Skip to content
Snippets Groups Projects
  1. Feb 14, 2024
    • Jean-Baptiste Kempf's avatar
      Update NEWS · bb645893
      Jean-Baptiste Kempf authored
      Fix typo
      1.4.0
      bb645893
    • Nathan E. Egge's avatar
      riscv64/itx: Add 16-point 8bpc RVV flipadst transform · 7b15ca13
      Nathan E. Egge authored
      inv_txfm_add_16x16_adst_flipadst_0_8bpc_c:        15272.2 ( 1.00x)
      inv_txfm_add_16x16_adst_flipadst_0_8bpc_rvv:       1824.4 ( 8.37x)
      inv_txfm_add_16x16_adst_flipadst_1_8bpc_c:        15261.2 ( 1.00x)
      inv_txfm_add_16x16_adst_flipadst_1_8bpc_rvv:       1824.5 ( 8.36x)
      inv_txfm_add_16x16_adst_flipadst_2_8bpc_c:        15260.0 ( 1.00x)
      inv_txfm_add_16x16_adst_flipadst_2_8bpc_rvv:       1824.5 ( 8.36x)
      inv_txfm_add_16x16_dct_flipadst_0_8bpc_c:         14497.2 ( 1.00x)
      inv_txfm_add_16x16_dct_flipadst_0_8bpc_rvv:        1637.3 ( 8.85x)
      inv_txfm_add_16x16_dct_flipadst_1_8bpc_c:         14490.5 ( 1.00x)
      inv_txfm_add_16x16_dct_flipadst_1_8bpc_rvv:        1637.3 ( 8.85x)
      inv_txfm_add_16x16_dct_flipadst_2_8bpc_c:         14486.4 ( 1.00x)
      inv_txfm_add_16x16_dct_flipadst_2_8bpc_rvv:        1637.3 ( 8.85x)
      inv_txfm_add_16x16_flipadst_adst_0_8bpc_c:        15307.7 ( 1.00x)
      inv_txfm_add_16x16_flipadst_adst_0_8bpc_rvv:       1808.0 ( 8.47x)
      inv_txfm_add_16x16_flipadst_adst_1_8bpc_c:        15341.0 ( 1.00x)
      inv_txfm_add_16x16_flipadst_adst_1_8bpc_rvv:       1808.1 ( 8.48x)
      inv_txfm_add_16x16_flipadst_adst_2_8bpc_c:        15333.5 ( 1.00x)
      inv_txfm_add_16x16_flipadst_adst_2_8bpc_rvv:       1808.1 ( 8.48x)
      inv_txfm_add_16x16_flipadst_dct_0_8bpc_c:         14530.0 ( 1.00x)
      inv_txfm_add_16x16_flipadst_dct_0_8bpc_rvv:        1636.4 ( 8.88x)
      inv_txfm_add_16x16_flipadst_dct_1_8bpc_c:         14510.3 ( 1.00x)
      inv_txfm_add_16x16_flipadst_dct_1_8bpc_rvv:        1636.3 ( 8.87x)
      inv_txfm_add_16x16_flipadst_dct_2_8bpc_c:         14504.7 ( 1.00x)
      inv_txfm_add_16x16_flipadst_dct_2_8bpc_rvv:        1636.3 ( 8.86x)
      inv_txfm_add_16x16_flipadst_flipadst_0_8bpc_c:    15315.5 ( 1.00x)
      inv_txfm_add_16x16_flipadst_flipadst_0_8bpc_rvv:   1823.5 ( 8.40x)
      inv_txfm_add_16x16_flipadst_flipadst_1_8bpc_c:    15324.1 ( 1.00x)
      inv_txfm_add_16x16_flipadst_flipadst_1_8bpc_rvv:   1823.3 ( 8.40x)
      inv_txfm_add_16x16_flipadst_flipadst_2_8bpc_c:    15315.6 ( 1.00x)
      inv_txfm_add_16x16_flipadst_flipadst_2_8bpc_rvv:   1823.5 ( 8.40x)
      7b15ca13
    • Nathan E. Egge's avatar
    • Nathan E. Egge's avatar
      riscv64/itx: Add 16-point 8bpc RVV adst transform · 2685b409
