wandb: Using wandb-core as the SDK backend. Please refer to https://wandb.me/wandb-core for more information. wandb: Currently logged in as: kaiqiu. Use `wandb login --relogin` to force relogin wandb: Tracking run with wandb version 0.18.3 wandb: Run data is saved locally in /home/user/VAR/wandb/run-20241012_091010-cmdzz231 wandb: Run `wandb offline` to turn off syncing. wandb: Syncing run misty-star-188 wandb: View project at https://wandb.ai/kaiqiu/VAR wandb: View run at https://wandb.ai/kaiqiu/VAR/runs/cmdzz231 /home/user/VAR/tokenizer/tokenizer_image/dino_enc/dinov2.py:122: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. with torch.cuda.amp.autocast(enabled=False): AUTOTUNE addmm(6840x1088, 6840x96, 96x1088) triton_mm_13 0.0358 ms 100.0% triton_mm_16 0.0377 ms 95.0% triton_mm_9 0.0387 ms 92.5% triton_mm_6 0.0421 ms 85.0% triton_mm_5 0.0424 ms 84.4% triton_mm_10 0.0430 ms 83.1% triton_mm_14 0.0434 ms 82.5% bias_addmm 0.0473 ms 75.7% triton_mm_15 0.0503 ms 71.1% triton_mm_11 0.0559 ms 64.0% SingleProcess AUTOTUNE benchmarking takes 2.3196 seconds and 0.0207 seconds precompiling AUTOTUNE addmm(24x6528, 24x1088, 1088x6528) triton_mm_31 0.0216 ms 100.0% triton_mm_35 0.0224 ms 96.3% triton_mm_26 0.0226 ms 95.5% triton_mm_27 0.0229 ms 94.1% triton_mm_23 0.0231 ms 93.4% triton_mm_20 0.0233 ms 92.7% bias_addmm 0.0240 ms 90.1% triton_mm_30 0.0246 ms 87.8% triton_mm_21 0.0251 ms 86.1% triton_mm_22 0.0251 ms 86.1% SingleProcess AUTOTUNE benchmarking takes 2.1974 seconds and 0.0008 seconds precompiling AUTOTUNE addmm(6864x3264, 6864x1088, 1088x3264) triton_mm_52 0.2191 ms 100.0% triton_mm_53 0.2255 ms 97.2% triton_mm_54 0.2528 ms 86.7% triton_mm_47 0.2621 ms 83.6% triton_mm_49 0.2710 ms 80.9% triton_mm_45 0.2740 ms 80.0% bias_addmm 0.3009 ms 72.8% triton_mm_46 0.3046 ms 71.9% triton_mm_50 0.3088 ms 71.0% addmm 0.3235 ms 67.7% SingleProcess AUTOTUNE benchmarking takes 2.5599 seconds and 0.0078 seconds precompiling AUTOTUNE addmm(6864x1088, 6864x1088, 1088x1088) triton_mm_72 0.0900 ms 100.0% triton_mm_71 0.0903 ms 99.6% bias_addmm 0.0967 ms 93.0% triton_mm_68 0.0971 ms 92.6% triton_mm_66 0.0984 ms 91.4% triton_mm_73 0.0991 ms 90.8% triton_mm_64 0.1015 ms 88.6% addmm 0.1101 ms 81.7% triton_mm_65 0.1114 ms 80.7% triton_mm_69 0.1121 ms 80.3% SingleProcess AUTOTUNE benchmarking takes 2.5114 seconds and 0.0010 seconds precompiling AUTOTUNE addmm(6864x4352, 6864x1088, 1088x4352) triton_mm_90 0.2884 ms 100.0% triton_mm_91 0.2930 ms 98.4% bias_addmm 0.3250 ms 88.8% triton_mm_92 0.3307 ms 87.2% triton_mm_85 0.3374 ms 85.5% triton_mm_83 0.3566 ms 80.9% triton_mm_87 0.3597 ms 80.2% addmm 0.3617 ms 79.7% triton_mm_84 0.3972 ms 72.6% triton_mm_88 0.4065 ms 71.0% SingleProcess AUTOTUNE benchmarking takes 2.6752 seconds and 0.0065 seconds precompiling AUTOTUNE addmm(6864x1088, 6864x4352, 4352x1088) bias_addmm 0.2879 ms 100.0% triton_mm_110 0.3036 ms 94.8% triton_mm_109 0.3132 ms 91.9% triton_mm_111 0.3220 ms 89.4% addmm 0.3228 ms 89.2% triton_mm_106 0.3519 ms 81.8% triton_mm_104 0.3551 ms 81.1% triton_mm_102 0.3676 ms 78.3% triton_mm_105 0.3683 ms 78.2% triton_mm_107 0.4031 ms 71.4% SingleProcess AUTOTUNE benchmarking takes 2.6612 seconds and 0.0009 seconds precompiling AUTOTUNE addmm(6864x4352, 6864x1088, 1088x4352) triton_mm_183 0.2883 ms 100.0% triton_mm_184 0.2925 ms 98.6% bias_addmm 0.3273 ms 88.1% triton_mm_185 0.3310 ms 87.1% triton_mm_178 0.3370 ms 85.5% triton_mm_176 0.3565 ms 80.8% triton_mm_180 0.3592 ms 80.3% triton_mm_177 0.3954 ms 72.9% triton_mm_181 0.4057 ms 71.0% addmm 0.4161 ms 69.3% SingleProcess AUTOTUNE benchmarking takes 2.6494 seconds and 0.0000 seconds precompiling AUTOTUNE addmm(6864x1088, 6864x4352, 4352x1088) bias_addmm 0.2887 ms 100.0% triton_mm_203 0.3031 ms 95.2% triton_mm_202 0.3124 ms 92.4% triton_mm_204 0.3235 ms 89.2% addmm 0.3238 ms 89.2% triton_mm_199 0.3523 ms 81.9% triton_mm_197 0.3544 ms 81.5% triton_mm_195 0.3673 ms 78.6% triton_mm_198 0.3685 ms 78.3% triton_mm_200 0.4028 ms 71.7% SingleProcess AUTOTUNE benchmarking takes 2.6511 seconds and 0.0000 seconds precompiling AUTOTUNE addmm(6864x4352, 6864x1088, 1088x4352) triton_mm_276 0.2871 ms 100.0% triton_mm_277 0.2913 ms 98.6% bias_addmm 0.3262 ms 88.0% triton_mm_278 0.3290 ms 87.3% triton_mm_271 0.3356 ms 85.5% triton_mm_269 0.3547 ms 80.9% triton_mm_273 0.3595 ms 79.8% addmm 0.3601 ms 79.7% triton_mm_270 0.3933 ms 73.0% triton_mm_274 0.4074 ms 70.5% SingleProcess AUTOTUNE benchmarking takes 2.6451 seconds and 0.0000 seconds precompiling AUTOTUNE addmm(6864x1088, 6864x4352, 4352x1088) bias_addmm 0.2874 ms 100.0% triton_mm_296 0.3024 ms 95.0% triton_mm_295 0.3114 ms 92.3% triton_mm_297 0.3214 ms 89.4% addmm 0.3223 ms 89.2% triton_mm_292 0.3517 ms 81.7% triton_mm_290 0.3525 ms 81.5% triton_mm_291 0.3669 ms 78.3% triton_mm_288 0.3674 ms 78.2% triton_mm_293 0.4018 ms 71.5% SingleProcess AUTOTUNE benchmarking takes 2.6588 seconds and 0.0000 seconds precompiling AUTOTUNE mm(24x1088, 1088x2176) triton_mm_1608 0.0139 ms 100.0% triton_mm_1604 0.0145 ms 96.2% triton_mm_1616 0.0160 ms 87.2% triton_mm_1612 0.0162 ms 85.8% triton_mm_1603 0.0167 ms 83.2% mm 0.0180 ms 77.5% triton_mm_1607 0.0183 ms 76.2% triton_mm_1602 0.0184 ms 75.8% triton_mm_1601 0.0188 ms 74.0% triton_mm_1611 0.0201 ms 69.2% SingleProcess AUTOTUNE benchmarking takes 7.6826 seconds and 0.0000 seconds precompiling AUTOTUNE addmm(6864x12288, 6864x1088, 1088x12288) bias_addmm 0.7993 ms 100.0% triton_mm_1634 0.8164 ms 97.9% triton_mm_1633 0.8193 ms 97.6% triton_mm_1635 0.9268 ms 86.2% triton_mm_1628 0.9677 ms 82.6% triton_mm_1630 1.0119 ms 79.0% addmm 1.0191 ms 78.4% triton_mm_1626 1.0440 ms 76.6% triton_mm_1631 1.1459 ms 69.8% triton_mm_1627 1.1494 ms 69.5% SingleProcess AUTOTUNE benchmarking takes 11.5765 seconds and 0.0000 seconds precompiling