Skip to content

x86: Add high bitdepth CFL SSSE3 asm

Henrik Gramner requested to merge gramner/dav1d:cfl16_ssse3 into master
cfl_ac_420_w4_16bpc_c: 267.5
cfl_ac_420_w4_16bpc_ssse3: 28.4
cfl_ac_420_w4_16bpc_avx2: 36.2
cfl_ac_420_w8_16bpc_c: 360.3
cfl_ac_420_w8_16bpc_ssse3: 49.9
cfl_ac_420_w8_16bpc_avx2: 45.1
cfl_ac_420_w16_16bpc_c: 625.3
cfl_ac_420_w16_16bpc_ssse3: 84.0
cfl_ac_420_w16_16bpc_avx2: 55.9

cfl_ac_422_w4_16bpc_c: 214.0
cfl_ac_422_w4_16bpc_ssse3: 20.8
cfl_ac_422_w4_16bpc_avx2: 21.0
cfl_ac_422_w8_16bpc_c: 412.7
cfl_ac_422_w8_16bpc_ssse3: 52.7
cfl_ac_422_w8_16bpc_avx2: 36.6
cfl_ac_422_w16_16bpc_c: 710.2
cfl_ac_422_w16_16bpc_ssse3: 104.6
cfl_ac_422_w16_16bpc_avx2: 65.1

cfl_ac_444_w4_16bpc_c: 201.4
cfl_ac_444_w4_16bpc_ssse3: 18.7
cfl_ac_444_w8_16bpc_c: 335.1
cfl_ac_444_w8_16bpc_ssse3: 40.9
cfl_ac_444_w16_16bpc_c: 496.1
cfl_ac_444_w16_16bpc_ssse3: 81.0
cfl_ac_444_w32_16bpc_c: 1412.9
cfl_ac_444_w32_16bpc_ssse3: 194.3

cfl_pred_cfl_128_w4_16bpc_c: 206.0
cfl_pred_cfl_128_w4_16bpc_ssse3: 18.3
cfl_pred_cfl_128_w4_16bpc_avx2: 15.1
cfl_pred_cfl_128_w8_16bpc_c: 364.0
cfl_pred_cfl_128_w8_16bpc_ssse3: 48.1
cfl_pred_cfl_128_w8_16bpc_avx2: 26.8
cfl_pred_cfl_128_w16_16bpc_c: 692.0
cfl_pred_cfl_128_w16_16bpc_ssse3: 81.0
cfl_pred_cfl_128_w16_16bpc_avx2: 41.3
cfl_pred_cfl_128_w32_16bpc_c: 1677.1
cfl_pred_cfl_128_w32_16bpc_ssse3: 184.2
cfl_pred_cfl_128_w32_16bpc_avx2: 97.2

cfl_pred_cfl_left_w4_16bpc_c: 222.2
cfl_pred_cfl_left_w4_16bpc_ssse3: 22.4
cfl_pred_cfl_left_w4_16bpc_avx2: 20.1
cfl_pred_cfl_left_w8_16bpc_c: 384.9
cfl_pred_cfl_left_w8_16bpc_ssse3: 52.0
cfl_pred_cfl_left_w8_16bpc_avx2: 31.4
cfl_pred_cfl_left_w16_16bpc_c: 716.5
cfl_pred_cfl_left_w16_16bpc_ssse3: 84.8
cfl_pred_cfl_left_w16_16bpc_avx2: 46.6
cfl_pred_cfl_left_w32_16bpc_c: 1705.3
cfl_pred_cfl_left_w32_16bpc_ssse3: 188.0
cfl_pred_cfl_left_w32_16bpc_avx2: 107.7

cfl_pred_cfl_top_w4_16bpc_c: 212.6
cfl_pred_cfl_top_w4_16bpc_ssse3: 21.5
cfl_pred_cfl_top_w4_16bpc_avx2: 18.3
cfl_pred_cfl_top_w8_16bpc_c: 371.0
cfl_pred_cfl_top_w8_16bpc_ssse3: 51.3
cfl_pred_cfl_top_w8_16bpc_avx2: 30.1
cfl_pred_cfl_top_w16_16bpc_c: 700.5
cfl_pred_cfl_top_w16_16bpc_ssse3: 84.7
cfl_pred_cfl_top_w16_16bpc_avx2: 45.6
cfl_pred_cfl_top_w32_16bpc_c: 1692.4
cfl_pred_cfl_top_w32_16bpc_ssse3: 188.7
cfl_pred_cfl_top_w32_16bpc_avx2: 107.6

cfl_pred_cfl_w4_16bpc_c: 223.0
cfl_pred_cfl_w4_16bpc_ssse3: 24.4
cfl_pred_cfl_w4_16bpc_avx2: 23.9
cfl_pred_cfl_w8_16bpc_c: 387.7
cfl_pred_cfl_w8_16bpc_ssse3: 53.5
cfl_pred_cfl_w8_16bpc_avx2: 33.0
cfl_pred_cfl_w16_16bpc_c: 727.9
cfl_pred_cfl_w16_16bpc_ssse3: 86.2
cfl_pred_cfl_w16_16bpc_avx2: 50.3
cfl_pred_cfl_w32_16bpc_c: 1717.3
cfl_pred_cfl_w32_16bpc_ssse3: 190.7
cfl_pred_cfl_w32_16bpc_avx2: 109.7

Merge request reports