Skip to content

itx: restrict number of columns iterated over based on EOB

Ronald S. Bultje requested to merge rbultje/dav1d:itx-c-shortcut into master

(Average) runtime for check --test=itx_8bpc --bench before vs. after changes from 10.30 sec to 8.86 sec, or a 16.3% overall improvement.

Detailed checkasm --test=itx_8bpc --bench 12345 before vs. after shows that most of the improvement is in partial non-DC-only inverse transform calls. I've only shown the 8bpc runtimes here, but the bigger relevance here is for 12bpc content (including fuzzing and CI testing), because we have limited SIMD there. Both before and after were run 5 times and the lowest number was selected. Average improvement is 32.0%.

Function Before After Improvement
inv_txfm_add_4x4_adst_adst_0_8bpc_c 159.5 120.8 32.04%
inv_txfm_add_4x4_adst_adst_1_8bpc_c 163.2 151.2 7.94%
inv_txfm_add_4x4_adst_dct_0_8bpc_c 166.3 126.3 31.67%
inv_txfm_add_4x4_adst_dct_1_8bpc_c 166.6 156 6.79%
inv_txfm_add_4x4_adst_flipadst_0_8bpc_c 163.9 124.4 31.75%
inv_txfm_add_4x4_adst_flipadst_1_8bpc_c 163.8 152.9 7.13%
inv_txfm_add_4x4_adst_identity_0_8bpc_c 145.7 103.4 40.91%
inv_txfm_add_4x4_adst_identity_1_8bpc_c 145.1 133 9.10%
inv_txfm_add_4x4_dct_adst_0_8bpc_c 167.6 120.6 38.97%
inv_txfm_add_4x4_dct_adst_1_8bpc_c 167.6 156.5 7.09%
inv_txfm_add_4x4_dct_dct_0_8bpc_c 38.1 37 2.97%
inv_txfm_add_4x4_dct_dct_1_8bpc_c 172.9 161.7 6.93%
inv_txfm_add_4x4_dct_flipadst_0_8bpc_c 169.1 123.8 36.59%
inv_txfm_add_4x4_dct_flipadst_1_8bpc_c 169.4 157.6 7.49%
inv_txfm_add_4x4_dct_identity_0_8bpc_c 150 104.5 43.54%
inv_txfm_add_4x4_dct_identity_1_8bpc_c 149.9 138.7 8.07%
inv_txfm_add_4x4_flipadst_adst_0_8bpc_c 164 122.1 34.32%
inv_txfm_add_4x4_flipadst_adst_1_8bpc_c 163.9 153.5 6.78%
inv_txfm_add_4x4_flipadst_dct_0_8bpc_c 169.2 127.3 32.91%
inv_txfm_add_4x4_flipadst_dct_1_8bpc_c 169.2 156.3 8.25%
inv_txfm_add_4x4_flipadst_flipadst_0_8bpc_c 165.4 121.6 36.02%
inv_txfm_add_4x4_flipadst_flipadst_1_8bpc_c 165.4 152.9 8.18%
inv_txfm_add_4x4_flipadst_identity_0_8bpc_c 146.1 102.7 42.26%
inv_txfm_add_4x4_flipadst_identity_1_8bpc_c 146.2 133.7 9.35%
inv_txfm_add_4x4_identity_adst_0_8bpc_c 143.5 116.3 23.39%
inv_txfm_add_4x4_identity_adst_1_8bpc_c 143.7 128.2 12.09%
inv_txfm_add_4x4_identity_dct_0_8bpc_c 148.4 123.1 20.55%
inv_txfm_add_4x4_identity_dct_1_8bpc_c 149.1 133.6 11.60%
inv_txfm_add_4x4_identity_flipadst_0_8bpc_c 145 119.7 21.14%
inv_txfm_add_4x4_identity_flipadst_1_8bpc_c 145.1 130.7 11.02%
inv_txfm_add_4x4_identity_identity_0_8bpc_c 125.7 100.2 25.45%
inv_txfm_add_4x4_identity_identity_1_8bpc_c 125.3 112.4 11.48%
inv_txfm_add_4x4_wht_wht_0_8bpc_c 53 52.4 1.15%
inv_txfm_add_4x4_wht_wht_1_8bpc_c 53 52.2 1.53%
inv_txfm_add_4x8_adst_adst_0_8bpc_c 390.3 259.9 50.17%
inv_txfm_add_4x8_adst_adst_1_8bpc_c 390.3 352.2 10.82%
inv_txfm_add_4x8_adst_dct_0_8bpc_c 408 279.5 45.97%
inv_txfm_add_4x8_adst_dct_1_8bpc_c 407.9 369.7 10.33%
inv_txfm_add_4x8_adst_flipadst_0_8bpc_c 392.9 262.5 49.68%
inv_txfm_add_4x8_adst_flipadst_1_8bpc_c 392.9 353.6 11.11%
inv_txfm_add_4x8_adst_identity_0_8bpc_c 265.5 141.4 87.77%
inv_txfm_add_4x8_adst_identity_1_8bpc_c 265.5 238.8 11.18%
inv_txfm_add_4x8_dct_adst_0_8bpc_c 409.1 261.1 56.68%
inv_txfm_add_4x8_dct_adst_1_8bpc_c 407 364.1 11.78%
inv_txfm_add_4x8_dct_dct_0_8bpc_c 64.4 58.8 9.52%