AUTOTUNE mm(12288x6864, 6864x1088) triton_mm_1671 0.7462 ms 100.0% triton_mm_1672 0.7707 ms 96.8% triton_mm_1673 0.8072 ms 92.4% mm 0.8670 ms 86.1% triton_mm_1668 0.9241 ms 80.8% triton_mm_1666 0.9668 ms 77.2% triton_mm_1664 0.9768 ms 76.4% triton_mm_1667 1.0199 ms 73.2% triton_mm_1669 1.0364 ms 72.0% triton_mm_1665 1.0702 ms 69.7% SingleProcess AUTOTUNE benchmarking takes 9.9694 seconds and 0.0000 seconds precompiling AUTOTUNE mm(1088x6864, 6864x4352) mm 0.2844 ms 100.0% triton_mm_1743 0.2915 ms 97.6% triton_mm_1744 0.2922 ms 97.4% triton_mm_1745 0.2959 ms 96.1% triton_mm_1738 0.3440 ms 82.7% triton_mm_1736 0.3467 ms 82.0% triton_mm_1740 0.3577 ms 79.5% triton_mm_1739 0.3730 ms 76.3% triton_mm_1741 0.3894 ms 73.0% triton_mm_1737 0.3940 ms 72.2% SingleProcess AUTOTUNE benchmarking takes 9.9612 seconds and 0.0000 seconds precompiling AUTOTUNE mm(2176x24, 24x1088) triton_mm_1700 0.0111 ms 100.0% triton_mm_1702 0.0111 ms 100.0% triton_mm_1696 0.0112 ms 99.7% triton_mm_1697 0.0114 ms 97.5% triton_mm_1703 0.0116 ms 95.7% triton_mm_1704 0.0117 ms 95.3% triton_mm_1706 0.0119 ms 93.3% triton_mm_1691 0.0121 ms 92.1% triton_mm_1695 0.0123 ms 90.6% triton_mm_1699 0.0124 ms 89.7% SingleProcess AUTOTUNE benchmarking takes 6.0154 seconds and 0.0000 seconds precompiling AUTOTUNE mm(24x2176, 2176x1088) triton_mm_1678 0.0155 ms 100.0% mm 0.0172 ms 90.3% triton_mm_1682 0.0178 ms 87.2% triton_mm_1686 0.0200 ms 77.3% triton_mm_1690 0.0223 ms 69.3% triton_mm_1677 0.0234 ms 66.2% triton_mm_1676 0.0242 ms 64.0% triton_mm_1685 0.0256 ms 60.4% triton_mm_1681 0.0258 ms 60.1% triton_mm_1675 0.0261 ms 59.2% SingleProcess AUTOTUNE benchmarking takes 7.8149 seconds and 0.0000 seconds precompiling AUTOTUNE mm(6864x12288, 12288x1088) mm 0.8157 ms 100.0% triton_mm_1652 0.8352 ms 97.7% triton_mm_1653 0.8487 ms 96.1% triton_mm_1654 0.8732 ms 93.4% triton_mm_1649 0.9558 ms 85.3% triton_mm_1645 0.9744 ms 83.7% triton_mm_1647 0.9833 ms 83.0% triton_mm_1648 1.0123 ms 80.6% triton_mm_1646 1.0816 ms 75.4% triton_mm_1650 1.0944 ms 74.5% SingleProcess AUTOTUNE benchmarking takes 2.6644 seconds and 0.0011 seconds precompiling AUTOTUNE mm(6864x1088, 1088x4352) triton_mm_1724 0.2731 ms 100.0% mm 0.2742 ms 99.6% triton_mm_1725 0.2760 ms 99.0% triton_mm_1726 0.3128 ms 87.3% triton_mm_1717 0.3344 ms 81.7% triton_mm_1719 0.3370 ms 81.0% triton_mm_1721 0.3417 ms 79.9% triton_mm_1718 0.3589 ms 76.1% triton_mm_1722 0.3805 ms 71.8% triton_mm_1720 0.4234 ms 64.5% SingleProcess AUTOTUNE benchmarking takes 2.5249 seconds and 0.0010 seconds precompiling AUTOTUNE mm(6864x3264, 3264x1088) triton_mm_1839 0.2188 ms 100.0% triton_mm_1838 0.2244 ms 97.5% triton_mm_1840 0.2361 ms 92.7% triton_mm_1835 0.2504 ms 87.4% triton_mm_1831 0.2547 ms 85.9% triton_mm_1833 0.2548 ms 85.9% mm 0.2577 ms 84.9% triton_mm_1834 0.2814 ms 77.8% triton_mm_1832 0.2824 ms 77.5% triton_mm_1836 0.2891 ms 75.7% SingleProcess AUTOTUNE benchmarking takes 9.6172 seconds and 0.0000 seconds precompiling AUTOTUNE mm(4352x6864, 6864x1088) triton_mm_1781 0.2823 ms 100.0% triton_mm_1782 0.2859 ms 98.7% mm 0.2874 ms 98.2% triton_mm_1783 0.2913 ms 96.9% triton_mm_1778 0.3407 ms 82.9% triton_mm_1774 0.3436 ms 82.2% triton_mm_1776 0.3470 ms 81.3% triton_mm_1777 0.3771 ms 74.9% triton_mm_1779 0.3816 ms 74.0% triton_mm_1775 0.3892 ms 72.5% SingleProcess AUTOTUNE benchmarking takes 9.9481 seconds and 0.0000 seconds precompiling AUTOTUNE mm(3264x6864, 6864x1088) mm 0.2203 ms 100.0% triton_mm_1857 0.2727 ms 80.8% triton_mm_1852 0.2765 ms 79.7% triton_mm_1854 0.2780 ms 79.2% triton_mm_1858 0.2804 ms 78.6% triton_mm_1850 0.2805 ms 78.5% triton_mm_1859 0.2836 ms 77.7% triton_mm_1853 0.3056 ms 72.1% triton_mm_1855 0.3186 ms 69.1% triton_mm_1851 0.3242 ms 67.9% SingleProcess AUTOTUNE benchmarking takes 9.9611 seconds and 0.0000 seconds precompiling AUTOTUNE mm(1088x6864, 6864x1088) mm 0.0968 ms 100.0% triton_mm_1821 0.0985 ms 98.3% triton_mm_1814 0.1040 ms 93.1% triton_mm_1820 0.1077 ms 89.8% triton_mm_1812 0.1186 ms 81.6% triton_mm_1811 0.1188 ms 81.5% triton_mm_1816 0.1195 ms 81.0% triton_mm_1819 0.1204 ms 80.4% triton_mm_1815 0.1271 ms 76.2% triton_mm_1817 0.1310 ms 73.9% SingleProcess AUTOTUNE benchmarking takes 12.5599 seconds and 0.0000 seconds precompiling AUTOTUNE mm(6864x4352, 4352x1088) mm 0.2934 ms 100.0% triton_mm_1762 0.2985 ms 98.3% triton_mm_1763 0.3061 ms 95.8% triton_mm_1764 0.3155 ms 93.0% triton_mm_1759 0.3385 ms 86.7% triton_mm_1755 0.3427 ms 85.6% triton_mm_1757 0.3441 ms 85.3% triton_mm_1758 0.3606 ms 81.3% triton_mm_1756 0.3795 ms 77.3% triton_mm_1760 0.3889 ms 75.4% SingleProcess AUTOTUNE benchmarking takes 2.5288 seconds and 0.0010 seconds precompiling AUTOTUNE mm(6864x1088, 1088x1088) triton_mm_1801 0.0834 ms 100.0% mm 0.0844 ms 98.7% triton_mm_1800 0.0849 ms 98.2% triton_mm_1797 0.0926 ms 90.0% triton_mm_1802 0.0944 ms 88.3% triton_mm_1793 0.0952 ms 87.6% triton_mm_1795 0.0952 ms 87.6% triton_mm_1794 0.1017 ms 82.0% triton_mm_1798 0.1049 ms 79.5% triton_mm_1796 0.1228 ms 67.9% SingleProcess AUTOTUNE benchmarking takes 2.4009 seconds and 0.0010 seconds precompiling AUTOTUNE mm(6864x1088, 1088x4352) mm 0.2712 ms 100.0% triton_mm_1910 0.2725 ms 99.5% triton_mm_1911 0.2758 ms 98.3% triton_mm_1912 0.3135 ms 86.5% triton_mm_1903 0.3340 ms 81.2% triton_mm_1905 0.3365 ms 80.6% triton_mm_1907 0.3425 ms 79.2% triton_mm_1904 0.3576 ms 75.8% triton_mm_1908 0.3808 ms 71.2% triton_mm_1906 0.4228 ms 64.1% SingleProcess AUTOTUNE benchmarking takes 2.5341 seconds and 0.0000 seconds precompiling AUTOTUNE mm(6864x4352, 4352x1088) mm 0.2945 ms 100.0% triton_mm_1948 0.2984 ms 98.7% triton_mm_1949 0.3064 ms 96.1% triton_mm_1950 0.3157 ms 93.3% triton_mm_1945 0.3400 ms 86.6% triton_mm_1943 0.3425 ms 86.0% triton_mm_1941 0.3445 ms 85.5% triton_mm_1944 0.3601 ms 81.8% triton_mm_1942 