      Nathan E. Egge authored
      inv_txfm_add_16x16_adst_adst_0_8bpc_c:      15364.4 ( 1.00x)
      inv_txfm_add_16x16_adst_adst_0_8bpc_rvv:     1814.1 ( 8.47x)
      inv_txfm_add_16x16_adst_adst_1_8bpc_c:      15363.7 ( 1.00x)
      inv_txfm_add_16x16_adst_adst_1_8bpc_rvv:     1814.5 ( 8.47x)
      inv_txfm_add_16x16_adst_adst_2_8bpc_c:      15368.9 ( 1.00x)
      inv_txfm_add_16x16_adst_adst_2_8bpc_rvv:     1814.5 ( 8.47x)
      inv_txfm_add_16x16_adst_dct_0_8bpc_c:       14560.0 ( 1.00x)
      inv_txfm_add_16x16_adst_dct_0_8bpc_rvv:      1644.4 ( 8.85x)
      inv_txfm_add_16x16_adst_dct_1_8bpc_c:       14578.9 ( 1.00x)
      inv_txfm_add_16x16_adst_dct_1_8bpc_rvv:      1644.5 ( 8.87x)
      inv_txfm_add_16x16_adst_dct_2_8bpc_c:       14575.0 ( 1.00x)
      inv_txfm_add_16x16_adst_dct_2_8bpc_rvv:      1644.6 ( 8.86x)
      inv_txfm_add_16x16_dct_adst_0_8bpc_c:       14550.7 ( 1.00x)
      inv_txfm_add_16x16_dct_adst_0_8bpc_rvv:      1622.7 ( 8.97x)
      inv_txfm_add_16x16_dct_adst_1_8bpc_c:       14556.0 ( 1.00x)
      inv_txfm_add_16x16_dct_adst_1_8bpc_rvv:      1622.6 ( 8.97x)
      inv_txfm_add_16x16_dct_adst_2_8bpc_c:       14543.3 ( 1.00x)
      inv_txfm_add_16x16_dct_adst_2_8bpc_rvv:      1622.6 ( 8.96x)
      2685b409
    • Nathan E. Egge's avatar
      riscv64/itx: Add 4x4 8bpc RVV wht_wht transform · 72dba22e
      Nathan E. Egge authored
      inv_txfm_add_4x4_wht_wht_0_8bpc_c:      265.6 ( 1.00x)
      inv_txfm_add_4x4_wht_wht_0_8bpc_rvv:     66.9 ( 3.97x)
      inv_txfm_add_4x4_wht_wht_1_8bpc_c:      265.5 ( 1.00x)
      inv_txfm_add_4x4_wht_wht_1_8bpc_rvv:     66.9 ( 3.97x)
      72dba22e
    • Nathan E. Egge's avatar
      riscv64/itx: Add 16x16 8bpc dct_identity and identity_dct · cc29b231
      Nathan E. Egge authored
      inv_txfm_add_16x16_dct_identity_0_8bpc_c:    10593.3 ( 1.00x)
      inv_txfm_add_16x16_dct_identity_0_8bpc_rvv:   1163.3 ( 9.11x)
      inv_txfm_add_16x16_dct_identity_1_8bpc_c:    10584.9 ( 1.00x)
      inv_txfm_add_16x16_dct_identity_1_8bpc_rvv:   1163.3 ( 9.10x)
      inv_txfm_add_16x16_dct_identity_2_8bpc_c:    10590.3 ( 1.00x)
      inv_txfm_add_16x16_dct_identity_2_8bpc_rvv:   1163.6 ( 9.10x)
      inv_txfm_add_16x16_identity_dct_0_8bpc_c:     9945.9 ( 1.00x)
      inv_txfm_add_16x16_identity_dct_0_8bpc_rvv:   1150.2 ( 8.65x)
      inv_txfm_add_16x16_identity_dct_1_8bpc_c:     9937.0 ( 1.00x)
      inv_txfm_add_16x16_identity_dct_1_8bpc_rvv:   1150.3 ( 8.64x)
      inv_txfm_add_16x16_identity_dct_2_8bpc_c:     9934.6 ( 1.00x)
      inv_txfm_add_16x16_identity_dct_2_8bpc_rvv:   1150.4 ( 8.64x)
      cc29b231