inv_txfm_add_4x8_dct_dct_1_8bpc_c 428.1 381.4 12.24%
inv_txfm_add_4x8_dct_flipadst_0_8bpc_c 411 263.9 55.74%
inv_txfm_add_4x8_dct_flipadst_1_8bpc_c 410.9 367.4 11.84%
inv_txfm_add_4x8_dct_identity_0_8bpc_c 284.5 141.7 100.78%
inv_txfm_add_4x8_dct_identity_1_8bpc_c 285.1 254.6 11.98%
inv_txfm_add_4x8_flipadst_adst_0_8bpc_c 394 260.3 51.36%
inv_txfm_add_4x8_flipadst_adst_1_8bpc_c 394.2 351.3 12.21%
inv_txfm_add_4x8_flipadst_dct_0_8bpc_c 412 280.4 46.93%
inv_txfm_add_4x8_flipadst_dct_1_8bpc_c 412.2 370.2 11.35%
inv_txfm_add_4x8_flipadst_flipadst_0_8bpc_c 396.4 263.1 50.67%
inv_txfm_add_4x8_flipadst_flipadst_1_8bpc_c 396.2 354.4 11.79%
inv_txfm_add_4x8_flipadst_identity_0_8bpc_c 268.6 141.8 89.42%
inv_txfm_add_4x8_flipadst_identity_1_8bpc_c 269.1 240 12.13%
inv_txfm_add_4x8_identity_adst_0_8bpc_c 350.5 256.3 36.75%
inv_txfm_add_4x8_identity_adst_1_8bpc_c 350 307.1 13.97%
inv_txfm_add_4x8_identity_dct_0_8bpc_c 366.1 272.9 34.15%
inv_txfm_add_4x8_identity_dct_1_8bpc_c 366.2 323.3 13.27%
inv_txfm_add_4x8_identity_flipadst_0_8bpc_c 352.9 259.4 36.04%
inv_txfm_add_4x8_identity_flipadst_1_8bpc_c 352.7 309.5 13.96%
inv_txfm_add_4x8_identity_identity_0_8bpc_c 222.7 136.4 63.27%
inv_txfm_add_4x8_identity_identity_1_8bpc_c 223.5 196.1 13.97%
inv_txfm_add_4x16_adst_adst_0_8bpc_c 820.7 560.4 46.45%
inv_txfm_add_4x16_adst_adst_1_8bpc_c 820.4 587.4 39.67%
inv_txfm_add_4x16_adst_adst_2_8bpc_c 820.6 680.9 20.52%
inv_txfm_add_4x16_adst_dct_0_8bpc_c 889.3 632.5 40.60%
inv_txfm_add_4x16_adst_dct_1_8bpc_c 889.2 667.3 33.25%
inv_txfm_add_4x16_adst_dct_2_8bpc_c 889.3 763.4 16.49%
inv_txfm_add_4x16_adst_flipadst_0_8bpc_c 819 563.6 45.32%
inv_txfm_add_4x16_adst_flipadst_1_8bpc_c 818.9 589.1 39.01%
inv_txfm_add_4x16_adst_flipadst_2_8bpc_c 815.9 682.7 19.51%
inv_txfm_add_4x16_adst_identity_0_8bpc_c 492.7 241.1 104.36%
inv_txfm_add_4x16_adst_identity_1_8bpc_c 492.9 293.7 67.82%
inv_txfm_add_4x16_adst_identity_2_8bpc_c 493 431.6 14.23%
inv_txfm_add_4x16_dct_adst_0_8bpc_c 856.3 556.8 53.79%
inv_txfm_add_4x16_dct_adst_1_8bpc_c 858.7 587.6 46.14%
inv_txfm_add_4x16_dct_adst_2_8bpc_c 861.9 694 24.19%
inv_txfm_add_4x16_dct_dct_0_8bpc_c 105.6 102.7 2.82%
inv_txfm_add_4x16_dct_dct_1_8bpc_c 918.6 663.1 38.53%
inv_txfm_add_4x16_dct_dct_2_8bpc_c 917.2 771 18.96%
inv_txfm_add_4x16_dct_flipadst_0_8bpc_c 858.9 562.7 52.64%
inv_txfm_add_4x16_dct_flipadst_1_8bpc_c 859.8 592.8 45.04%
inv_txfm_add_4x16_dct_flipadst_2_8bpc_c 858.8 698.9 22.88%
inv_txfm_add_4x16_dct_identity_0_8bpc_c 538.8 243.3 121.45%
inv_txfm_add_4x16_dct_identity_1_8bpc_c 539.3 302.3 78.40%
inv_txfm_add_4x16_dct_identity_2_8bpc_c 538.3 460.2 16.97%
inv_txfm_add_4x16_flipadst_adst_0_8bpc_c 819.3 558.6 46.67%
inv_txfm_add_4x16_flipadst_adst_1_8bpc_c 819.2 586 39.80%
inv_txfm_add_4x16_flipadst_adst_2_8bpc_c 819.2 681.9 20.13%
inv_txfm_add_4x16_flipadst_dct_0_8bpc_c 887.6 632.6 40.31%
inv_txfm_add_4x16_flipadst_dct_1_8bpc_c 888.1 665.3 33.49%
inv_txfm_add_4x16_flipadst_dct_2_8bpc_c 887 763.6 16.16%
inv_txfm_add_4x16_flipadst_flipadst_0_8bpc_c 816.9 563.7 44.92%
inv_txfm_add_4x16_flipadst_flipadst_1_8bpc_c 817.1 590.8 38.30%
inv_txfm_add_4x16_flipadst_flipadst_2_8bpc_c 817.1 685.9 19.13%
inv_txfm_add_4x16_flipadst_identity_0_8bpc_c 493.3 241.8 104.01%
inv_txfm_add_4x16_flipadst_identity_1_8bpc_c 493.1 295.2 67.04%
