Skip to content

x86: Add 10-bit 8x8/8x16/16x8/16x16 itx AVX-512 (Ice Lake) asm

Henrik Gramner requested to merge gramner/dav1d:itx16_avx512icl into master
inv_txfm_add_8x8_adst_adst_1_10bpc_avx2:      123.4
inv_txfm_add_8x8_adst_adst_1_10bpc_avx512icl:  99.8

inv_txfm_add_8x8_adst_dct_1_10bpc_avx2:      92.5
inv_txfm_add_8x8_adst_dct_1_10bpc_avx512icl: 93.0

inv_txfm_add_8x8_adst_flipadst_1_10bpc_avx2:      102.6
inv_txfm_add_8x8_adst_flipadst_1_10bpc_avx512icl: 100.0

inv_txfm_add_8x8_adst_identity_1_10bpc_avx2:      69.8
inv_txfm_add_8x8_adst_identity_1_10bpc_avx512icl: 61.1

inv_txfm_add_8x8_dct_adst_1_10bpc_avx2:      83.6
inv_txfm_add_8x8_dct_adst_1_10bpc_avx512icl: 79.8

inv_txfm_add_8x8_dct_dct_0_10bpc_avx2:      14.1
inv_txfm_add_8x8_dct_dct_0_10bpc_avx512icl: 12.4
inv_txfm_add_8x8_dct_dct_1_10bpc_avx2:      72.4
inv_txfm_add_8x8_dct_dct_1_10bpc_avx512icl: 71.4

inv_txfm_add_8x8_dct_flipadst_1_10bpc_avx2:      105.4
inv_txfm_add_8x8_dct_flipadst_1_10bpc_avx512icl:  79.3

inv_txfm_add_8x8_dct_identity_1_10bpc_avx2:      51.0
inv_txfm_add_8x8_dct_identity_1_10bpc_avx512icl: 44.7

inv_txfm_add_8x8_flipadst_adst_1_10bpc_avx2:      103.0
inv_txfm_add_8x8_flipadst_adst_1_10bpc_avx512icl:  99.8

inv_txfm_add_8x8_flipadst_dct_1_10bpc_avx2:      92.1
inv_txfm_add_8x8_flipadst_dct_1_10bpc_avx512icl: 92.7

inv_txfm_add_8x8_flipadst_flipadst_1_10bpc_avx2:      103.1
inv_txfm_add_8x8_flipadst_flipadst_1_10bpc_avx512icl:  99.9

inv_txfm_add_8x8_flipadst_identity_1_10bpc_avx2:      70.0
inv_txfm_add_8x8_flipadst_identity_1_10bpc_avx512icl: 60.7

inv_txfm_add_8x8_identity_adst_1_10bpc_avx2:      52.6
inv_txfm_add_8x8_identity_adst_1_10bpc_avx512icl: 44.3

inv_txfm_add_8x8_identity_dct_1_10bpc_avx2:      41.0
inv_txfm_add_8x8_identity_dct_1_10bpc_avx512icl: 33.3

inv_txfm_add_8x8_identity_flipadst_1_10bpc_avx2:      52.5
inv_txfm_add_8x8_identity_flipadst_1_10bpc_avx512icl: 43.9

inv_txfm_add_8x8_identity_identity_1_10bpc_avx2:      21.3
inv_txfm_add_8x8_identity_identity_1_10bpc_avx512icl: 15.3



inv_txfm_add_8x16_adst_adst_1_10bpc_avx2:      152.5
inv_txfm_add_8x16_adst_adst_1_10bpc_avx512icl: 126.4
inv_txfm_add_8x16_adst_adst_2_10bpc_avx2:      206.1
inv_txfm_add_8x16_adst_adst_2_10bpc_avx512icl: 150.3