    • Nathan E. Egge's avatar
      riscv64/itx: Add 16-point 8bpc RVV dct transform · 8e82093e
      Nathan E. Egge authored
      inv_txfm_add_16x16_dct_dct_0_8bpc_c:     1574.4 ( 1.00x)
      inv_txfm_add_16x16_dct_dct_0_8bpc_rvv:   1450.3 ( 1.09x)
      inv_txfm_add_16x16_dct_dct_1_8bpc_c:    13614.4 ( 1.00x)
      inv_txfm_add_16x16_dct_dct_1_8bpc_rvv:   1450.5 ( 9.39x)
      inv_txfm_add_16x16_dct_dct_2_8bpc_c:    13613.2 ( 1.00x)
      inv_txfm_add_16x16_dct_dct_2_8bpc_rvv:   1450.4 ( 9.39x)
      8e82093e
    • Nathan E. Egge's avatar
    • Nathan E. Egge's avatar
      57d5729c
    • Nathan E. Egge's avatar
    • Nathan E. Egge's avatar
      riscv64/itx: Add 16-point 8bpc RVV idtx transform · 64c9d160
      Nathan E. Egge authored
      inv_txfm_add_16x16_identity_identity_0_8bpc_c:     6933.8 ( 1.00x)
      inv_txfm_add_16x16_identity_identity_0_8bpc_rvv:    866.0 ( 8.01x)
      inv_txfm_add_16x16_identity_identity_1_8bpc_c:     6933.4 ( 1.00x)
      inv_txfm_add_16x16_identity_identity_1_8bpc_rvv:    866.1 ( 8.01x)
      inv_txfm_add_16x16_identity_identity_2_8bpc_c:     6934.2 ( 1.00x)
      inv_txfm_add_16x16_identity_identity_2_8bpc_rvv:    866.1 ( 8.01x)
      64c9d160
    • Henrik Gramner's avatar
      x86: Add high bit-depth ipred z2 AVX-512 (Ice Lake) asm · 97744bdc
      Henrik Gramner authored and Jean-Baptiste Kempf's avatar Jean-Baptiste Kempf committed
      97744bdc
  2. Feb 13, 2024
  3. Feb 06, 2024
  4. Feb 05, 2024
  5. Feb 02, 2024
  6. Jan 31, 2024
    • Nathan E. Egge's avatar
      a6878be7
    • Nathan E. Egge's avatar
      riscv64/itx: Add 8-point 8bpc RVV flipadst transform · 219befef
      Nathan E. Egge authored
      inv_txfm_add_8x8_adst_flipadst_0_8bpc_c:         3323.1 ( 1.00x)
      inv_txfm_add_8x8_adst_flipadst_0_8bpc_rvv:        402.1 ( 8.26x)
      inv_txfm_add_8x8_adst_flipadst_1_8bpc_c:         3322.8 ( 1.00x)
      inv_txfm_add_8x8_adst_flipadst_1_8bpc_rvv:        402.2 ( 8.26x)
      inv_txfm_add_8x8_dct_flipadst_0_8bpc_c:          3074.3 ( 1.00x)
      inv_txfm_add_8x8_dct_flipadst_0_8bpc_rvv:         359.5 ( 8.55x)
      inv_txfm_add_8x8_dct_flipadst_1_8bpc_c:          3074.4 ( 1.00x)
      inv_txfm_add_8x8_dct_flipadst_1_8bpc_rvv:         359.4 ( 8.56x)
      inv_txfm_add_8x8_flipadst_adst_0_8bpc_c:         3314.8 ( 1.00x)
      inv_txfm_add_8x8_flipadst_adst_0_8bpc_rvv:        403.3 ( 8.22x)
      inv_txfm_add_8x8_flipadst_adst_1_8bpc_c:         3315.3 ( 1.00x)
      inv_txfm_add_8x8_flipadst_adst_1_8bpc_rvv:        403.3 ( 8.22x)
      inv_txfm_add_8x8_flipadst_dct_0_8bpc_c:          3071.7 ( 1.00x)
      inv_txfm_add_8x8_flipadst_dct_0_8bpc_rvv:         359.1 ( 8.55x)