inv_txfm_add_4x16_flipadst_identity_2_8bpc_c 493.3 435.8 13.19%
inv_txfm_add_4x16_identity_adst_0_8bpc_c 731 551.3 32.60%
inv_txfm_add_4x16_identity_adst_1_8bpc_c 731 559.2 30.72%
inv_txfm_add_4x16_identity_adst_2_8bpc_c 731.1 658 11.11%
inv_txfm_add_4x16_identity_dct_0_8bpc_c 792.1 620.2 27.72%
inv_txfm_add_4x16_identity_dct_1_8bpc_c 790.5 633.8 24.72%
inv_txfm_add_4x16_identity_dct_2_8bpc_c 791.9 733.7 7.93%
inv_txfm_add_4x16_identity_flipadst_0_8bpc_c 731.8 551.8 32.62%
inv_txfm_add_4x16_identity_flipadst_1_8bpc_c 731.7 563.2 29.92%
inv_txfm_add_4x16_identity_flipadst_2_8bpc_c 731.8 660.7 10.76%
inv_txfm_add_4x16_identity_identity_0_8bpc_c 408.9 236.2 73.12%
inv_txfm_add_4x16_identity_identity_1_8bpc_c 408.9 253.2 61.49%
inv_txfm_add_4x16_identity_identity_2_8bpc_c 409.3 308.6 32.63%
inv_txfm_add_8x4_adst_adst_0_8bpc_c 327.5 196.6 66.58%
inv_txfm_add_8x4_adst_adst_1_8bpc_c 327.5 330.8 -1.00%
inv_txfm_add_8x4_adst_dct_0_8bpc_c 340.6 205.5 65.74%
inv_txfm_add_8x4_adst_dct_1_8bpc_c 340.4 339.8 0.18%
inv_txfm_add_8x4_adst_flipadst_0_8bpc_c 330.8 198.9 66.31%
inv_txfm_add_8x4_adst_flipadst_1_8bpc_c 330.8 331.5 -0.21%
inv_txfm_add_8x4_adst_identity_0_8bpc_c 294.3 160.2 83.71%
inv_txfm_add_8x4_adst_identity_1_8bpc_c 293.9 293.4 0.17%
inv_txfm_add_8x4_dct_adst_0_8bpc_c 342.3 201.8 69.62%
inv_txfm_add_8x4_dct_adst_1_8bpc_c 342.9 346.5 -1.04%
inv_txfm_add_8x4_dct_dct_0_8bpc_c 28.2 22.1 27.60%
inv_txfm_add_8x4_dct_dct_1_8bpc_c 354.3 356.8 -0.70%
inv_txfm_add_8x4_dct_flipadst_0_8bpc_c 345.2 204.1 69.13%
inv_txfm_add_8x4_dct_flipadst_1_8bpc_c 345.9 350.2 -1.23%
inv_txfm_add_8x4_dct_identity_0_8bpc_c 308.8 166.1 85.91%
inv_txfm_add_8x4_dct_identity_1_8bpc_c 309.2 307.8 0.45%
inv_txfm_add_8x4_flipadst_adst_0_8bpc_c 328.1 197.5 66.13%
inv_txfm_add_8x4_flipadst_adst_1_8bpc_c 328.3 331.4 -0.94%
inv_txfm_add_8x4_flipadst_dct_0_8bpc_c 339.2 214.4 58.21%
inv_txfm_add_8x4_flipadst_dct_1_8bpc_c 340.4 342.9 -0.73%
inv_txfm_add_8x4_flipadst_flipadst_0_8bpc_c 331.3 199.6 65.98%
inv_txfm_add_8x4_flipadst_flipadst_1_8bpc_c 331.2 333.6 -0.72%
inv_txfm_add_8x4_flipadst_identity_0_8bpc_c 294 160.6 83.06%
inv_txfm_add_8x4_flipadst_identity_1_8bpc_c 294 294.3 -0.10%
inv_txfm_add_8x4_identity_adst_0_8bpc_c 205 169.8 20.73%
inv_txfm_add_8x4_identity_adst_1_8bpc_c 204.9 197.4 3.80%
inv_txfm_add_8x4_identity_dct_0_8bpc_c 216.7 178.2 21.60%
inv_txfm_add_8x4_identity_dct_1_8bpc_c 216.7 208.7 3.83%
inv_txfm_add_8x4_identity_flipadst_0_8bpc_c 208 171.4 21.35%
inv_txfm_add_8x4_identity_flipadst_1_8bpc_c 207.2 201.2 2.98%
inv_txfm_add_8x4_identity_identity_0_8bpc_c 171.4 131.5 30.34%
inv_txfm_add_8x4_identity_identity_1_8bpc_c 171.2 170.9 0.18%
inv_txfm_add_8x8_adst_adst_0_8bpc_c 758.4 463.8 63.52%
inv_txfm_add_8x8_adst_adst_1_8bpc_c 758 725.5 4.48%
inv_txfm_add_8x8_adst_dct_0_8bpc_c 799.4 501.8 59.31%
inv_txfm_add_8x8_adst_dct_1_8bpc_c 796.1 763.3 4.30%
inv_txfm_add_8x8_adst_flipadst_0_8bpc_c 761.7 470.7 61.82%
inv_txfm_add_8x8_adst_flipadst_1_8bpc_c 761 729 4.39%
inv_txfm_add_8x8_adst_identity_0_8bpc_c 507 218.2 132.36%
inv_txfm_add_8x8_adst_identity_1_8bpc_c 507.6 518.4 -2.08%
inv_txfm_add_8x8_dct_adst_0_8bpc_c 802.7 467.3 71.77%
inv_txfm_add_8x8_dct_adst_1_8bpc_c 799.8 754.2 6.05%
inv_txfm_add_8x8_dct_dct_0_8bpc_c 35.9 29.6 21.28%
inv_txfm_add_8x8_dct_dct_1_8bpc_c 828.7 785.2 5.54%