inv_txfm_add_8x16_adst_dct_1_10bpc_avx2:      137.9
inv_txfm_add_8x16_adst_dct_1_10bpc_avx512icl: 135.7
inv_txfm_add_8x16_adst_dct_2_10bpc_avx2:      193.5
inv_txfm_add_8x16_adst_dct_2_10bpc_avx512icl: 158.0

inv_txfm_add_8x16_adst_flipadst_1_10bpc_avx2:      152.4
inv_txfm_add_8x16_adst_flipadst_1_10bpc_avx512icl: 126.6
inv_txfm_add_8x16_adst_flipadst_2_10bpc_avx2:      206.8
inv_txfm_add_8x16_adst_flipadst_2_10bpc_avx512icl: 149.5

inv_txfm_add_8x16_adst_identity_1_10bpc_avx2:       93.9
inv_txfm_add_8x16_adst_identity_1_10bpc_avx512icl:  90.3
inv_txfm_add_8x16_adst_identity_2_10bpc_avx2:      149.2
inv_txfm_add_8x16_adst_identity_2_10bpc_avx512icl: 111.9

inv_txfm_add_8x16_dct_adst_1_10bpc_avx2:      135.0
inv_txfm_add_8x16_dct_adst_1_10bpc_avx512icl: 107.9
inv_txfm_add_8x16_dct_adst_2_10bpc_avx2:      176.6
inv_txfm_add_8x16_dct_adst_2_10bpc_avx512icl: 122.7

inv_txfm_add_8x16_dct_dct_0_10bpc_avx2:       19.8
inv_txfm_add_8x16_dct_dct_0_10bpc_avx512icl:  19.7
inv_txfm_add_8x16_dct_dct_1_10bpc_avx2:      121.7
inv_txfm_add_8x16_dct_dct_1_10bpc_avx512icl: 116.9
inv_txfm_add_8x16_dct_dct_2_10bpc_avx2:      162.7
inv_txfm_add_8x16_dct_dct_2_10bpc_avx512icl: 132.5

inv_txfm_add_8x16_dct_flipadst_1_10bpc_avx2:      135.0
inv_txfm_add_8x16_dct_flipadst_1_10bpc_avx512icl: 107.7
inv_txfm_add_8x16_dct_flipadst_2_10bpc_avx2:      176.5
inv_txfm_add_8x16_dct_flipadst_2_10bpc_avx512icl: 123.0

inv_txfm_add_8x16_dct_identity_1_10bpc_avx2:       79.4
inv_txfm_add_8x16_dct_identity_1_10bpc_avx512icl:  74.0
inv_txfm_add_8x16_dct_identity_2_10bpc_avx2:      121.4
inv_txfm_add_8x16_dct_identity_2_10bpc_avx512icl:  87.9

inv_txfm_add_8x16_flipadst_adst_1_10bpc_avx2:      153.1
inv_txfm_add_8x16_flipadst_adst_1_10bpc_avx512icl: 126.5
inv_txfm_add_8x16_flipadst_adst_2_10bpc_avx2:      207.1
inv_txfm_add_8x16_flipadst_adst_2_10bpc_avx512icl: 149.0

inv_txfm_add_8x16_flipadst_dct_1_10bpc_avx2:      139.1
inv_txfm_add_8x16_flipadst_dct_1_10bpc_avx512icl: 135.5
inv_txfm_add_8x16_flipadst_dct_2_10bpc_avx2:      193.6
inv_txfm_add_8x16_flipadst_dct_2_10bpc_avx512icl: 157.6

inv_txfm_add_8x16_flipadst_flipadst_1_10bpc_avx2:      158.7
inv_txfm_add_8x16_flipadst_flipadst_1_10bpc_avx512icl: 127.2
inv_txfm_add_8x16_flipadst_flipadst_2_10bpc_avx2:      207.0
inv_txfm_add_8x16_flipadst_flipadst_2_10bpc_avx512icl: 149.1