      inv_txfm_add_8x8_flipadst_dct_1_8bpc_c:          3072.5 ( 1.00x)
      inv_txfm_add_8x8_flipadst_dct_1_8bpc_rvv:         359.3 ( 8.55x)
      inv_txfm_add_8x8_flipadst_flipadst_0_8bpc_c:     3325.2 ( 1.00x)
      inv_txfm_add_8x8_flipadst_flipadst_0_8bpc_rvv:    405.2 ( 8.21x)
      inv_txfm_add_8x8_flipadst_flipadst_1_8bpc_c:     3325.0 ( 1.00x)
      inv_txfm_add_8x8_flipadst_flipadst_1_8bpc_rvv:    405.2 ( 8.21x)
      inv_txfm_add_8x8_flipadst_identity_0_8bpc_c:     2356.2 ( 1.00x)
      inv_txfm_add_8x8_flipadst_identity_0_8bpc_rvv:    283.7 ( 8.31x)
      inv_txfm_add_8x8_flipadst_identity_1_8bpc_c:     2356.2 ( 1.00x)
      inv_txfm_add_8x8_flipadst_identity_1_8bpc_rvv:    283.5 ( 8.31x)
      inv_txfm_add_8x8_identity_flipadst_0_8bpc_c:     2332.8 ( 1.00x)
      inv_txfm_add_8x8_identity_flipadst_0_8bpc_rvv:    268.0 ( 8.71x)
      inv_txfm_add_8x8_identity_flipadst_1_8bpc_c:     2331.5 ( 1.00x)
      inv_txfm_add_8x8_identity_flipadst_1_8bpc_rvv:    268.0 ( 8.70x)
      219befef
    • Nathan E. Egge's avatar
      b5747aee
    • Nathan E. Egge's avatar
      riscv64/itx: Add 8-point 8bpc RVV adst transform · 64f9fd02
      Nathan E. Egge authored
      inv_txfm_add_8x8_adst_adst_0_8bpc_c:         3338.5 ( 1.00x)
      inv_txfm_add_8x8_adst_adst_0_8bpc_rvv:        400.4 ( 8.34x)
      inv_txfm_add_8x8_adst_adst_1_8bpc_c:         3338.1 ( 1.00x)
      inv_txfm_add_8x8_adst_adst_1_8bpc_rvv:        399.8 ( 8.35x)
      inv_txfm_add_8x8_adst_dct_0_8bpc_c:          3112.5 ( 1.00x)
      inv_txfm_add_8x8_adst_dct_0_8bpc_rvv:         357.2 ( 8.71x)
      inv_txfm_add_8x8_adst_dct_1_8bpc_c:          3111.4 ( 1.00x)
      inv_txfm_add_8x8_adst_dct_1_8bpc_rvv:         357.0 ( 8.71x)
      inv_txfm_add_8x8_adst_identity_0_8bpc_c:     2375.0 ( 1.00x)
      inv_txfm_add_8x8_adst_identity_0_8bpc_rvv:    281.0 ( 8.45x)
      inv_txfm_add_8x8_adst_identity_1_8bpc_c:     2375.6 ( 1.00x)
      inv_txfm_add_8x8_adst_identity_1_8bpc_rvv:    281.0 ( 8.45x)
      inv_txfm_add_8x8_dct_adst_0_8bpc_c:          3113.3 ( 1.00x)
      inv_txfm_add_8x8_dct_adst_0_8bpc_rvv:         357.2 ( 8.72x)
      inv_txfm_add_8x8_dct_adst_1_8bpc_c:          3112.1 ( 1.00x)
      inv_txfm_add_8x8_dct_adst_1_8bpc_rvv:         357.2 ( 8.71x)
      inv_txfm_add_8x8_identity_adst_0_8bpc_c:     2346.7 ( 1.00x)
      inv_txfm_add_8x8_identity_adst_0_8bpc_rvv:    265.6 ( 8.83x)
      inv_txfm_add_8x8_identity_adst_1_8bpc_c:     2348.3 ( 1.00x)
      inv_txfm_add_8x8_identity_adst_1_8bpc_rvv:    265.8 ( 8.84x)
      64f9fd02
    • Nathan E. Egge's avatar
      riscv64/itx: Add 8x8 8bpc dct_identity and identity_dct · 3e8260d6