inv_txfm_add_8x8_dct_flipadst_0_8bpc_c 805 472.2 70.48%
inv_txfm_add_8x8_dct_flipadst_1_8bpc_c 806.7 759.6 6.20%
inv_txfm_add_8x8_dct_identity_0_8bpc_c 543 218.5 148.51%
inv_txfm_add_8x8_dct_identity_1_8bpc_c 542.9 549.5 -1.20%
inv_txfm_add_8x8_flipadst_adst_0_8bpc_c 760.7 462.3 64.55%
inv_txfm_add_8x8_flipadst_adst_1_8bpc_c 761.4 728.4 4.53%
inv_txfm_add_8x8_flipadst_dct_0_8bpc_c 805.4 500.8 60.82%
inv_txfm_add_8x8_flipadst_dct_1_8bpc_c 814.6 765.8 6.37%
inv_txfm_add_8x8_flipadst_flipadst_0_8bpc_c 762.9 469.7 62.42%
inv_txfm_add_8x8_flipadst_flipadst_1_8bpc_c 763.3 733.4 4.08%
inv_txfm_add_8x8_flipadst_identity_0_8bpc_c 506.6 216.2 134.32%
inv_txfm_add_8x8_flipadst_identity_1_8bpc_c 519.1 521.3 -0.42%
inv_txfm_add_8x8_identity_adst_0_8bpc_c 514.3 427.9 20.19%
inv_txfm_add_8x8_identity_adst_1_8bpc_c 513.6 462.3 11.10%
inv_txfm_add_8x8_identity_dct_0_8bpc_c 539.2 465.9 15.73%
inv_txfm_add_8x8_identity_dct_1_8bpc_c 539.9 497.1 8.61%
inv_txfm_add_8x8_identity_flipadst_0_8bpc_c 511.5 436.9 17.07%
inv_txfm_add_8x8_identity_flipadst_1_8bpc_c 511.5 467.7 9.36%
inv_txfm_add_8x8_identity_identity_0_8bpc_c 255.8 190.5 34.28%
inv_txfm_add_8x8_identity_identity_1_8bpc_c 256 257.2 -0.47%
inv_txfm_add_8x16_adst_adst_0_8bpc_c 1737.4 1015.7 71.05%
inv_txfm_add_8x16_adst_adst_1_8bpc_c 1737.1 1213 43.21%
inv_txfm_add_8x16_adst_adst_2_8bpc_c 1736.2 1583.7 9.63%
inv_txfm_add_8x16_adst_dct_0_8bpc_c 1874.2 1145.1 63.67%
inv_txfm_add_8x16_adst_dct_1_8bpc_c 1873.7 1362 37.57%
inv_txfm_add_8x16_adst_dct_2_8bpc_c 1875 1738.3 7.86%
inv_txfm_add_8x16_adst_flipadst_0_8bpc_c 1737.5 1025.8 69.38%
inv_txfm_add_8x16_adst_flipadst_1_8bpc_c 1739.4 1223 42.22%
inv_txfm_add_8x16_adst_flipadst_2_8bpc_c 1735.8 1593.1 8.96%
inv_txfm_add_8x16_adst_identity_0_8bpc_c 1098.9 389.4 182.20%
inv_txfm_add_8x16_adst_identity_1_8bpc_c 1098.8 572.5 91.93%
inv_txfm_add_8x16_adst_identity_2_8bpc_c 1099 1095.8 0.29%
inv_txfm_add_8x16_dct_adst_0_8bpc_c 1795.4 1020.7 75.90%
inv_txfm_add_8x16_dct_adst_1_8bpc_c 1795.3 1228.8 46.10%
inv_txfm_add_8x16_dct_adst_2_8bpc_c 1795.2 1657.5 8.31%
inv_txfm_add_8x16_dct_dct_0_8bpc_c 54.3 46.2 17.53%
inv_txfm_add_8x16_dct_dct_1_8bpc_c 1912.2 1345.8 42.09%
inv_txfm_add_8x16_dct_dct_2_8bpc_c 1912.4 1777.7 7.58%
inv_txfm_add_8x16_dct_flipadst_0_8bpc_c 1790.7 1029.5 73.94%
inv_txfm_add_8x16_dct_flipadst_1_8bpc_c 1790.6 1239.7 44.44%
inv_txfm_add_8x16_dct_flipadst_2_8bpc_c 1792.6 1670.4 7.32%
inv_txfm_add_8x16_dct_identity_0_8bpc_c 1155.1 391.1 195.35%
inv_txfm_add_8x16_dct_identity_1_8bpc_c 1155.1 590.2 95.71%
inv_txfm_add_8x16_dct_identity_2_8bpc_c 1155.8 1157.3 -0.13%
inv_txfm_add_8x16_flipadst_adst_0_8bpc_c 1738.6 1010.9 71.99%
inv_txfm_add_8x16_flipadst_adst_1_8bpc_c 1738.8 1217 42.88%
inv_txfm_add_8x16_flipadst_adst_2_8bpc_c 1739.1 1591.8 9.25%
inv_txfm_add_8x16_flipadst_dct_0_8bpc_c 1876.5 1162.6 61.41%
inv_txfm_add_8x16_flipadst_dct_1_8bpc_c 1875.8 1372.7 36.65%
inv_txfm_add_8x16_flipadst_dct_2_8bpc_c 1876 1743.6 7.59%
inv_txfm_add_8x16_flipadst_flipadst_0_8bpc_c 1736.1 1021.8 69.91%
inv_txfm_add_8x16_flipadst_flipadst_1_8bpc_c 1736.7 1223.9 41.90%
inv_txfm_add_8x16_flipadst_flipadst_2_8bpc_c 1736.5 1596.6 8.76%
inv_txfm_add_8x16_flipadst_identity_0_8bpc_c 1101 388.2 183.62%
inv_txfm_add_8x16_flipadst_identity_1_8bpc_c 1101.5 574.6 91.70%
inv_txfm_add_8x16_flipadst_identity_2_8bpc_c 1102.7 1102.2 0.05%