inv_txfm_add_8x16_flipadst_identity_1_10bpc_avx2:       94.2
inv_txfm_add_8x16_flipadst_identity_1_10bpc_avx512icl:  90.5
inv_txfm_add_8x16_flipadst_identity_2_10bpc_avx2:      149.7
inv_txfm_add_8x16_flipadst_identity_2_10bpc_avx512icl: 111.5

inv_txfm_add_8x16_identity_adst_2_10bpc_avx2:      117.3
inv_txfm_add_8x16_identity_adst_2_10bpc_avx512icl:  76.3

inv_txfm_add_8x16_identity_dct_2_10bpc_avx2:      100.3
inv_txfm_add_8x16_identity_dct_2_10bpc_avx512icl:  84.2

inv_txfm_add_8x16_identity_flipadst_2_10bpc_avx2:      116.2
inv_txfm_add_8x16_identity_flipadst_2_10bpc_avx512icl:  76.7

inv_txfm_add_8x16_identity_identity_2_10bpc_avx2:      57.7
inv_txfm_add_8x16_identity_identity_2_10bpc_avx512icl: 39.1



inv_txfm_add_16x8_adst_adst_1_10bpc_avx2:      210.4
inv_txfm_add_16x8_adst_adst_1_10bpc_avx512icl: 155.6
inv_txfm_add_16x8_adst_adst_2_10bpc_avx2:      214.7
inv_txfm_add_16x8_adst_adst_2_10bpc_avx512icl: 174.0

inv_txfm_add_16x8_adst_dct_1_10bpc_avx2:      203.4
inv_txfm_add_16x8_adst_dct_1_10bpc_avx512icl: 147.7
inv_txfm_add_16x8_adst_dct_2_10bpc_avx2:      207.5
inv_txfm_add_16x8_adst_dct_2_10bpc_avx512icl: 166.1

inv_txfm_add_16x8_adst_flipadst_1_10bpc_avx2:      210.9
inv_txfm_add_16x8_adst_flipadst_1_10bpc_avx512icl: 155.2
inv_txfm_add_16x8_adst_flipadst_2_10bpc_avx2:      210.9
inv_txfm_add_16x8_adst_flipadst_2_10bpc_avx512icl: 173.7

inv_txfm_add_16x8_adst_identity_1_10bpc_avx2:      168.9
inv_txfm_add_16x8_adst_identity_1_10bpc_avx512icl: 122.8
inv_txfm_add_16x8_adst_identity_2_10bpc_avx2:      169.3
inv_txfm_add_16x8_adst_identity_2_10bpc_avx512icl: 136.7

inv_txfm_add_16x8_dct_adst_1_10bpc_avx2:      179.9
inv_txfm_add_16x8_dct_adst_1_10bpc_avx512icl: 126.4
inv_txfm_add_16x8_dct_adst_2_10bpc_avx2:      180.5
inv_txfm_add_16x8_dct_adst_2_10bpc_avx512icl: 144.3

inv_txfm_add_16x8_dct_dct_0_10bpc_avx2:       16.6
inv_txfm_add_16x8_dct_dct_0_10bpc_avx512icl:  16.8
inv_txfm_add_16x8_dct_dct_1_10bpc_avx2:      173.3
inv_txfm_add_16x8_dct_dct_1_10bpc_avx512icl: 119.4
inv_txfm_add_16x8_dct_dct_2_10bpc_avx2:      172.9
inv_txfm_add_16x8_dct_dct_2_10bpc_avx512icl: 136.1

inv_txfm_add_16x8_dct_flipadst_1_10bpc_avx2:      181.3
inv_txfm_add_16x8_dct_flipadst_1_10bpc_avx512icl: 126.4
inv_txfm_add_16x8_dct_flipadst_2_10bpc_avx2:      185.6
inv_txfm_add_16x8_dct_flipadst_2_10bpc_avx512icl: 144.4

inv_txfm_add_16x8_dct_identity_1_10bpc_avx2:      139.0
inv_txfm_add_16x8_dct_identity_1_10bpc_avx512icl:  93.8
inv_txfm_add_16x8_dct_identity_2_10bpc_avx2:      139.6
inv_txfm_add_16x8_dct_identity_2_10bpc_avx512icl: 106.9