      Nathan E. Egge authored
      inv_txfm_add_8x8_dct_identity_0_8bpc_c:     2122.2 ( 1.00x)
      inv_txfm_add_8x8_dct_identity_0_8bpc_rvv:    239.6 ( 8.86x)
      inv_txfm_add_8x8_dct_identity_1_8bpc_c:     2122.6 ( 1.00x)
      inv_txfm_add_8x8_dct_identity_1_8bpc_rvv:    239.4 ( 8.87x)
      inv_txfm_add_8x8_identity_dct_0_8bpc_c:     2093.6 ( 1.00x)
      inv_txfm_add_8x8_identity_dct_0_8bpc_rvv:    224.2 ( 9.34x)
      inv_txfm_add_8x8_identity_dct_1_8bpc_c:     2092.7 ( 1.00x)
      inv_txfm_add_8x8_identity_dct_1_8bpc_rvv:    224.2 ( 9.34x)
      3e8260d6
    • Nathan E. Egge's avatar
      riscv64/itx: Special case 8x8 8bpc dct_dct eob = 0 · 877486e0
      Nathan E. Egge authored
      inv_txfm_add_8x8_dct_dct_0_8bpc_c:      443.1 ( 1.00x)
      inv_txfm_add_8x8_dct_dct_0_8bpc_rvv:     75.3 ( 5.88x)
      inv_txfm_add_8x8_dct_dct_1_8bpc_c:     2796.8 ( 1.00x)
      inv_txfm_add_8x8_dct_dct_1_8bpc_rvv:    315.6 ( 8.86x)
      877486e0
    • Nathan E. Egge's avatar
      riscv64/itx: Add 8-point 8bpc RVV dct transform · 1eaff185
      Nathan E. Egge authored
      inv_txfm_add_8x8_dct_dct_0_8bpc_c:      443.8 ( 1.00x)
      inv_txfm_add_8x8_dct_dct_0_8bpc_rvv:    310.6 ( 1.43x)
      inv_txfm_add_8x8_dct_dct_1_8bpc_c:     2813.0 ( 1.00x)
      inv_txfm_add_8x8_dct_dct_1_8bpc_rvv:    312.3 ( 9.01x)
      1eaff185
    • Nathan E. Egge's avatar
      6ba7e570
    • Nathan E. Egge's avatar
      riscv64/itx: Add 4-point 8bpc RVV flipadst transform · 9b9f2284
      Nathan E. Egge authored
      inv_txfm_add_4x4_adst_flipadst_0_8bpc_c:          686.6 ( 1.00x)
      inv_txfm_add_4x4_adst_flipadst_0_8bpc_rvv:        104.9 ( 6.55x)
      inv_txfm_add_4x4_adst_flipadst_1_8bpc_c:          686.7 ( 1.00x)
      inv_txfm_add_4x4_adst_flipadst_1_8bpc_rvv:        104.9 ( 6.55x)
      inv_txfm_add_4x4_dct_flipadst_0_8bpc_c:           664.3 ( 1.00x)
      inv_txfm_add_4x4_dct_flipadst_0_8bpc_rvv:          94.2 ( 7.05x)
      inv_txfm_add_4x4_dct_flipadst_1_8bpc_c:           663.5 ( 1.00x)
      inv_txfm_add_4x4_dct_flipadst_1_8bpc_rvv:          94.2 ( 7.04x)
      inv_txfm_add_4x4_flipadst_adst_0_8bpc_c:          686.5 ( 1.00x)
      inv_txfm_add_4x4_flipadst_adst_0_8bpc_rvv:        101.4 ( 6.77x)
      inv_txfm_add_4x4_flipadst_adst_1_8bpc_c:          685.6 ( 1.00x)
      inv_txfm_add_4x4_flipadst_adst_1_8bpc_rvv:        101.4 ( 6.76x)
      inv_txfm_add_4x4_flipadst_dct_0_8bpc_c:           664.4 ( 1.00x)
      inv_txfm_add_4x4_flipadst_dct_0_8bpc_rvv:          93.7 ( 7.09x)
      inv_txfm_add_4x4_flipadst_dct_1_8bpc_c:           664.4 ( 1.00x)