inv_txfm_add_8x16_identity_adst_0_8bpc_c 1222.5 977.2 25.10%
inv_txfm_add_8x16_identity_adst_1_8bpc_c 1223.4 1056.9 15.75%
inv_txfm_add_8x16_identity_adst_2_8bpc_c 1221.7 1198.8 1.91%
inv_txfm_add_8x16_identity_dct_0_8bpc_c 1363 1127.1 20.93%
inv_txfm_add_8x16_identity_dct_1_8bpc_c 1361.3 1199.5 13.49%
inv_txfm_add_8x16_identity_dct_2_8bpc_c 1361 1340.3 1.54%
inv_txfm_add_8x16_identity_flipadst_0_8bpc_c 1234.1 987.9 24.92%
inv_txfm_add_8x16_identity_flipadst_1_8bpc_c 1234 1067.1 15.64%
inv_txfm_add_8x16_identity_flipadst_2_8bpc_c 1233.8 1210.3 1.94%
inv_txfm_add_8x16_identity_identity_0_8bpc_c 600 358.4 67.41%
inv_txfm_add_8x16_identity_identity_1_8bpc_c 600.6 414.7 44.83%
inv_txfm_add_8x16_identity_identity_2_8bpc_c 601.7 530.5 13.42%
inv_txfm_add_8x32_dct_dct_0_8bpc_c 89.2 81.4 9.58%
inv_txfm_add_8x32_dct_dct_1_8bpc_c 4296.8 2746.4 56.45%
inv_txfm_add_8x32_dct_dct_2_8bpc_c 4298.1 3280 31.04%
inv_txfm_add_8x32_dct_dct_3_8bpc_c 4296.6 3690.3 16.43%
inv_txfm_add_8x32_dct_dct_4_8bpc_c 4295.1 4185.9 2.61%
inv_txfm_add_8x32_identity_identity_0_8bpc_c 911.2 556.8 63.65%
inv_txfm_add_8x32_identity_identity_1_8bpc_c 910.8 567.6 60.47%
inv_txfm_add_8x32_identity_identity_2_8bpc_c 911 681.8 33.62%
inv_txfm_add_8x32_identity_identity_3_8bpc_c 911.1 765.5 19.02%
inv_txfm_add_8x32_identity_identity_4_8bpc_c 911.2 928.5 -1.86%
inv_txfm_add_16x4_adst_adst_0_8bpc_c 658.9 358.5 83.79%
inv_txfm_add_16x4_adst_adst_1_8bpc_c 658.9 668.2 -1.39%
inv_txfm_add_16x4_adst_adst_2_8bpc_c 658.9 667.4 -1.27%
inv_txfm_add_16x4_adst_dct_0_8bpc_c 683.8 380.7 79.62%
inv_txfm_add_16x4_adst_dct_1_8bpc_c 685.7 693.1 -1.07%
inv_txfm_add_16x4_adst_dct_2_8bpc_c 685.6 693.3 -1.11%
inv_txfm_add_16x4_adst_flipadst_0_8bpc_c 670.9 363 84.82%
inv_txfm_add_16x4_adst_flipadst_1_8bpc_c 672.9 674 -0.16%
inv_txfm_add_16x4_adst_flipadst_2_8bpc_c 672.9 673 -0.01%
inv_txfm_add_16x4_adst_identity_0_8bpc_c 592.5 288.7 105.23%
inv_txfm_add_16x4_adst_identity_1_8bpc_c 590.9 595.1 -0.71%
inv_txfm_add_16x4_adst_identity_2_8bpc_c 591.4 595.3 -0.66%
inv_txfm_add_16x4_dct_adst_0_8bpc_c 749.5 384 95.18%
inv_txfm_add_16x4_dct_adst_1_8bpc_c 748.8 759 -1.34%
inv_txfm_add_16x4_dct_adst_2_8bpc_c 749.1 758.6 -1.25%
inv_txfm_add_16x4_dct_dct_0_8bpc_c 36.1 27.3 32.23%
inv_txfm_add_16x4_dct_dct_1_8bpc_c 761.2 765.3 -0.54%
inv_txfm_add_16x4_dct_dct_2_8bpc_c 761.1 770.5 -1.22%
inv_txfm_add_16x4_dct_flipadst_0_8bpc_c 752.8 387.5 94.27%
inv_txfm_add_16x4_dct_flipadst_1_8bpc_c 753.2 758.7 -0.72%
inv_txfm_add_16x4_dct_flipadst_2_8bpc_c 753.1 758.6 -0.73%
inv_txfm_add_16x4_dct_identity_0_8bpc_c 671.2 311 115.82%
inv_txfm_add_16x4_dct_identity_1_8bpc_c 671.1 678.4 -1.08%
inv_txfm_add_16x4_dct_identity_2_8bpc_c 670.9 678.7 -1.15%
inv_txfm_add_16x4_flipadst_adst_0_8bpc_c 666 359.3 85.36%
inv_txfm_add_16x4_flipadst_adst_1_8bpc_c 665.2 669.5 -0.64%
inv_txfm_add_16x4_flipadst_adst_2_8bpc_c 666 668.3 -0.34%
inv_txfm_add_16x4_flipadst_dct_0_8bpc_c 686.4 381.3 80.02%
inv_txfm_add_16x4_flipadst_dct_1_8bpc_c 686.5 690.9 -0.64%
inv_txfm_add_16x4_flipadst_dct_2_8bpc_c 686.4 691.2 -0.69%
inv_txfm_add_16x4_flipadst_flipadst_0_8bpc_c 675 362.1 86.41%
inv_txfm_add_16x4_flipadst_flipadst_1_8bpc_c 675.4 673.8 0.24%
inv_txfm_add_16x4_flipadst_flipadst_2_8bpc_c 673.5 673.7 -0.03%