0.3809 ms 77.3% triton_mm_1946 0.3899 ms 75.5% SingleProcess AUTOTUNE benchmarking takes 2.5427 seconds and 0.0000 seconds precompiling AUTOTUNE mm(6528x24, 24x1088) triton_mm_1885 0.0173 ms 100.0% triton_mm_1886 0.0174 ms 99.6% triton_mm_1882 0.0179 ms 97.0% triton_mm_1887 0.0182 ms 95.1% triton_mm_1890 0.0182 ms 94.9% triton_mm_1888 0.0184 ms 94.2% triton_mm_1889 0.0185 ms 93.8% triton_mm_1883 0.0186 ms 93.3% triton_mm_1893 0.0192 ms 90.2% triton_mm_1892 0.0196 ms 88.2% SingleProcess AUTOTUNE benchmarking takes 5.9094 seconds and 0.0000 seconds precompiling AUTOTUNE mm(1088x6864, 6864x4352) mm 0.2834 ms 100.0% triton_mm_1930 0.2925 ms 96.9% triton_mm_1929 0.2934 ms 96.6% triton_mm_1931 0.2985 ms 94.9% triton_mm_1924 0.3480 ms 81.5% triton_mm_1922 0.3522 ms 80.5% triton_mm_1926 0.3597 ms 78.8% triton_mm_1925 0.3764 ms 75.3% triton_mm_1927 0.3943 ms 71.9% triton_mm_1923 0.3998 ms 70.9% SingleProcess AUTOTUNE benchmarking takes 2.5343 seconds and 0.0000 seconds precompiling AUTOTUNE mm(4352x6864, 6864x1088) triton_mm_1967 0.2860 ms 100.0% mm 0.2874 ms 99.5% triton_mm_1968 0.2913 ms 98.2% triton_mm_1969 0.2962 ms 96.5% triton_mm_1964 0.3423 ms 83.5% triton_mm_1960 0.3495 ms 81.8% triton_mm_1962 0.3530 ms 81.0% triton_mm_1963 0.3786 ms 75.5% triton_mm_1965 0.3887 ms 73.6% triton_mm_1961 0.3958 ms 72.2% SingleProcess AUTOTUNE benchmarking takes 2.5307 seconds and 0.0000 seconds precompiling AUTOTUNE mm(24x6528, 6528x1088) triton_mm_1864 0.0267 ms 100.0% mm 0.0269 ms 99.0% triton_mm_1868 0.0308 ms 86.6% triton_mm_1872 0.0399 ms 66.8% triton_mm_1876 0.0521 ms 51.2% triton_mm_1863 0.0524 ms 50.9% triton_mm_1862 0.0552 ms 48.3% triton_mm_1867 0.0604 ms 44.2% triton_mm_1871 0.0626 ms 42.6% triton_mm_1861 0.0660 ms 40.4% SingleProcess AUTOTUNE benchmarking takes 6.6684 seconds and 0.0000 seconds precompiling AUTOTUNE mm(3264x6864, 6864x1088) mm 0.2201 ms 100.0% triton_mm_2043 0.2748 ms 80.1% triton_mm_2038 0.2797 ms 78.7% triton_mm_2044 0.2812 ms 78.3% triton_mm_2040 0.2818 ms 78.1% triton_mm_2036 0.2844 ms 77.4% triton_mm_2045 0.2846 ms 77.3% triton_mm_2039 0.3080 ms 71.5% triton_mm_2041 0.3210 ms 68.6% triton_mm_2037 0.3250 ms 67.7% SingleProcess AUTOTUNE benchmarking takes 2.5072 seconds and 0.0000 seconds precompiling AUTOTUNE mm(1088x6864, 6864x1088) mm 0.0968 ms 100.0% triton_mm_2007 0.0985 ms 98.2% triton_mm_2000 0.1040 ms 93.1% triton_mm_2006 0.1078 ms 89.8% triton_mm_1998 0.1180 ms 82.1% triton_mm_1997 0.1181 ms 82.0% triton_mm_2002 0.1203 ms 80.4% triton_mm_2005 0.1204 ms 80.4% triton_mm_2001 0.1260 ms 76.8% triton_mm_2003 0.1302 ms 74.3% SingleProcess AUTOTUNE benchmarking takes 2.4159 seconds and 0.0000 seconds precompiling AUTOTUNE mm(6864x1088, 1088x1088) triton_mm_1987 0.0833 ms 100.0% mm 0.0834 ms 99.9% triton_mm_1986 0.0849 ms 98.2% triton_mm_1983 0.0919 ms 90.6% triton_mm_1988 0.0944 ms 88.3% triton_mm_1979 0.0945 ms 88.2% triton_mm_1981 0.0961 ms 86.7% triton_mm_1980 0.1024 ms 81.4% triton_mm_1984 0.1058 ms 78.7% triton_mm_1982 0.1227 ms 67.9% SingleProcess AUTOTUNE benchmarking takes 2.4121 seconds and 0.0000 seconds precompiling AUTOTUNE mm(6864x3264, 3264x1088) mm 0.2207 ms 100.0% triton_mm_2025 0.2217 ms 99.6% triton_mm_2024 0.2278 ms 96.9% triton_mm_2026 0.2396 ms 92.1% triton_mm_2021 0.2538 ms 87.0% triton_mm_2017 0.2592 ms 85.2% triton_mm_2019 0.2592 ms 85.1% triton_mm_2020 0.2850 ms 77.4% triton_mm_2018 0.2865 ms 77.0% triton_mm_2022 0.2937 ms 75.2% SingleProcess AUTOTUNE benchmarking takes 2.5107 seconds and 0.0000 seconds precompiling AUTOTUNE mm(6528x24, 24x1088) triton_mm_2075 0.0172 ms 100.0% triton_mm_2069 0.0172 ms 99.6% triton_mm_2072 0.0174 ms 98.7% triton_mm_2074 0.0174 ms 98.7% triton_mm_2073 0.0182 ms 94.2% triton_mm_2070 0.0184 ms 93.2% triton_mm_2071 0.0187 ms 91.9% triton_mm_2068 0.0188 ms 91.3% triton_mm_2077 0.0192 ms 89.3% triton_mm_2076 0.0194 ms 88.6% SingleProcess AUTOTUNE benchmarking takes 2.1207 seconds and 0.0000 seconds precompiling AUTOTUNE mm(1088x6864, 6864x4352) mm 0.2842 ms 100.0% triton_mm_2116 0.2914 ms 97.5% triton_mm_2115 0.2935 ms 96.8% triton_mm_2117 0.2980 ms 95.4% triton_mm_2110 0.3483 ms 81.6% triton_mm_2108 0.3505 ms 81.1% triton_mm_2112 0.3602 ms 78.9% triton_mm_2111 0.3755 ms 75.7% triton_mm_2113 0.3938 ms 72.2% triton_mm_2109 0.3998 ms 71.1% SingleProcess AUTOTUNE benchmarking takes 2.5303 seconds and 0.0000 seconds precompiling AUTOTUNE mm(4352x6864, 6864x1088) triton_mm_2153 0.2861 ms 100.0% mm 0.2875 ms 99.5% triton_mm_2154 0.2917 ms 98.1% triton_mm_2155 0.2958 ms 96.7% triton_mm_2150 0.3433 ms 83.4% triton_mm_2146 0.3487 ms 82.1% triton_mm_2148 0.3531 ms 81.0% triton_mm_2149 0.3790 ms 75.5% triton_mm_2151 0.3884 ms 73.7% triton_mm_2147 0.3956 ms 72.3% SingleProcess AUTOTUNE benchmarking takes 2.5312 seconds and 0.0000 seconds precompiling AUTOTUNE mm(24x6528, 6528x1088) triton_mm_2050 0.0267 ms 100.0% mm 0.0269 ms 99.3% triton_mm_2054 0.0320 ms 83.3% triton_mm_2058 0.0412 ms 64.7% triton_mm_2062 0.0508 ms 52.5% triton_mm_2049 0.0524 ms 50.9% triton_mm_2048 0.0562 ms 47.5% triton_mm_2053 0.0595 ms 44.9% triton_mm_2057 0.0618 ms 43.2% triton_mm_2047 0.0652 ms 41.0% SingleProcess AUTOTUNE benchmarking takes 2.0810 seconds and 0.0000 seconds precompiling AUTOTUNE mm(3264x6864, 6864x1088) mm 0.2209 ms 100.0% triton_mm_2229 0.2751 ms 80.3% triton_mm_2224 0.2798 ms 79.0% triton_mm_2230 0.2814 ms 78.5% triton_mm_2226 0.2819 ms 78.4% triton_mm_2222 0.2842 ms 77.7% triton_mm_2231 0.2849 ms 77.5% triton_mm_2225 0.3073 ms 71.9% triton_mm_2227 0.3213 ms 68.8% triton_mm_2223 0.3264 ms 67.7% SingleProcess AUTOTUNE benchmarking takes 2.5091 seconds and 0.0000 seconds precompiling AUTOTUNE mm(1088x6864, 6864x1088) mm 0.0968 ms 100.0% triton_mm_2193 0.0994 ms 97.3% triton_mm_2186 0.1040 ms 93.0% triton_mm_2192 0.1086 