      inv_txfm_add_4x4_flipadst_dct_1_8bpc_rvv:          93.6 ( 7.10x)
      inv_txfm_add_4x4_flipadst_flipadst_0_8bpc_c:      691.5 ( 1.00x)
      inv_txfm_add_4x4_flipadst_flipadst_0_8bpc_rvv:    102.5 ( 6.74x)
      inv_txfm_add_4x4_flipadst_flipadst_1_8bpc_c:      691.5 ( 1.00x)
      inv_txfm_add_4x4_flipadst_flipadst_1_8bpc_rvv:    102.5 ( 6.74x)
      inv_txfm_add_4x4_flipadst_identity_0_8bpc_c:      641.8 ( 1.00x)
      inv_txfm_add_4x4_flipadst_identity_0_8bpc_rvv:     80.5 ( 7.97x)
      inv_txfm_add_4x4_flipadst_identity_1_8bpc_c:      641.7 ( 1.00x)
      inv_txfm_add_4x4_flipadst_identity_1_8bpc_rvv:     80.5 ( 7.97x)
      inv_txfm_add_4x4_identity_flipadst_0_8bpc_c:      605.3 ( 1.00x)
      inv_txfm_add_4x4_identity_flipadst_0_8bpc_rvv:     80.5 ( 7.52x)
      inv_txfm_add_4x4_identity_flipadst_1_8bpc_c:      606.3 ( 1.00x)
      inv_txfm_add_4x4_identity_flipadst_1_8bpc_rvv:     80.5 ( 7.53x)
      9b9f2284
    • Nathan E. Egge's avatar
      91c9e6fe
    • Nathan E. Egge's avatar
      riscv64/itx: Add 4-point 8bpc RVV adst transform · 5305eb36
      Nathan E. Egge authored
      inv_txfm_add_4x4_adst_adst_0_8bpc_c:          674.2 ( 1.00x)
      inv_txfm_add_4x4_adst_adst_0_8bpc_rvv:         98.5 ( 6.84x)
      inv_txfm_add_4x4_adst_adst_1_8bpc_c:          674.1 ( 1.00x)
      inv_txfm_add_4x4_adst_adst_1_8bpc_rvv:         98.5 ( 6.84x)
      inv_txfm_add_4x4_adst_dct_0_8bpc_c:           650.3 ( 1.00x)
      inv_txfm_add_4x4_adst_dct_0_8bpc_rvv:          93.1 ( 6.99x)
      inv_txfm_add_4x4_adst_dct_1_8bpc_c:           650.6 ( 1.00x)
      inv_txfm_add_4x4_adst_dct_1_8bpc_rvv:          93.1 ( 6.99x)
      inv_txfm_add_4x4_adst_identity_0_8bpc_c:      635.6 ( 1.00x)
      inv_txfm_add_4x4_adst_identity_0_8bpc_rvv:     79.9 ( 7.95x)
      inv_txfm_add_4x4_adst_identity_1_8bpc_c:      635.4 ( 1.00x)
      inv_txfm_add_4x4_adst_identity_1_8bpc_rvv:     79.9 ( 7.95x)
      inv_txfm_add_4x4_dct_adst_0_8bpc_c:           649.7 ( 1.00x)
      inv_txfm_add_4x4_dct_adst_0_8bpc_rvv:          93.7 ( 6.93x)
      inv_txfm_add_4x4_dct_adst_1_8bpc_c:           649.6 ( 1.00x)
      inv_txfm_add_4x4_dct_adst_1_8bpc_rvv:          93.7 ( 6.93x)
      inv_txfm_add_4x4_identity_adst_0_8bpc_c:      594.1 ( 1.00x)
      inv_txfm_add_4x4_identity_adst_0_8bpc_rvv:     79.8 ( 7.45x)
      inv_txfm_add_4x4_identity_adst_1_8bpc_c:      592.6 ( 1.00x)
      inv_txfm_add_4x4_identity_adst_1_8bpc_rvv:     79.8 ( 7.43x)
      5305eb36
    • Nathan E. Egge's avatar
      riscv64/itx: Improve inv_txfm_add_4x4_rvv performance · c07ccee5
      Nathan E. Egge authored
      Use a higher SEW/LMUL ratio to reduce data path costs.