inv_txfm_add_16x4_flipadst_identity_0_8bpc_c 594.9 288.5 106.20%
inv_txfm_add_16x4_flipadst_identity_1_8bpc_c 594.2 598.9 -0.78%
inv_txfm_add_16x4_flipadst_identity_2_8bpc_c 595.1 598.4 -0.55%
inv_txfm_add_16x4_identity_adst_0_8bpc_c 342.5 278.8 22.85%
inv_txfm_add_16x4_identity_adst_1_8bpc_c 342.5 278.1 23.16%
inv_txfm_add_16x4_identity_adst_2_8bpc_c 342 330 3.64%
inv_txfm_add_16x4_identity_dct_0_8bpc_c 366.9 301.1 21.85%
inv_txfm_add_16x4_identity_dct_1_8bpc_c 366.7 300.1 22.19%
inv_txfm_add_16x4_identity_dct_2_8bpc_c 366.9 351.4 4.41%
inv_txfm_add_16x4_identity_flipadst_0_8bpc_c 347.2 282.6 22.86%
inv_txfm_add_16x4_identity_flipadst_1_8bpc_c 347.4 282.6 22.93%
inv_txfm_add_16x4_identity_flipadst_2_8bpc_c 347.4 334.2 3.95%
inv_txfm_add_16x4_identity_identity_0_8bpc_c 275.3 208.8 31.85%
inv_txfm_add_16x4_identity_identity_1_8bpc_c 275.1 281.1 -2.13%
inv_txfm_add_16x4_identity_identity_2_8bpc_c 275.4 282.2 -2.41%
inv_txfm_add_16x8_adst_adst_0_8bpc_c 1692.3 899 88.24%
inv_txfm_add_16x8_adst_adst_1_8bpc_c 1690.7 1462.8 15.58%
inv_txfm_add_16x8_adst_adst_2_8bpc_c 1689.8 1679.1 0.64%
inv_txfm_add_16x8_adst_dct_0_8bpc_c 1766 985.3 79.23%
inv_txfm_add_16x8_adst_dct_1_8bpc_c 1765.7 1545.4 14.26%
inv_txfm_add_16x8_adst_dct_2_8bpc_c 1765.8 1757.6 0.47%
inv_txfm_add_16x8_adst_flipadst_0_8bpc_c 1687.4 908.5 85.73%
inv_txfm_add_16x8_adst_flipadst_1_8bpc_c 1688.3 1479.3 14.13%
inv_txfm_add_16x8_adst_flipadst_2_8bpc_c 1690.3 1695.5 -0.31%
inv_txfm_add_16x8_adst_identity_0_8bpc_c 1148.3 402.4 185.36%
inv_txfm_add_16x8_adst_identity_1_8bpc_c 1149.3 1173.6 -2.07%
inv_txfm_add_16x8_adst_identity_2_8bpc_c 1147.6 1172.5 -2.12%
inv_txfm_add_16x8_dct_adst_0_8bpc_c 1872.2 918.9 103.74%
inv_txfm_add_16x8_dct_adst_1_8bpc_c 1871.3 1591.7 17.57%
inv_txfm_add_16x8_dct_adst_2_8bpc_c 1872 1864.4 0.41%
inv_txfm_add_16x8_dct_dct_0_8bpc_c 51.4 40.2 27.86%
inv_txfm_add_16x8_dct_dct_1_8bpc_c 1930.9 1636.4 18.00%
inv_txfm_add_16x8_dct_dct_2_8bpc_c 1930.7 1894.7 1.90%
inv_txfm_add_16x8_dct_flipadst_0_8bpc_c 1868 924.1 102.14%
inv_txfm_add_16x8_dct_flipadst_1_8bpc_c 1872.3 1598 17.17%
inv_txfm_add_16x8_dct_flipadst_2_8bpc_c 1872.3 1872.6 -0.02%
inv_txfm_add_16x8_dct_identity_0_8bpc_c 1341.9 424.6 216.04%
inv_txfm_add_16x8_dct_identity_1_8bpc_c 1342.1 1333.2 0.67%
inv_txfm_add_16x8_dct_identity_2_8bpc_c 1346.6 1329.9 1.26%
inv_txfm_add_16x8_flipadst_adst_0_8bpc_c 1691 899 88.10%
inv_txfm_add_16x8_flipadst_adst_1_8bpc_c 1698 1459.1 16.37%
inv_txfm_add_16x8_flipadst_adst_2_8bpc_c 1698.1 1674.4 1.42%
inv_txfm_add_16x8_flipadst_dct_0_8bpc_c 1761.6 985 78.84%
inv_txfm_add_16x8_flipadst_dct_1_8bpc_c 1760.7 1542.8 14.12%
inv_txfm_add_16x8_flipadst_dct_2_8bpc_c 1757.1 1754.5 0.15%
inv_txfm_add_16x8_flipadst_flipadst_0_8bpc_c 1698.1 906.3 87.37%
inv_txfm_add_16x8_flipadst_flipadst_1_8bpc_c 1698 1468.6 15.62%
inv_txfm_add_16x8_flipadst_flipadst_2_8bpc_c 1698.2 1679.1 1.14%
inv_txfm_add_16x8_flipadst_identity_0_8bpc_c 1168.4 408.5 186.02%
inv_txfm_add_16x8_flipadst_identity_1_8bpc_c 1169.1 1173.6 -0.38%
inv_txfm_add_16x8_flipadst_identity_2_8bpc_c 1168.5 1175.4 -0.59%
inv_txfm_add_16x8_identity_adst_0_8bpc_c 1032.4 812.2 27.11%
inv_txfm_add_16x8_identity_adst_1_8bpc_c 1032.5 811.9 27.17%
inv_txfm_add_16x8_identity_adst_2_8bpc_c 1032.7 938.6 10.03%
inv_txfm_add_16x8_identity_dct_0_8bpc_c 1118 901.9 23.96%