inv_txfm_add_16x8_flipadst_adst_1_10bpc_avx2:      214.7
inv_txfm_add_16x8_flipadst_adst_1_10bpc_avx512icl: 155.2
inv_txfm_add_16x8_flipadst_adst_2_10bpc_avx2:      210.7
inv_txfm_add_16x8_flipadst_adst_2_10bpc_avx512icl: 174.5

inv_txfm_add_16x8_flipadst_dct_1_10bpc_avx2:      203.7
inv_txfm_add_16x8_flipadst_dct_1_10bpc_avx512icl: 147.8
inv_txfm_add_16x8_flipadst_dct_2_10bpc_avx2:      203.7
inv_txfm_add_16x8_flipadst_dct_2_10bpc_avx512icl: 166.3

inv_txfm_add_16x8_flipadst_flipadst_1_10bpc_avx2:      211.6
inv_txfm_add_16x8_flipadst_flipadst_1_10bpc_avx512icl: 155.1
inv_txfm_add_16x8_flipadst_flipadst_2_10bpc_avx2:      211.4
inv_txfm_add_16x8_flipadst_flipadst_2_10bpc_avx512icl: 174.9

inv_txfm_add_16x8_flipadst_identity_1_10bpc_avx2:      169.5
inv_txfm_add_16x8_flipadst_identity_1_10bpc_avx512icl: 122.5
inv_txfm_add_16x8_flipadst_identity_2_10bpc_avx2:      169.6
inv_txfm_add_16x8_flipadst_identity_2_10bpc_avx512icl: 137.2

inv_txfm_add_16x8_identity_adst_1_10bpc_avx2:      108.2
inv_txfm_add_16x8_identity_adst_1_10bpc_avx512icl:  87.0
inv_txfm_add_16x8_identity_adst_2_10bpc_avx2:      108.0
inv_txfm_add_16x8_identity_adst_2_10bpc_avx512icl:  86.9

inv_txfm_add_16x8_identity_dct_1_10bpc_avx2:       99.7
inv_txfm_add_16x8_identity_dct_1_10bpc_avx512icl:  78.3
inv_txfm_add_16x8_identity_dct_2_10bpc_avx2:      100.1
inv_txfm_add_16x8_identity_dct_2_10bpc_avx512icl:  78.5

inv_txfm_add_16x8_identity_flipadst_1_10bpc_avx2:      108.6
inv_txfm_add_16x8_identity_flipadst_1_10bpc_avx512icl:  87.0
inv_txfm_add_16x8_identity_flipadst_2_10bpc_avx2:      108.9
inv_txfm_add_16x8_identity_flipadst_2_10bpc_avx512icl:  86.8

inv_txfm_add_16x8_identity_identity_1_10bpc_avx2:      67.1
inv_txfm_add_16x8_identity_identity_1_10bpc_avx512icl: 48.2
inv_txfm_add_16x8_identity_identity_2_10bpc_avx2:      66.5
inv_txfm_add_16x8_identity_identity_2_10bpc_avx512icl: 48.3



inv_txfm_add_16x16_adst_adst_1_10bpc_avx2:      258.5
inv_txfm_add_16x16_adst_adst_1_10bpc_avx512icl: 188.0
inv_txfm_add_16x16_adst_adst_2_10bpc_avx2:      391.9
inv_txfm_add_16x16_adst_adst_2_10bpc_avx512icl: 299.0

inv_txfm_add_16x16_adst_dct_1_10bpc_avx2:      233.6
inv_txfm_add_16x16_adst_dct_1_10bpc_avx512icl: 170.8
inv_txfm_add_16x16_adst_dct_2_10bpc_avx2:      365.6
inv_txfm_add_16x16_adst_dct_2_10bpc_avx512icl: 280.6

