bit_row[txw_idx][txh_idx], 0, bd, -shift[0]); - row_txfm(buf0 + 4, buf0 + 4, inv_cos_bit_row[txw_idx][txh_idx], 0, bd, + row_txfm(buf0, buf0, av1_inv_cos_bit_row[txw_idx][txh_idx], 0, bd, -shift[0]); + row_txfm(buf0 + 4, buf0 + 4, av1_inv_cos_bit_row[txw_idx][txh_idx], 0, bd, -shift[0]); if (lr_flip) { @@ -5659,7 +5671,7 @@ } // 2nd stage: column transform - col_txfm(buf1, buf1, inv_cos_bit_col[txw_idx][txh_idx], 1, bd, 0); + col_txfm(buf1, buf1, av1_inv_cos_bit_col[txw_idx][txh_idx], 1, bd, 0); av1_round_shift_array_32_sse4_1(buf1, buf1, txfm_size_row, -shift[1]); @@ -5674,7 +5686,7 @@ int eob, const int bd) { (void)eob; __m128i buf1[8]; - const int8_t *shift = inv_txfm_shift_ls[tx_size]; + const int8_t *shift = av1_inv_txfm_shift_ls[tx_size]; const int txw_idx = get_txw_idx(tx_size); const int txh_idx = get_txh_idx(tx_size); const int txfm_size_col = tx_size_wide[tx_size]; @@ -5701,7 +5713,7 @@ av1_round_shift_rect_array_32_sse4_1(buf1, buf0, txfm_size_col, 0, NewInvSqrt2); - row_txfm(buf0, buf0, inv_cos_bit_row[txw_idx][txh_idx], 0, bd, -shift[0]); + row_txfm(buf0, buf0, av1_inv_cos_bit_row[txw_idx][txh_idx], 0, bd, -shift[0]); __m128i *buf1_ptr; if (lr_flip) { @@ -5714,7 +5726,7 @@ // 2nd stage: column transform for (int i = 0; i < 2; i++) { col_txfm(buf1_ptr + i * txfm_size_row, buf1_ptr + i * txfm_size_row, - inv_cos_bit_col[txw_idx][txh_idx], 1, bd, 0); + av1_inv_cos_bit_col[txw_idx][txh_idx], 1, bd, 0); } av1_round_shift_array_32_sse4_1(buf1_ptr, buf1_ptr, txfm_size_col, -shift[1]); // write to buffer @@ -5728,7 +5740,7 @@ int eob, const int bd) { (void)eob; __m128i buf1[16]; - const int8_t *shift = inv_txfm_shift_ls[tx_size]; + const int8_t *shift = av1_inv_txfm_shift_ls[tx_size]; const int txw_idx = get_txw_idx(tx_size); const int txh_idx = get_txh_idx(tx_size); const int txfm_size_col = tx_size_wide[tx_size]; @@ -5752,7 +5764,7 @@ load_buffer_32bit_input(input_row, input_stride, buf0_cur, txfm_size_row); for (int i = 0; i < (txfm_size_row >> 2); i++) { row_txfm(buf0 + (i << 2), buf0 + (i << 2), - inv_cos_bit_row[txw_idx][txh_idx], 0, bd, -shift[0]); + av1_inv_cos_bit_row[txw_idx][txh_idx], 0, bd, -shift[0]); } av1_round_shift_array_32_sse4_1(buf0, buf0, txfm_size_row, -shift[0]); @@ -5772,7 +5784,7 @@ } // 2nd stage: column transform - col_txfm(buf1, buf1, inv_cos_bit_col[txw_idx][txh_idx], 1, bd, 0); + col_txfm(buf1, buf1, av1_inv_cos_bit_col[txw_idx][txh_idx], 1, bd, 0); av1_round_shift_array_32_sse4_1(buf1, buf1, txfm_size_row, -shift[1]); @@ -5787,7 +5799,7 @@ int eob, const int bd) { (void)eob; __m128i buf1[16]; - const int8_t *shift = inv_txfm_shift_ls[tx_size]; + const int8_t *shift = av1_inv_txfm_shift_ls[tx_size]; const int txw_idx = get_txw_idx(tx_size); const int txh_idx = get_txh_idx(tx_size); const int txfm_size_col = tx_size_wide[tx_size]; @@ -5812,7 +5824,7 @@ TRANSPOSE_4X4(buf0[j], buf0[j + 4], buf0[j + 8], buf0[j + 12], buf1[4 * j], buf1[4 * j + 1], buf1[4 * j + 2], buf1[4 * j + 3]); } - row_txfm(buf1, buf0, inv_cos_bit_row[txw_idx][txh_idx], 0, bd, -shift[0]); + row_txfm(buf1, buf0, av1_inv_cos_bit_row[txw_idx][txh_idx], 0, bd, -shift[0]); __m128i *buf1_ptr; if (lr_flip) { @@ -5825,7 +5837,7 @@ // 2nd stage: column transform for (int i = 0; i < buf_size_w_div8; i++) { col_txfm(buf1_ptr + i * txfm_size_row, buf1_ptr + i * txfm_size_row, - inv_cos_bit_col[txw_idx][txh_idx], 1, bd, 0); + av1_inv_cos_bit_col[txw_idx][txh_idx], 1, bd, 0); } av1_round_shift_array_32_sse4_1(buf1_ptr, buf1_ptr, txfm_size_col, -shift[1]);