inv_txfm_add_16x8_identity_dct_1_8bpc_c 1117.7 904.7 23.54%
inv_txfm_add_16x8_identity_dct_2_8bpc_c 1118.6 1030.9 8.51%
inv_txfm_add_16x8_identity_flipadst_0_8bpc_c 1041.1 825.1 26.18%
inv_txfm_add_16x8_identity_flipadst_1_8bpc_c 1041.1 825.1 26.18%
inv_txfm_add_16x8_identity_flipadst_2_8bpc_c 1040.4 953.2 9.15%
inv_txfm_add_16x8_identity_identity_0_8bpc_c 523.6 324.7 61.26%
inv_txfm_add_16x8_identity_identity_1_8bpc_c 523.7 480.4 9.01%
inv_txfm_add_16x8_identity_identity_2_8bpc_c 522.7 539.5 -3.11%
inv_txfm_add_16x16_adst_adst_0_8bpc_c 3581.9 1916.6 86.89%
inv_txfm_add_16x16_adst_adst_1_8bpc_c 3594.7 2321.2 54.86%
inv_txfm_add_16x16_adst_adst_2_8bpc_c 3596.4 3576.6 0.55%
inv_txfm_add_16x16_adst_dct_0_8bpc_c 3915.6 2240.8 74.74%
inv_txfm_add_16x16_adst_dct_1_8bpc_c 3913.8 2705.8 44.64%
inv_txfm_add_16x16_adst_dct_2_8bpc_c 3913.8 4014.7 -2.51%
inv_txfm_add_16x16_adst_flipadst_0_8bpc_c 3593.5 1935.8 85.63%
inv_txfm_add_16x16_adst_flipadst_1_8bpc_c 3598.6 2344.5 53.49%
inv_txfm_add_16x16_adst_flipadst_2_8bpc_c 3596.5 3589.4 0.20%
inv_txfm_add_16x16_dct_adst_0_8bpc_c 3981 1967.6 102.33%
inv_txfm_add_16x16_dct_adst_1_8bpc_c 3978.4 2516.4 58.10%
inv_txfm_add_16x16_dct_adst_2_8bpc_c 3978.7 4037 -1.44%
inv_txfm_add_16x16_dct_dct_0_8bpc_c 82 70.6 16.15%
inv_txfm_add_16x16_dct_dct_1_8bpc_c 4159.7 2763.6 50.52%
inv_txfm_add_16x16_dct_dct_2_8bpc_c 4165.5 4201.5 -0.86%
inv_txfm_add_16x16_dct_flipadst_0_8bpc_c 3962.8 1973.2 100.83%
inv_txfm_add_16x16_dct_flipadst_1_8bpc_c 3962.9 2529.9 56.64%
inv_txfm_add_16x16_dct_flipadst_2_8bpc_c 3962.2 4038.3 -1.88%
inv_txfm_add_16x16_dct_identity_0_8bpc_c 2569.9 736.4 248.98%
inv_txfm_add_16x16_dct_identity_1_8bpc_c 2571 1126.4 128.25%
inv_txfm_add_16x16_dct_identity_2_8bpc_c 2569.4 2633.2 -2.42%
inv_txfm_add_16x16_flipadst_adst_0_8bpc_c 3609.9 1919.9 88.03%
inv_txfm_add_16x16_flipadst_adst_1_8bpc_c 3594.8 2315.6 55.24%
inv_txfm_add_16x16_flipadst_adst_2_8bpc_c 3600.3 3566.5 0.95%
inv_txfm_add_16x16_flipadst_dct_0_8bpc_c 3914 2242.2 74.56%
inv_txfm_add_16x16_flipadst_dct_1_8bpc_c 3916.6 2669.1 46.74%
inv_txfm_add_16x16_flipadst_dct_2_8bpc_c 3913.3 3953.2 -1.01%
inv_txfm_add_16x16_flipadst_flipadst_0_8bpc_c 3611.5 1932.3 86.90%
inv_txfm_add_16x16_flipadst_flipadst_1_8bpc_c 3611.9 2343 54.16%
inv_txfm_add_16x16_flipadst_flipadst_2_8bpc_c 3608.3 3590.9 0.48%
inv_txfm_add_16x16_identity_dct_0_8bpc_c 2560.2 2167 18.14%
inv_txfm_add_16x16_identity_dct_1_8bpc_c 2559.7 2161.8 18.41%
inv_txfm_add_16x16_identity_dct_2_8bpc_c 2560.2 2558.5 0.07%
inv_txfm_add_16x16_identity_identity_0_8bpc_c 1014.1 629.6 61.07%
inv_txfm_add_16x16_identity_identity_1_8bpc_c 1014.5 734.7 38.08%
inv_txfm_add_16x16_identity_identity_2_8bpc_c 1015.3 1019.5 -0.41%
inv_txfm_add_16x32_dct_dct_0_8bpc_c 158.1 131.9 19.86%
inv_txfm_add_16x32_dct_dct_1_8bpc_c 9326.4 6015.3 55.04%
inv_txfm_add_16x32_dct_dct_2_8bpc_c 9332.1 6900.9 35.23%
inv_txfm_add_16x32_dct_dct_3_8bpc_c 9327 7774.1 19.98%
inv_txfm_add_16x32_dct_dct_4_8bpc_c 9328.5 9153 1.92%
inv_txfm_add_16x32_identity_identity_0_8bpc_c 2013.7 1036.2 94.34%
inv_txfm_add_16x32_identity_identity_1_8bpc_c 2012.9 1188.7 69.34%
inv_txfm_add_16x32_identity_identity_2_8bpc_c 2015.7 1441.3 39.85%
inv_txfm_add_16x32_identity_identity_3_8bpc_c 2010.6 1650.2 21.84%
inv_txfm_add_16x32_identity_identity_4_8bpc_c 2012.3 2005.7 0.33%