ms 89.1% triton_mm_2183 0.1180 ms 82.0% triton_mm_2184 0.1187 ms 81.5% triton_mm_2188 0.1204 ms 80.4% triton_mm_2191 0.1211 ms 79.9% triton_mm_2187 0.1261 ms 76.8% triton_mm_2189 0.1307 ms 74.1% SingleProcess AUTOTUNE benchmarking takes 2.4108 seconds and 0.0000 seconds precompiling AUTOTUNE mm(6864x1088, 1088x1088) mm 0.0834 ms 100.0% triton_mm_2173 0.0842 ms 99.0% triton_mm_2172 0.0849 ms 98.2% triton_mm_2169 0.0919 ms 90.8% triton_mm_2165 0.0945 ms 88.3% triton_mm_2174 0.0955 ms 87.4% triton_mm_2167 0.0960 ms 86.8% triton_mm_2166 0.1026 ms 81.3% triton_mm_2170 0.1057 ms 78.9% triton_mm_2168 0.1227 ms 68.0% SingleProcess AUTOTUNE benchmarking takes 3.3216 seconds and 0.0000 seconds precompiling AUTOTUNE mm(6864x3264, 3264x1088) mm 0.2196 ms 100.0% triton_mm_2211 0.2224 ms 98.7% triton_mm_2210 0.2282 ms 96.2% triton_mm_2212 0.2399 ms 91.5% triton_mm_2207 0.2531 ms 86.7% triton_mm_2205 0.2582 ms 85.0% triton_mm_2203 0.2602 ms 84.4% triton_mm_2206 0.2836 ms 77.4% triton_mm_2204 0.2874 ms 76.4% triton_mm_2208 0.2936 ms 74.8% SingleProcess AUTOTUNE benchmarking takes 2.5083 seconds and 0.0000 seconds precompiling AUTOTUNE mm(4352x6864, 6864x1088) mm 0.2844 ms 100.0% triton_mm_2339 0.2856 ms 99.6% triton_mm_2340 0.2915 ms 97.6% triton_mm_2341 0.2953 ms 96.3% triton_mm_2336 0.3407 ms 83.5% triton_mm_2332 0.3484 ms 81.6% triton_mm_2334 0.3518 ms 80.8% triton_mm_2335 0.3786 ms 75.1% triton_mm_2337 0.3867 ms 73.6% triton_mm_2333 0.3940 ms 72.2% SingleProcess AUTOTUNE benchmarking takes 2.5361 seconds and 0.0000 seconds precompiling AUTOTUNE mm(24x6528, 6528x1088) triton_mm_2236 0.0267 ms 100.0% mm 0.0270 ms 98.8% triton_mm_2240 0.0321 ms 83.1% triton_mm_2244 0.0413 ms 64.7% triton_mm_2248 0.0508 ms 52.5% triton_mm_2235 0.0524 ms 51.0% triton_mm_2234 0.0552 ms 48.3% triton_mm_2239 0.0595 ms 44.9% triton_mm_2243 0.0618 ms 43.2% triton_mm_2233 0.0652 ms 40.9% SingleProcess AUTOTUNE benchmarking takes 2.0946 seconds and 0.0000 seconds precompiling AUTOTUNE mm(6864x3264, 3264x1088) mm 0.2188 ms 100.0% triton_mm_2397 0.2218 ms 98.7% triton_mm_2396 0.2274 ms 96.3% triton_mm_2398 0.2395 ms 91.4% triton_mm_2393 0.2526 ms 86.6% triton_mm_2391 0.2578 ms 84.9% triton_mm_2389 0.2590 ms 84.5% triton_mm_2392 0.2838 ms 77.1% triton_mm_2390 0.2873 ms 76.2% triton_mm_2394 0.2928 ms 74.8% SingleProcess AUTOTUNE benchmarking takes 2.5152 seconds and 0.0000 seconds precompiling AUTOTUNE mm(6864x3264, 3264x1088) mm 0.2200 ms 100.0% triton_mm_2583 0.2206 ms 99.7% triton_mm_2582 0.2263 ms 97.2% triton_mm_2584 0.2384 ms 92.3% triton_mm_2579 0.2529 ms 87.0% triton_mm_2577 0.2573 ms 85.5% triton_mm_2575 0.2576 ms 85.4% triton_mm_2578 0.2838 ms 77.5% triton_mm_2576 0.2856 ms 77.0% triton_mm_2580 0.2917 ms 75.4% SingleProcess AUTOTUNE benchmarking takes 2.5132 seconds and 0.0000 seconds precompiling AUTOTUNE mm(1088x6840, 6840x96) mm 0.0313 ms 100.0% triton_mm_4874 0.0454 ms 68.9% triton_mm_4878 0.0539 ms 58.0% triton_mm_4872 0.0599 ms 52.2% triton_mm_4873 0.0620 ms 50.4% triton_mm_4871 0.0724 ms 43.2% triton_mm_4877 0.0740 ms 42.2% triton_mm_4882 0.0789 ms 39.6% triton_mm_4881 0.0810 ms 38.6% triton_mm_4880 0.0884 ms 35.3% SingleProcess AUTOTUNE benchmarking takes 9.7599 seconds and 0.0000 seconds precompiling /home/user/VAR/tokenizer/tokenizer_image/dino_enc/dinov2.py:122: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. with torch.cuda.amp.autocast(enabled=False): /home/user/VAR/tokenizer/tokenizer_image/dino_enc/dinov2.py:122: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. with torch.cuda.amp.autocast(enabled=False): /home/user/VAR/tokenizer/tokenizer_image/dino_enc/dinov2.py:122: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. with torch.cuda.amp.autocast(enabled=False): AUTOTUNE addmm(6840x1088, 6840x96, 96x1088) triton_mm_13 0.0358 ms 100.0% triton_mm_9 0.0379 ms 94.7% triton_mm_16 0.0380 ms 94.3% triton_mm_6 0.0423 ms 84.7% triton_mm_14 0.0427 ms 84.0% triton_mm_5 0.0433 ms 82.8% triton_mm_10 0.0440 ms 81.5% bias_addmm 0.0489 ms 73.3% triton_mm_15 0.0495 ms 72.4% triton_mm_7 0.0564 ms 63.5% SingleProcess AUTOTUNE benchmarking takes 2.2998 seconds and 0.0167 seconds precompiling AUTOTUNE addmm(24x6528, 24x1088, 1088x6528) triton_mm_31 0.0217 ms 100.0% triton_mm_35 0.0226 ms 96.2% triton_mm_23 0.0227 ms 95.5% triton_mm_27 0.0228 ms 95.4% triton_mm_30 0.0237 ms 91.5% triton_mm_26 0.0239 ms 90.6% bias_addmm 0.0241 ms 90.2% triton_mm_20 0.0245 ms 88.6% triton_mm_22 0.0253 ms 85.9% triton_mm_21 0.0254 ms 85.5% SingleProcess AUTOTUNE benchmarking takes 2.2001 seconds and 0.0008 seconds precompiling AUTOTUNE addmm(6864x3264, 6864x1088, 1088x3264) triton_mm_52 0.2168 ms 100.0% triton_mm_53 0.2236 ms 97.0% triton_mm_54 0.2517 ms 86.1% triton_mm_47 0.2596 ms 83.5% triton_mm_49 0.2679 ms 80.9% triton_mm_45 0.2721 ms 79.7% addmm 0.2757 ms 78.6% bias_addmm 0.3002 ms 72.2% triton_mm_46 0.3016 ms 71.9% triton_mm_50 0.3054 ms 71.0% SingleProcess AUTOTUNE benchmarking takes 2.6368 seconds and 0.0114 seconds precompiling AUTOTUNE addmm(6864x1088, 6864x1088, 1088x1088) triton_mm_72 0.0903 ms 100.0% triton_mm_71 0.0911 ms 99.1% bias_addmm 0.0958 ms 94.3% triton_mm_68 0.0980 ms 92.1% triton_mm_66 0.0987 ms 91.5% triton_mm_73 0.1004 ms 89.9% triton_mm_64 0.1013 ms 89.1% addmm 0.1101 ms 82.0% triton_mm_65 0.1116 ms 80.9% triton_mm_69 0.1132 ms 79.7% SingleProcess AUTOTUNE benchmarking takes 2.5133 seconds and 0.0010 seconds precompiling AUTOTUNE addmm(6864x4352, 6864x1088, 1088x4352) triton_mm_90 0.2859 ms 100.0% triton_mm_91 0.2905 ms 98.4% bias_addmm 0.3258 ms 87.7% triton_mm_92 0.3283 ms 87.1% triton_mm_85 0.3352 ms 85.3% triton_mm_83 0.3554 ms 80.5% triton_mm_87 0.3566 ms 80.2% triton_mm_84 0.3937 ms 72.6% triton_mm_88 0.4043 ms 70.7% addmm 0.4172 ms 68.5% SingleProcess AUTOTUNE benchmarking takes 