inv_txfm_add_16x16_adst_flipadst_1_10bpc_avx2:      258.6
inv_txfm_add_16x16_adst_flipadst_1_10bpc_avx512icl: 188.3
inv_txfm_add_16x16_adst_flipadst_2_10bpc_avx2:      392.0
inv_txfm_add_16x16_adst_flipadst_2_10bpc_avx512icl: 300.3

inv_txfm_add_16x16_dct_adst_1_10bpc_avx2:      228.3
inv_txfm_add_16x16_dct_adst_1_10bpc_avx512icl: 159.6
inv_txfm_add_16x16_dct_adst_2_10bpc_avx2:      329.0
inv_txfm_add_16x16_dct_adst_2_10bpc_avx512icl: 246.7

inv_txfm_add_16x16_dct_dct_0_10bpc_avx2:       21.5
inv_txfm_add_16x16_dct_dct_0_10bpc_avx512icl:  20.7
inv_txfm_add_16x16_dct_dct_1_10bpc_avx2:      203.0
inv_txfm_add_16x16_dct_dct_1_10bpc_avx512icl: 142.2
inv_txfm_add_16x16_dct_dct_2_10bpc_avx2:      306.7
inv_txfm_add_16x16_dct_dct_2_10bpc_avx512icl: 228.3

inv_txfm_add_16x16_dct_flipadst_1_10bpc_avx2:      228.5
inv_txfm_add_16x16_dct_flipadst_1_10bpc_avx512icl: 159.4
inv_txfm_add_16x16_dct_flipadst_2_10bpc_avx2:      328.4
inv_txfm_add_16x16_dct_flipadst_2_10bpc_avx512icl: 246.6

inv_txfm_add_16x16_dct_identity_1_10bpc_avx2:      140.7
inv_txfm_add_16x16_dct_identity_1_10bpc_avx512icl: 110.8
inv_txfm_add_16x16_dct_identity_2_10bpc_avx2:      245.3
inv_txfm_add_16x16_dct_identity_2_10bpc_avx512icl: 201.2

inv_txfm_add_16x16_flipadst_adst_1_10bpc_avx2:      258.9
inv_txfm_add_16x16_flipadst_adst_1_10bpc_avx512icl: 188.2
inv_txfm_add_16x16_flipadst_adst_2_10bpc_avx2:      390.4
inv_txfm_add_16x16_flipadst_adst_2_10bpc_avx512icl: 301.3

inv_txfm_add_16x16_flipadst_dct_1_10bpc_avx2:      234.1
inv_txfm_add_16x16_flipadst_dct_1_10bpc_avx512icl: 171.1
inv_txfm_add_16x16_flipadst_dct_2_10bpc_avx2:      365.4
inv_txfm_add_16x16_flipadst_dct_2_10bpc_avx512icl: 282.8

inv_txfm_add_16x16_flipadst_flipadst_1_10bpc_avx2:      259.1
inv_txfm_add_16x16_flipadst_flipadst_1_10bpc_avx512icl: 188.0
inv_txfm_add_16x16_flipadst_flipadst_2_10bpc_avx2:      391.8
inv_txfm_add_16x16_flipadst_flipadst_2_10bpc_avx512icl: 301.3

inv_txfm_add_16x16_identity_dct_1_10bpc_avx2:      132.6
inv_txfm_add_16x16_identity_dct_1_10bpc_avx512icl:  99.3
inv_txfm_add_16x16_identity_dct_2_10bpc_avx2:      165.8
inv_txfm_add_16x16_identity_dct_2_10bpc_avx512icl: 122.9

inv_txfm_add_16x16_identity_identity_1_10bpc_avx2:       69.2
inv_txfm_add_16x16_identity_identity_1_10bpc_avx512icl:  60.0
inv_txfm_add_16x16_identity_identity_2_10bpc_avx2:      106.7
inv_txfm_add_16x16_identity_identity_2_10bpc_avx512icl:  93.7

Merge request reports

Loading