      
      inv_txfm_add_4x4_dct_dct_0_8bpc_c:                153.5 ( 1.00x)
      inv_txfm_add_4x4_dct_dct_0_8bpc_rvv:               39.4 ( 3.90x)
      inv_txfm_add_4x4_dct_dct_1_8bpc_c:                634.4 ( 1.00x)
      inv_txfm_add_4x4_dct_dct_1_8bpc_rvv:               90.6 ( 7.00x)
      inv_txfm_add_4x4_dct_identity_0_8bpc_c:           621.0 ( 1.00x)
      inv_txfm_add_4x4_dct_identity_0_8bpc_rvv:          74.2 ( 8.37x)
      inv_txfm_add_4x4_dct_identity_1_8bpc_c:           619.8 ( 1.00x)
      inv_txfm_add_4x4_dct_identity_1_8bpc_rvv:          74.2 ( 8.35x)
      inv_txfm_add_4x4_identity_dct_0_8bpc_c:           574.3 ( 1.00x)
      inv_txfm_add_4x4_identity_dct_0_8bpc_rvv:          73.6 ( 7.80x)
      inv_txfm_add_4x4_identity_dct_1_8bpc_c:           574.5 ( 1.00x)
      inv_txfm_add_4x4_identity_dct_1_8bpc_rvv:          73.6 ( 7.80x)
      inv_txfm_add_4x4_identity_identity_0_8bpc_c:      548.8 ( 1.00x)
      inv_txfm_add_4x4_identity_identity_0_8bpc_rvv:     60.5 ( 9.08x)
      inv_txfm_add_4x4_identity_identity_1_8bpc_c:      548.7 ( 1.00x)
      inv_txfm_add_4x4_identity_identity_1_8bpc_rvv:     60.5 ( 9.07x)
      c07ccee5
    • Nathan E. Egge's avatar
      riscv64/itx: Add 4x4 8bpc dct_identity and identity_dct · 802d9257
      Nathan E. Egge authored
      inv_txfm_add_4x4_dct_identity_0_8bpc_c:      619.6 ( 1.00x)
      inv_txfm_add_4x4_dct_identity_0_8bpc_rvv:     95.6 ( 6.48x)
      inv_txfm_add_4x4_dct_identity_1_8bpc_c:      620.0 ( 1.00x)
      inv_txfm_add_4x4_dct_identity_1_8bpc_rvv:     95.6 ( 6.49x)
      inv_txfm_add_4x4_identity_dct_0_8bpc_c:      573.9 ( 1.00x)
      inv_txfm_add_4x4_identity_dct_0_8bpc_rvv:     98.1 ( 5.85x)
      inv_txfm_add_4x4_identity_dct_1_8bpc_c:      573.1 ( 1.00x)
      inv_txfm_add_4x4_identity_dct_1_8bpc_rvv:     98.1 ( 5.84x)
      802d9257
    • Nathan E. Egge's avatar
    • Nathan E. Egge's avatar
      riscv64/itx: Special case 4x4 8bpc dct_dct eob = 0 · c436b8e3
      Nathan E. Egge authored
      inv_txfm_add_4x4_dct_dct_0_8bpc_c:      152.1 ( 1.00x)
      inv_txfm_add_4x4_dct_dct_0_8bpc_rvv:     46.9 ( 3.25x)
      inv_txfm_add_4x4_dct_dct_1_8bpc_c:      622.4 ( 1.00x)
      inv_txfm_add_4x4_dct_dct_1_8bpc_rvv:    120.7 ( 5.16x)
      c436b8e3
    • Nathan E. Egge's avatar
      riscv64/itx: Add 4-point 8bpc RVV dct transform · 04d4d50b
      Nathan E. Egge authored
      inv_txfm_add_4x4_dct_dct_0_8bpc_c:      154.6 ( 1.00x)
      inv_txfm_add_4x4_dct_dct_0_8bpc_rvv:    121.3 ( 1.27x)
      inv_txfm_add_4x4_dct_dct_1_8bpc_c:      619.5 ( 1.00x)
      inv_txfm_add_4x4_dct_dct_1_8bpc_rvv:    121.3 ( 5.11x)
      04d4d50b
    • Nathan E. Egge's avatar
      riscv64/itx: Add 8-point 8bpc RVV idtx transform · 80d78c80
      Nathan E. Egge authored
      inv_txfm_add_8x8_identity_identity_0_8bpc_c:     1344.6 ( 1.00x)
      inv_txfm_add_8x8_identity_identity_0_8bpc_rvv:    144.4 ( 9.31x)
      inv_txfm_add_8x8_identity_identity_1_8bpc_c:     1344.4 ( 1.00x)
      inv_txfm_add_8x8_identity_identity_1_8bpc_rvv:    144.4 ( 9.31x)
      80d78c80
    • Matthias Dressel's avatar
      CI: Use cross-compiling libc instead of multi-arch · a7edb029
      Matthias Dressel authored and Nathan E. Egge's avatar Nathan E. Egge committed
      See videolan/docker-images!272
      for more context.
      a7edb029
    • Nathan E. Egge's avatar
      1042008e
    • Nathan E. Egge's avatar
      e67f6306
Loading