2.6673 seconds and 0.0061 seconds precompiling AUTOTUNE addmm(6864x1088, 6864x4352, 4352x1088) bias_addmm 0.2862 ms 100.0% triton_mm_110 0.3023 ms 94.7% triton_mm_109 0.3121 ms 91.7% triton_mm_111 0.3210 ms 89.2% addmm 0.3214 ms 89.0% triton_mm_106 0.3503 ms 81.7% triton_mm_104 0.3527 ms 81.1% triton_mm_105 0.3639 ms 78.6% triton_mm_102 0.3669 ms 78.0% triton_mm_107 0.4009 ms 71.4% SingleProcess AUTOTUNE benchmarking takes 2.6557 seconds and 0.0010 seconds precompiling AUTOTUNE addmm(6864x4352, 6864x1088, 1088x4352) triton_mm_183 0.2873 ms 100.0% triton_mm_184 0.2901 ms 99.0% bias_addmm 0.3255 ms 88.3% triton_mm_185 0.3275 ms 87.7% triton_mm_178 0.3348 ms 85.8% triton_mm_176 0.3545 ms 81.0% triton_mm_180 0.3558 ms 80.7% addmm 0.3643 ms 78.9% triton_mm_177 0.3938 ms 73.0% triton_mm_181 0.4016 ms 71.5% SingleProcess AUTOTUNE benchmarking takes 2.6395 seconds and 0.0000 seconds precompiling AUTOTUNE addmm(6864x1088, 6864x4352, 4352x1088) bias_addmm 0.2871 ms 100.0% triton_mm_203 0.3015 ms 95.2% triton_mm_202 0.3111 ms 92.3% triton_mm_204 0.3219 ms 89.2% addmm 0.3228 ms 89.0% triton_mm_199 0.3511 ms 81.8% triton_mm_197 0.3538 ms 81.1% triton_mm_198 0.3656 ms 78.5% triton_mm_195 0.3666 ms 78.3% triton_mm_200 0.4012 ms 71.6% SingleProcess AUTOTUNE benchmarking takes 2.6466 seconds and 0.0000 seconds precompiling AUTOTUNE mm(24x1088, 1088x2176) triton_mm_1604 0.0133 ms 100.0% triton_mm_1608 0.0139 ms 95.6% triton_mm_1616 0.0170 ms 78.2% mm 0.0171 ms 77.6% triton_mm_1603 0.0175 ms 76.0% triton_mm_1602 0.0176 ms 75.5% triton_mm_1601 0.0186 ms 71.4% triton_mm_1612 0.0186 ms 71.4% triton_mm_1607 0.0187 ms 71.1% triton_mm_1611 0.0190 ms 69.9% SingleProcess AUTOTUNE benchmarking takes 7.7637 seconds and 0.0000 seconds precompiling AUTOTUNE addmm(6864x12288, 6864x1088, 1088x12288) bias_addmm 0.7996 ms 100.0% triton_mm_1634 0.8165 ms 97.9% triton_mm_1633 0.8200 ms 97.5% triton_mm_1635 0.9222 ms 86.7% triton_mm_1628 0.9664 ms 82.7% triton_mm_1630 1.0122 ms 79.0% addmm 1.0191 ms 78.5% triton_mm_1626 1.0444 ms 76.6% triton_mm_1631 1.1396 ms 70.2% triton_mm_1627 1.1497 ms 69.5% SingleProcess AUTOTUNE benchmarking takes 11.2543 seconds and 0.0000 seconds precompiling AUTOTUNE mm(12288x6864, 6864x1088) triton_mm_1671 0.7486 ms 100.0% triton_mm_1672 0.7774 ms 96.3% triton_mm_1673 0.8189 ms 91.4% mm 0.8695 ms 86.1% triton_mm_1668 0.9330 ms 80.2% triton_mm_1666 0.9787 ms 76.5% triton_mm_1664 0.9821 ms 76.2% triton_mm_1669 1.0421 ms 71.8% triton_mm_1667 1.0425 ms 71.8% triton_mm_1665 1.0836 ms 69.1% SingleProcess AUTOTUNE benchmarking takes 2.6498 seconds and 0.0139 seconds precompiling AUTOTUNE mm(1088x6864, 6864x4352) mm 0.2843 ms 100.0% triton_mm_1744 0.2926 ms 97.2% triton_mm_1743 0.2944 ms 96.6% triton_mm_1745 0.2979 ms 95.5% triton_mm_1738 0.3472 ms 81.9% triton_mm_1736 0.3496 ms 81.3% triton_mm_1740 0.3581 ms 79.4% triton_mm_1739 0.3749 ms 75.8% triton_mm_1741 0.3930 ms 72.4% triton_mm_1737 0.3992 ms 71.2% SingleProcess AUTOTUNE benchmarking takes 2.5334 seconds and 0.0010 seconds precompiling AUTOTUNE mm(2176x24, 24x1088) triton_mm_1702 0.0110 ms 100.0% triton_mm_1696 0.0117 ms 94.3% triton_mm_1706 0.0117 ms 94.0% triton_mm_1701 0.0119 ms 92.5% triton_mm_1699 0.0120 ms 92.2% triton_mm_1700 0.0121 ms 91.1% triton_mm_1697 0.0124 ms 88.8% triton_mm_1695 0.0124 ms 88.7% triton_mm_1703 0.0126 ms 87.5% triton_mm_1698 0.0127 ms 87.1% SingleProcess AUTOTUNE benchmarking takes 2.1236 seconds and 0.0009 seconds precompiling AUTOTUNE mm(24x2176, 2176x1088) triton_mm_1678 0.0164 ms 100.0% mm 0.0172 ms 95.4% triton_mm_1682 0.0175 ms 94.0% triton_mm_1686 0.0208 ms 78.8% triton_mm_1677 0.0224 ms 73.4% triton_mm_1676 0.0234 ms 70.1% triton_mm_1690 0.0236 ms 69.7% triton_mm_1681 0.0258 ms 63.7% triton_mm_1685 0.0263 ms 62.4% triton_mm_1675 0.0269 ms 61.0% SingleProcess AUTOTUNE benchmarking takes 7.4573 seconds and 0.0000 seconds precompiling AUTOTUNE mm(6864x12288, 12288x1088) mm 0.8232 ms 100.0% triton_mm_1652 0.8347 ms 98.6% triton_mm_1653 0.8462 ms 97.3% triton_mm_1654 0.8675 ms 94.9% triton_mm_1649 0.9502 ms 86.6% triton_mm_1645 0.9649 ms 85.3% triton_mm_1647 0.9704 ms 84.8% triton_mm_1648 0.9989 ms 82.4% triton_mm_1646 1.0764 ms 76.5% triton_mm_1650 1.0901 ms 75.5% SingleProcess AUTOTUNE benchmarking takes 9.6839 seconds and 0.0000 seconds precompiling AUTOTUNE mm(6864x1088, 1088x4352) triton_mm_1724 0.2723 ms 100.0% triton_mm_1725 0.2739 ms 99.4% mm 0.2764 ms 98.5% triton_mm_1726 0.3114 ms 87.5% triton_mm_1717 0.3340 ms 81.5% triton_mm_1719 0.3345 ms 81.4% triton_mm_1721 0.3388 ms 80.4% triton_mm_1718 0.3561 ms 76.5% triton_mm_1722 0.3781 ms 72.0% triton_mm_1720 0.4212 ms 64.6% SingleProcess AUTOTUNE benchmarking takes 10.2589 seconds and 0.0000 seconds precompiling AUTOTUNE mm(6864x3264, 3264x1088) triton_mm_1839 0.2200 ms 100.0% triton_mm_1838 0.2260 ms 97.4% triton_mm_1840 0.2375 ms 92.7% triton_mm_1835 0.2516 ms 87.4% triton_mm_1833 0.2554 ms 86.1% triton_mm_1831 0.2572 ms 85.5% mm 0.2576 ms 85.4% triton_mm_1834 0.2813 ms 78.2% triton_mm_1832 0.2848 ms 77.3% triton_mm_1836 0.2920 ms 75.3% SingleProcess AUTOTUNE benchmarking takes 2.4884 seconds and 0.0008 seconds precompiling AUTOTUNE mm(4352x6864, 6864x1088) triton_mm_1781 0.2857 ms 100.0% mm 0.2872 ms 99.5% triton_mm_1782 0.2901 ms 98.5% triton_mm_1783 0.2962 ms 96.4% triton_mm_1778 0.3407 ms 83.8% triton_mm_1774 0.3466 ms 82.4% triton_mm_1776 0.3517 ms 81.2% triton_mm_1777 0.3771 ms 75.7% triton_mm_1779 0.3870 ms 73.8% triton_mm_1775 0.3939 ms 72.5% SingleProcess AUTOTUNE benchmarking takes 2.5268 seconds and 0.0009 seconds precompiling AUTOTUNE mm(3264x6864, 6864x1088) mm 0.2214 ms 100.0% triton_mm_1857 0.2752 ms 80.5% triton_mm_1852 0.2782 ms 79.6% triton_mm_1854 0.2800 ms 79.1% triton_mm_1858 0.2813 ms 78.7% triton_mm_1850 0.2830 