inv_txfm_add_16x64_dct_dct_0_8bpc_c 281.5 266.4 5.67%
inv_txfm_add_16x64_dct_dct_1_8bpc_c 14921.1 11916.5 25.21%
inv_txfm_add_16x64_dct_dct_2_8bpc_c 14918.7 12551.8 18.86%
inv_txfm_add_16x64_dct_dct_3_8bpc_c 14918.5 14058.1 6.12%
inv_txfm_add_16x64_dct_dct_4_8bpc_c 14929.1 15037 -0.72%
inv_txfm_add_32x8_dct_dct_0_8bpc_c 73.4 69.4 5.76%
inv_txfm_add_32x8_dct_dct_1_8bpc_c 4164.9 2567.5 62.22%
inv_txfm_add_32x8_dct_dct_2_8bpc_c 4185.9 4212.2 -0.62%
inv_txfm_add_32x8_dct_dct_3_8bpc_c 4184.5 4213.5 -0.69%
inv_txfm_add_32x8_dct_dct_4_8bpc_c 4187.6 4210.8 -0.55%
inv_txfm_add_32x8_identity_identity_0_8bpc_c 770.6 552.2 39.55%
inv_txfm_add_32x8_identity_identity_1_8bpc_c 770.2 634 21.48%
inv_txfm_add_32x8_identity_identity_2_8bpc_c 770.4 824.1 -6.52%
inv_txfm_add_32x8_identity_identity_3_8bpc_c 770.2 826.9 -6.86%
inv_txfm_add_32x8_identity_identity_4_8bpc_c 771 824.9 -6.53%
inv_txfm_add_32x16_dct_dct_0_8bpc_c 134.9 130.9 3.06%
inv_txfm_add_32x16_dct_dct_1_8bpc_c 9271.2 6441 43.94%
inv_txfm_add_32x16_dct_dct_2_8bpc_c 9279 8631.3 7.50%
inv_txfm_add_32x16_dct_dct_3_8bpc_c 9271.5 9274.9 -0.04%
inv_txfm_add_32x16_dct_dct_4_8bpc_c 9273.2 9266.3 0.07%
inv_txfm_add_32x16_identity_identity_0_8bpc_c 1934.8 1190.8 62.48%
inv_txfm_add_32x16_identity_identity_1_8bpc_c 1934.4 1504.9 28.54%
inv_txfm_add_32x16_identity_identity_2_8bpc_c 1935.4 1840.6 5.15%
inv_txfm_add_32x16_identity_identity_3_8bpc_c 1935.1 1934.7 0.02%
inv_txfm_add_32x16_identity_identity_4_8bpc_c 1935.9 1934.5 0.07%
inv_txfm_add_32x32_dct_dct_0_8bpc_c 256 253.6 0.95%
inv_txfm_add_32x32_dct_dct_1_8bpc_c 19959.7 12449.5 60.33%
inv_txfm_add_32x32_dct_dct_2_8bpc_c 19916.4 13673 45.66%
inv_txfm_add_32x32_dct_dct_3_8bpc_c 19944 16689 19.50%
inv_txfm_add_32x32_dct_dct_4_8bpc_c 19955.6 19739.5 1.09%
inv_txfm_add_32x32_identity_identity_0_8bpc_c 3047.1 1918.1 58.86%
inv_txfm_add_32x32_identity_identity_1_8bpc_c 3035.9 2153.6 40.97%
inv_txfm_add_32x32_identity_identity_2_8bpc_c 3048 2323.5 31.18%
inv_txfm_add_32x32_identity_identity_3_8bpc_c 3047.5 2740 11.22%
inv_txfm_add_32x32_identity_identity_4_8bpc_c 3047 3257.2 -6.45%
inv_txfm_add_32x64_dct_dct_0_8bpc_c 517.8 512.1 1.11%
inv_txfm_add_32x64_dct_dct_1_8bpc_c 31726.6 23321.1 36.04%
inv_txfm_add_32x64_dct_dct_2_8bpc_c 31760.3 25779.4 23.20%
inv_txfm_add_32x64_dct_dct_3_8bpc_c 31705.3 28886.7 9.76%
inv_txfm_add_32x64_dct_dct_4_8bpc_c 31721 31711.9 0.03%
inv_txfm_add_64x16_dct_dct_0_8bpc_c 256 253.6 0.95%
inv_txfm_add_64x16_dct_dct_1_8bpc_c 18556.5 12950.2 43.29%
inv_txfm_add_64x16_dct_dct_2_8bpc_c 18560.2 16839.7 10.22%
inv_txfm_add_64x16_dct_dct_3_8bpc_c 18569 18711.8 -0.76%
inv_txfm_add_64x16_dct_dct_4_8bpc_c 18544.3 18751 -1.10%
inv_txfm_add_64x32_dct_dct_0_8bpc_c 507.6 498.3 1.87%
inv_txfm_add_64x32_dct_dct_1_8bpc_c 40934.2 23639.5 73.16%
inv_txfm_add_64x32_dct_dct_2_8bpc_c 40965.5 28842.7 42.03%
inv_txfm_add_64x32_dct_dct_3_8bpc_c 40989.8 35298 16.12%
inv_txfm_add_64x32_dct_dct_4_8bpc_c 40959 41171.7 -0.52%
inv_txfm_add_64x64_dct_dct_0_8bpc_c 1005.4 988.8 1.68%
inv_txfm_add_64x64_dct_dct_1_8bpc_c 62995.6 46838.1 34.50%
inv_txfm_add_64x64_dct_dct_2_8bpc_c 62992.9 52549.4 19.87%
inv_txfm_add_64x64_dct_dct_3_8bpc_c 62994.2 57695.8 9.18%
inv_txfm_add_64x64_dct_dct_4_8bpc_c 62970.4 63366.8 -0.63%
Average 32.02%
Edited by Ronald S. Bultje

Merge request reports