Skip to content

x86: Add high bit-depth ipred AVX-512 (Ice Lake) asm

Henrik Gramner requested to merge gramner/dav1d:avx512_ipred_16bpc into master
intra_pred_filter_w4_10bpc_avx2: 63.7
intra_pred_filter_w4_10bpc_avx512icl: 47.3
intra_pred_filter_w8_10bpc_avx2: 192.5
intra_pred_filter_w8_10bpc_avx512icl: 117.1
intra_pred_filter_w16_10bpc_avx2: 291.4
intra_pred_filter_w16_10bpc_avx512icl: 256.7
intra_pred_filter_w32_10bpc_avx2: 734.8
intra_pred_filter_w32_10bpc_avx512icl: 684.3

intra_pred_paeth_w4_16bpc_avx2: 15.4
intra_pred_paeth_w4_16bpc_avx512icl: 13.5
intra_pred_paeth_w8_16bpc_avx2: 33.1
intra_pred_paeth_w8_16bpc_avx512icl: 20.6
intra_pred_paeth_w16_16bpc_avx2: 89.5
intra_pred_paeth_w16_16bpc_avx512icl: 55.3
intra_pred_paeth_w32_16bpc_avx2: 203.1
intra_pred_paeth_w32_16bpc_avx512icl: 127.1
intra_pred_paeth_w64_16bpc_avx2: 492.2
intra_pred_paeth_w64_16bpc_avx512icl: 293.3

intra_pred_smooth_h_w4_16bpc_avx2: 10.4
intra_pred_smooth_h_w4_16bpc_avx512icl: 10.4
intra_pred_smooth_h_w8_16bpc_avx2: 15.7
intra_pred_smooth_h_w8_16bpc_avx512icl: 14.7
intra_pred_smooth_h_w16_16bpc_avx2: 28.6
intra_pred_smooth_h_w16_16bpc_avx512icl: 24.5
intra_pred_smooth_h_w32_16bpc_avx2: 55.5
intra_pred_smooth_h_w32_16bpc_avx512icl: 42.0
intra_pred_smooth_h_w64_16bpc_avx2: 121.5
intra_pred_smooth_h_w64_16bpc_avx512icl: 87.1

intra_pred_smooth_v_w4_16bpc_avx2: 10.5
intra_pred_smooth_v_w4_16bpc_avx512icl: 10.0
intra_pred_smooth_v_w8_16bpc_avx2: 15.0
intra_pred_smooth_v_w8_16bpc_avx512icl: 13.9
intra_pred_smooth_v_w16_16bpc_avx2: 28.7
intra_pred_smooth_v_w16_16bpc_avx512icl: 22.9
intra_pred_smooth_v_w32_16bpc_avx2: 53.2
intra_pred_smooth_v_w32_16bpc_avx512icl: 37.4
intra_pred_smooth_v_w64_16bpc_avx2: 109.0
intra_pred_smooth_v_w64_16bpc_avx512icl: 75.3

intra_pred_smooth_w4_16bpc_avx2: 18.8
intra_pred_smooth_w4_16bpc_avx512icl: 15.2
intra_pred_smooth_w8_16bpc_avx2: 36.5
intra_pred_smooth_w8_16bpc_avx512icl: 25.0
intra_pred_smooth_w16_16bpc_avx2: 82.2
intra_pred_smooth_w16_16bpc_avx512icl: 53.2
intra_pred_smooth_w32_16bpc_avx2: 182.5
intra_pred_smooth_w32_16bpc_avx512icl: 112.6
intra_pred_smooth_w64_16bpc_avx2: 449.9
intra_pred_smooth_w64_16bpc_avx512icl: 276.7

pal_pred_w4_16bpc_avx2: 7.9
pal_pred_w4_16bpc_avx512icl: 8.2
pal_pred_w8_16bpc_avx2: 12.0
pal_pred_w8_16bpc_avx512icl: 11.8
pal_pred_w16_16bpc_avx2: 27.8
pal_pred_w16_16bpc_avx512icl: 20.1
pal_pred_w32_16bpc_avx2: 63.3
pal_pred_w32_16bpc_avx512icl: 38.5
pal_pred_w64_16bpc_avx2: 146.8
pal_pred_w64_16bpc_avx512icl: 88.7

Merge request reports