ms 78.2% triton_mm_1859 0.2850 ms 77.7% triton_mm_1853 0.3072 ms 72.1% triton_mm_1855 0.3207 ms 69.0% triton_mm_1851 0.3256 ms 68.0% SingleProcess AUTOTUNE benchmarking takes 2.5088 seconds and 0.0009 seconds precompiling AUTOTUNE mm(1088x6864, 6864x1088) mm 0.0995 ms 100.0% triton_mm_1821 0.1000 ms 99.5% triton_mm_1814 0.1045 ms 95.2% triton_mm_1820 0.1092 ms 91.1% triton_mm_1811 0.1189 ms 83.7% triton_mm_1812 0.1193 ms 83.4% triton_mm_1816 0.1204 ms 82.7% triton_mm_1819 0.1220 ms 81.5% triton_mm_1815 0.1264 ms 78.7% triton_mm_1817 0.1318 ms 75.5% SingleProcess AUTOTUNE benchmarking takes 2.4126 seconds and 0.0010 seconds precompiling AUTOTUNE mm(6864x4352, 4352x1088) mm 0.2929 ms 100.0% triton_mm_1762 0.2985 ms 98.1% triton_mm_1763 0.3057 ms 95.8% triton_mm_1764 0.3147 ms 93.1% triton_mm_1759 0.3372 ms 86.8% triton_mm_1757 0.3435 ms 85.3% triton_mm_1755 0.3438 ms 85.2% triton_mm_1758 0.3587 ms 81.7% triton_mm_1756 0.3807 ms 76.9% triton_mm_1760 0.3886 ms 75.4% SingleProcess AUTOTUNE benchmarking takes 2.5264 seconds and 0.0008 seconds precompiling AUTOTUNE mm(6864x1088, 1088x1088) triton_mm_1801 0.0835 ms 100.0% mm 0.0843 ms 99.0% triton_mm_1800 0.0860 ms 97.1% triton_mm_1797 0.0929 ms 89.9% triton_mm_1793 0.0944 ms 88.4% triton_mm_1802 0.0947 ms 88.2% triton_mm_1795 0.0961 ms 86.8% triton_mm_1794 0.1026 ms 81.3% triton_mm_1798 0.1052 ms 79.4% triton_mm_1796 0.1234 ms 67.7% SingleProcess AUTOTUNE benchmarking takes 2.4059 seconds and 0.0010 seconds precompiling AUTOTUNE mm(6864x1088, 1088x4352) triton_mm_1910 0.2723 ms 100.0% triton_mm_1911 0.2751 ms 99.0% triton_mm_1912 0.3121 ms 87.2% mm 0.3205 ms 84.9% triton_mm_1903 0.3355 ms 81.2% triton_mm_1905 0.3364 ms 80.9% triton_mm_1907 0.3413 ms 79.8% triton_mm_1904 0.3569 ms 76.3% triton_mm_1908 0.3788 ms 71.9% triton_mm_1906 0.4213 ms 64.6% SingleProcess AUTOTUNE benchmarking takes 2.5254 seconds and 0.0000 seconds precompiling AUTOTUNE mm(6864x4352, 4352x1088) mm 0.2939 ms 100.0% triton_mm_1948 0.2979 ms 98.7% triton_mm_1949 0.3049 ms 96.4% triton_mm_1950 0.3145 ms 93.4% triton_mm_1945 0.3392 ms 86.6% triton_mm_1943 0.3436 ms 85.5% triton_mm_1941 0.3454 ms 85.1% triton_mm_1944 0.3597 ms 81.7% triton_mm_1942 0.3801 ms 77.3% triton_mm_1946 0.3892 ms 75.5% SingleProcess AUTOTUNE benchmarking takes 2.5315 seconds and 0.0000 seconds precompiling AUTOTUNE mm(6528x24, 24x1088) triton_mm_1888 0.0173 ms 100.0% triton_mm_1885 0.0174 ms 99.1% triton_mm_1889 0.0180 ms 96.3% triton_mm_1890 0.0183 ms 94.2% triton_mm_1883 0.0184 ms 94.1% triton_mm_1886 0.0187 ms 92.3% triton_mm_1882 0.0188 ms 91.8% triton_mm_1892 0.0192 ms 89.8% triton_mm_1887 0.0193 ms 89.4% triton_mm_1884 0.0196 ms 88.1% SingleProcess AUTOTUNE benchmarking takes 5.9547 seconds and 0.0000 seconds precompiling AUTOTUNE mm(4352x6864, 6864x1088) mm 0.2844 ms 100.0% triton_mm_1967 0.2847 ms 99.9% triton_mm_1968 0.2893 ms 98.3% triton_mm_1969 0.2963 ms 96.0% triton_mm_1964 0.3405 ms 83.5% triton_mm_1960 0.3463 ms 82.1% triton_mm_1962 0.3509 ms 81.0% triton_mm_1963 0.3765 ms 75.5% triton_mm_1965 0.3875 ms 73.4% triton_mm_1961 0.3945 ms 72.1% SingleProcess AUTOTUNE benchmarking takes 2.5313 seconds and 0.0000 seconds precompiling AUTOTUNE mm(24x6528, 6528x1088) triton_mm_1864 0.0265 ms 100.0% mm 0.0268 ms 98.8% triton_mm_1868 0.0307 ms 86.3% triton_mm_1872 0.0413 ms 64.1% triton_mm_1876 0.0518 ms 51.2% triton_mm_1863 0.0527 ms 50.3% triton_mm_1862 0.0559 ms 47.4% triton_mm_1867 0.0610 ms 43.5% triton_mm_1871 0.0627 ms 42.3% triton_mm_1861 0.0656 ms 40.4% SingleProcess AUTOTUNE benchmarking takes 2.0822 seconds and 0.0008 seconds precompiling AUTOTUNE mm(3264x6864, 6864x1088) mm 0.2224 ms 100.0% triton_mm_2043 0.2739 ms 81.2% triton_mm_2038 0.2792 ms 79.7% triton_mm_2040 0.2798 ms 79.5% triton_mm_2036 0.2815 ms 79.0% triton_mm_2044 0.2818 ms 78.9% triton_mm_2045 0.2853 ms 77.9% triton_mm_2039 0.3067 ms 72.5% triton_mm_2041 0.3199 ms 69.5% triton_mm_2037 0.3250 ms 68.4% SingleProcess AUTOTUNE benchmarking takes 2.5110 seconds and 0.0000 seconds precompiling AUTOTUNE mm(1088x6864, 6864x1088) triton_mm_2007 0.1001 ms 100.0% mm 0.1007 ms 99.3% triton_mm_2000 0.1045 ms 95.7% triton_mm_2006 0.1091 ms 91.7% triton_mm_1997 0.1190 ms 84.1% triton_mm_1998 0.1193 ms 83.9% triton_mm_2002 0.1206 ms 83.0% triton_mm_2005 0.1221 ms 81.9% triton_mm_2001 0.1266 ms 79.1% triton_mm_2003 0.1315 ms 76.1% SingleProcess AUTOTUNE benchmarking takes 3.4135 seconds and 0.0000 seconds precompiling AUTOTUNE mm(6864x1088, 1088x1088) mm 0.0834 ms 100.0% triton_mm_1987 0.0834 ms 100.0% triton_mm_1986 0.0852 ms 97.9% triton_mm_1983 0.0920 ms 90.6% triton_mm_1988 0.0946 ms 88.2% triton_mm_1979 0.0951 ms 87.7% triton_mm_1981 0.0953 ms 87.5% triton_mm_1980 0.1025 ms 81.4% triton_mm_1984 0.1060 ms 78.7% triton_mm_1982 0.1243 ms 67.1% SingleProcess AUTOTUNE benchmarking takes 2.4165 seconds and 0.0000 seconds precompiling AUTOTUNE mm(6864x3264, 3264x1088) mm 0.2196 ms 100.0% triton_mm_2025 0.2216 ms 99.1% triton_mm_2024 0.2273 ms 96.6% triton_mm_2026 0.2389 ms 91.9% triton_mm_2021 0.2530 ms 86.8% triton_mm_2019 0.2574 ms 85.3% triton_mm_2017 0.2576 ms 85.3% triton_mm_2020 0.2827 ms 77.7% triton_mm_2018 0.2852 ms 77.0% triton_mm_2022 0.2918 ms 75.3% SingleProcess AUTOTUNE benchmarking takes 2.5112 seconds and 0.0000 seconds precompiling AUTOTUNE mm(6864x1088, 1088x4352) triton_mm_2096 0.2719 ms 100.0% mm 0.2733 ms 99.5% triton_mm_2097 0.2740 ms 99.2% triton_mm_2098 0.3120 ms 87.2% triton_mm_2089 0.3342 ms 81.4% triton_mm_2091 0.3350 ms 81.2% triton_mm_2093 0.3400 ms 80.0% triton_mm_2090 0.3557 ms 76.4% triton_mm_2094 0.3786 ms 71.8% triton_mm_2092 0.4207 ms 64.6% SingleProcess AUTOTUNE benchmarking takes 2.5177 seconds and 0.0000 seconds precompiling AUTOTUNE mm(6864x4352, 4352x1088) mm 0.2943 ms 100.0% triton_mm_2134 0.2980 ms 98.8% triton_mm_2135 0.3049 ms 96.5% triton_mm_2136 0.3131 ms 94.0% triton_mm_2131 0.3369 ms 87.4% triton_mm_2129 0.3419 ms 86.1% triton_mm_2127 0.3421 ms 86.0% triton_mm_2130 0.3601 ms 81.7% triton_mm_2128 0.3794 ms 77.6% triton_mm_2132 0.3874 ms 76.0% SingleProcess AUTOTUNE benchmarking takes 2.5369 seconds and 0.0000 seconds precompiling AUTOTUNE mm(6528x24, 24x1088) triton_mm_2069 0.0174 ms 100.0% triton_mm_2071 0.0174 ms 99.8% triton_mm_2073 0.0181 ms 95.9% triton_mm_2078 0.0183 ms 95.1% triton_mm_2070 0.0184 ms 94.8% triton_mm_2075 0.0184 ms 94.6% triton_mm_2074 0.0184 ms 94.4% triton_mm_2072 0.0185 ms 94.0% triton_mm_2068 0.0186 ms 93.5% triton_mm_2076 0.0191 ms 91.0% SingleProcess AUTOTUNE benchmarking takes 2.1293 seconds and 0.0000 seconds precompiling AUTOTUNE mm(4352x6864, 6864x1088) triton_mm_2153 0.2846 ms 100.0% mm 0.2847 ms 100.0% triton_mm_2154 0.2892 ms 98.4% triton_mm_2155 0.2958 ms 96.2% triton_mm_2150 0.3409 ms 83.5% triton_mm_2146 0.3470 ms 82.0% triton_mm_2148 0.3507 ms 81.2% triton_mm_2149 0.3768 ms 75.5% triton_mm_2151 0.3874 ms 73.5% triton_mm_2147 0.3947 ms 72.1% SingleProcess AUTOTUNE benchmarking takes 2.5270 seconds and 0.0000 seconds precompiling AUTOTUNE mm(24x6528, 6528x1088) triton_mm_2050 0.0265 ms 100.0% mm 0.0267 ms 99.3% triton_mm_2054 0.0307 ms 86.3% triton_mm_2058 0.0412 ms 64.4% triton_mm_2049 0.0519 ms 51.1% triton_mm_2062 0.0521 ms 50.9% triton_mm_2048 0.0559 ms 47.4% triton_mm_2053 0.0600 ms 44.1% triton_mm_2057 0.0627 ms 42.3% triton_mm_2047 0.0656 ms 40.4% SingleProcess AUTOTUNE benchmarking takes 2.0731 seconds and 0.0000 seconds precompiling AUTOTUNE mm(3264x6864, 6864x1088) mm 0.2224 ms 100.0% triton_mm_2229 0.2739 ms 81.2% triton_mm_2224 0.2786 ms 79.9% triton_mm_2226 0.2808 ms 79.2% triton_mm_2230 0.2815 ms 79.0% triton_mm_2222 0.2819 ms 78.9% triton_mm_2231 0.2851 ms 78.0% triton_mm_2225 0.3067 ms 72.5% triton_mm_2227 0.3204 ms 69.4% triton_mm_2223 0.3247 ms 68.5% SingleProcess AUTOTUNE benchmarking takes 2.5068 seconds and 0.0000 seconds precompiling AUTOTUNE mm(1088x6864, 6864x1088) triton_mm_2193 0.0992 ms 100.0% mm 0.1003 ms 98.9% triton_mm_2186 0.1054 ms 94.2% triton_mm_2192 0.1084 ms 91.5% triton_mm_2183 0.1182 ms 84.0% triton_mm_2188 0.1197 ms 82.9% triton_mm_2184 0.1199 ms 82.7% triton_mm_2191 0.1213 ms 81.8% triton_mm_2187 0.1274 ms 77.9% triton_mm_2189 0.1308 ms 75.9% SingleProcess AUTOTUNE benchmarking takes 2.4165 seconds and 0.0000 seconds precompiling AUTOTUNE mm(6864x1088, 1088x1088) triton_mm_2173 0.0834 ms 100.0% mm 0.0844 ms 98.8% triton_mm_2172 0.0851 ms 98.0% triton_mm_2169 0.0920 ms 90.7% triton_mm_2165 0.0944 ms 88.4% triton_mm_2174 0.0945 ms 88.3% triton_mm_2167 0.0960 ms 86.9% triton_mm_2166 0.1025 ms 81.4% triton_mm_2170 0.1059 ms 78.7% triton_mm_2168 0.1232 ms 67.7% SingleProcess AUTOTUNE benchmarking takes 2.4075 seconds and 0.0000 seconds precompiling AUTOTUNE mm(6864x3264, 3264x1088) mm 0.2195 ms 100.0% triton_mm_2211 0.2211 ms 99.3% triton_mm_2210 0.2269 ms 96.7% triton_mm_2212 0.2384 ms 92.0% triton_mm_2207 0.2528 ms 86.8% triton_mm_2205 0.2572 ms 85.3% triton_mm_2203 0.2577 ms 85.2% triton_mm_2204 0.2856 ms 76.8% triton_mm_2206 0.2858 ms 76.8% triton_mm_2208 0.2919 ms 75.2% SingleProcess AUTOTUNE benchmarking takes 2.5028 seconds and 0.0000 seconds precompiling AUTOTUNE mm(6864x1088, 1088x4352) triton_mm_2282 0.2716 ms 100.0% mm 0.2732 ms 99.4% triton_mm_2283 0.2743 ms 99.0% triton_mm_2284 0.3117 ms 87.1% triton_mm_2275 0.3323 ms 81.7% triton_mm_2277 0.3347 ms 81.1% triton_mm_2279 0.3405 ms 79.8% triton_mm_2276 0.3560 ms 76.3% triton_mm_2280 0.3789 ms 71.7% triton_mm_2278 0.4229 ms 64.2% SingleProcess AUTOTUNE benchmarking takes 2.5301 seconds and 0.0000 seconds precompiling AUTOTUNE mm(6864x4352, 4352x1088) mm 0.2947 ms 100.0% triton_mm_2320 0.2967 ms 99.3% triton_mm_2321 0.3037 ms 97.0% triton_mm_2322 0.3143 ms 93.7% triton_mm_2317 0.3370 ms 87.4% triton_mm_2315 0.3419 ms 86.2% triton_mm_2313 0.3436 ms 85.8% triton_mm_2316 0.3588 ms 82.1% triton_mm_2314 0.3796 ms 77.6% triton_mm_2318 0.3879 ms 76.0% SingleProcess AUTOTUNE benchmarking takes 2.5357 seconds and 0.0000 seconds precompiling AUTOTUNE mm(6528x24, 24x1088) triton_mm_2258 0.0173 ms 100.0% triton_mm_2260 0.0173 ms 100.0% triton_mm_2257 0.0174 ms 99.1% triton_mm_2254 0.0177 ms 97.5% triton_mm_2264 0.0182 ms 94.7% triton_mm_2261 0.0183 ms 94.6% triton_mm_2255 0.0184 ms 93.7% triton_mm_2256 0.0193 ms 89.7% triton_mm_2259 0.0193 ms 89.6% triton_mm_2262 0.0194 ms 89.3% SingleProcess AUTOTUNE benchmarking takes 2.1328 seconds and 0.0000 seconds precompiling AUTOTUNE mm(4352x6864, 6864x1088) mm 0.2848 ms 100.0% triton_mm_2339 0.2856 ms 99.7% triton_mm_2340 0.2897 ms 98.3% triton_mm_2341 0.2946 ms 96.7% triton_mm_2336 0.3393 ms 84.0% triton_mm_2332 0.3466 ms 82.2% triton_mm_2334 0.3506 ms 81.3% triton_mm_2335 0.3760 ms 75.7% triton_mm_2337 0.3854 ms 73.9% triton_mm_2333 0.3941 ms 72.3% SingleProcess AUTOTUNE benchmarking takes 2.5372 seconds and 0.0000 seconds precompiling AUTOTUNE mm(24x6528, 6528x1088) mm 0.0266 ms 100.0% triton_mm_2236 0.0276 ms 96.2% triton_mm_2240 0.0307 ms 86.5% triton_mm_2244 0.0412 ms 64.5% triton_mm_2248 0.0509 ms 52.1% triton_mm_2235 0.0526 ms 50.5% triton_mm_2234 0.0559 ms 47.5% triton_mm_2239 0.0608 ms 43.7% triton_mm_2243 0.0627 ms 42.4% triton_mm_2233 0.0656 ms 40.5% SingleProcess AUTOTUNE benchmarking takes 2.0849 seconds and 0.0000 seconds precompiling AUTOTUNE mm(6864x3264, 3264x1088) mm 0.2193 ms 100.0% triton_mm_2397 0.2197 ms 99.8% triton_mm_2396 0.2257 ms 97.2% triton_mm_2398 0.2374 ms 92.4% triton_mm_2393 0.2514 ms 87.2% triton_mm_2391 0.2563 ms 85.6% triton_mm_2389 0.2569 ms 85.4% triton_mm_2392 0.2849 ms 77.0% triton_mm_2390 0.2858 ms 76.7% triton_mm_2394 0.2917 ms 75.2% SingleProcess AUTOTUNE benchmarking takes 2.5050 seconds and 0.0000 seconds precompiling AUTOTUNE mm(6528x24, 24x1088) triton_mm_2444 0.0173 ms 100.0% triton_mm_2446 0.0173 ms 100.0% triton_mm_2443 0.0174 ms 99.1% triton_mm_2441 0.0181 ms 95.2% triton_mm_2445 0.0181 ms 95.2% triton_mm_2442 0.0184 ms 94.1% triton_mm_2447 0.0185 ms 93.3% triton_mm_2440 0.0188 ms 91.8% triton_mm_2448 0.0192 ms 90.0% triton_mm_2449 0.0192 ms 89.9% SingleProcess AUTOTUNE benchmarking takes 2.1276 seconds and 0.0000 seconds precompiling AUTOTUNE mm(4352x6864, 6864x1088) triton_mm_2525 0.2848 ms 100.0% mm 0.2865 ms 99.4% triton_mm_2526 0.2884 ms 98.8% triton_mm_2527 0.2947 ms 96.6% triton_mm_2522 0.3386 ms 84.1% triton_mm_2518 0.3457 ms 82.4% triton_mm_2520 0.3491 ms 81.6% triton_mm_2521 0.3754 ms 75.9% triton_mm_2523 0.3853 ms 73.9% triton_mm_2519 0.3920 ms 72.6% SingleProcess AUTOTUNE benchmarking takes 2.5302 seconds and 0.0000 seconds precompiling AUTOTUNE mm(24x6528, 6528x1088) triton_mm_2422 0.0265 ms 100.0% mm 0.0267 ms 99.4% triton_mm_2426 0.0320 ms 82.7% triton_mm_2430 0.0409 ms 64.8% triton_mm_2421 0.0519 ms 51.0% triton_mm_2434 0.0522 ms 50.8% triton_mm_2420 0.0568 ms 46.7% triton_mm_2425 0.0600 ms 44.1% triton_mm_2429 0.0627 ms 42.3% triton_mm_2419 0.0666 ms 39.8% SingleProcess AUTOTUNE benchmarking takes 2.0798 seconds and 0.0000 seconds precompiling AUTOTUNE mm(6864x3264, 3264x1088) mm 0.2199 ms 100.0% triton_mm_2583 0.2209 ms 99.5% triton_mm_2582 0.2267 ms 97.0% triton_mm_2584 0.2385 ms 92.2% triton_mm_2579 0.2518 ms 87.3% triton_mm_2575 0.2569 ms 85.6% triton_mm_2577 0.2571 ms 85.5% triton_mm_2578 0.2823 ms 77.9% triton_mm_2576 0.2843 ms 77.3% triton_mm_2580 0.2907 ms 75.6% SingleProcess AUTOTUNE benchmarking takes 2.5039 seconds and 0.0000 seconds precompiling AUTOTUNE mm(4352x6864, 6864x1088) triton_mm_2711 0.2848 ms 100.0% mm 0.2859 ms 99.6% triton_mm_2712 0.2881 ms 98.9% triton_mm_2713 0.2932 ms 97.2% triton_mm_2708 0.3391 ms 84.0% triton_mm_2704 0.3456 ms 82.4% triton_mm_2706 0.3498 ms 81.4% triton_mm_2707 0.3757 ms 75.8% triton_mm_2709 0.3841 ms 74.2% triton_mm_2705 0.3925 ms 72.6% SingleProcess AUTOTUNE benchmarking takes 2.5346 seconds and 0.0000 seconds precompiling AUTOTUNE mm(24x6528, 6528x1088) mm 0.0268 ms 100.0% triton_mm_2608 0.0277 ms 96.5% triton_mm_2612 0.0308 ms 87.1% triton_mm_2616 0.0400 ms 66.9% triton_mm_2620 0.0510 ms 52.5% triton_mm_2607 0.0527 ms 50.8% triton_mm_2606 0.0560 ms 47.9% triton_mm_2611 0.0610 ms 43.9% triton_mm_2615 0.0635 ms 42.2% triton_mm_2605 0.0656 ms 40.8% SingleProcess AUTOTUNE benchmarking takes 2.0794 seconds and 0.0000 seconds precompiling AUTOTUNE mm(4352x6864, 6864x1088) triton_mm_2897 0.2833 ms 100.0% mm 0.2855 ms 99.2% triton_mm_2898 0.2876 ms 98.5% triton_mm_2899 0.2929 ms 96.7% triton_mm_2894 0.3377 ms 83.9% triton_mm_2890 0.3448 ms 82.2% triton_mm_2892 0.3471 ms 81.6% triton_mm_2893 0.3745 ms 75.6% triton_mm_2895 0.3830 ms 74.0% triton_mm_2891 0.3907 ms 72.5% SingleProcess AUTOTUNE benchmarking takes 2.5348 seconds and 0.0000 seconds precompiling AUTOTUNE mm(24x6528, 6528x1088) mm 0.0266 ms 100.0% triton_mm_2794 0.0276 ms 96.5% triton_mm_2798 0.0307 ms 86.6% triton_mm_2802 0.0411 ms 64.7% triton_mm_2806 0.0509 ms 52.2% triton_mm_2793 0.0519 ms 51.3% triton_mm_2792 0.0567 ms 46.9% triton_mm_2797 0.0609 ms 43.7% triton_mm_2801 0.0627 ms 42.4% triton_mm_2791 0.0663 ms 40.1% SingleProcess AUTOTUNE benchmarking takes 2.0866 seconds and 0.0000 seconds precompiling AUTOTUNE mm(4352x6864, 6864x1088) triton_mm_3083 0.2840 ms 100.0% mm 0.2853 ms 99.6% triton_mm_3084 0.2879 ms 98.6% triton_mm_3085 0.2940 ms 96.6% triton_mm_3080 0.3380 ms 84.0% triton_mm_3076 0.3451 ms 82.3% triton_mm_3078 0.3529 ms 80.5% triton_mm_3079 0.3755 ms 75.6% triton_mm_3081 0.3831 ms 74.1% triton_mm_3077 0.3919 ms 72.5% SingleProcess AUTOTUNE benchmarking takes 2.5387 seconds and 0.0000 seconds precompiling AUTOTUNE mm(24x6528, 6528x1088) triton_mm_2980 0.0265 ms 100.0% mm 0.0266 ms 99.5% triton_mm_2984 0.0320 ms 82.8% triton_mm_2988 0.0400 ms 66.2% triton_mm_2979 0.0518 ms 51.1% triton_mm_2992 0.0522 ms 50.7% triton_mm_2978 0.0568 ms 46.6% triton_mm_2983 0.0609 ms 43.5% triton_mm_2987 0.0636 ms 41.7% triton_mm_2977 0.0665 ms 39.8% SingleProcess AUTOTUNE benchmarking takes 2.0811 seconds and 0.0000 seconds precompiling AUTOTUNE mm(1088x6840, 6840x96) mm 0.0309 ms 100.0% triton_mm_4874 0.0454 ms 68.1% triton_mm_4878 0.0528 ms 58.5% triton_mm_4872 0.0600 ms 51.5% triton_mm_4873 0.0625 ms 49.4% triton_mm_4871 0.0725 ms 42.6% triton_mm_4877 0.0734 ms 42.1% triton_mm_4882 0.0792 ms 39.0% triton_mm_4881 0.0814 ms 37.9% triton_mm_4880 0.0889 ms 34.7% SingleProcess AUTOTUNE benchmarking takes 10.7123 seconds and 0.0000 seconds precompiling /home/user/VAR/tokenizer/tokenizer_image/dino_enc/dinov2.py:122: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. with torch.cuda.amp.autocast(enabled=False): /home/user/VAR/tokenizer/tokenizer_image/dino_enc/dinov2.py:122: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. with torch.cuda.amp.autocast(enabled=False): /home/user/VAR/tokenizer/tokenizer_image/dino_enc/dinov2.py:122: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. with torch.cuda.amp.autocast(enabled=False): AUTOTUNE addmm(6840x1088, 6840x96, 96x1088) triton_mm_13 0.0366 ms 100.0% triton_mm_9 0.0379 ms 96.6% triton_mm_16 0.0379 ms 96.6% triton_mm_6 0.0422 ms 86.7% triton_mm_5 0.0424 ms 86.4% triton_mm_14 0.0433 ms 84.5% triton_mm_10 0.0437 ms 83.7% bias_addmm 0.0489 ms 74.8% triton_mm_15 0.0503 ms 72.8% triton_mm_11 0.0561 ms 65.3% SingleProcess AUTOTUNE benchmarking takes 2.2977 seconds and 0.0268 seconds precompiling AUTOTUNE addmm(24x6528, 24x1088, 1088x6528) triton_mm_31 0.0218 ms 100.0% triton_mm_35 0.0224 ms 97.3% triton_mm_27 0.0230 ms 94.7% triton_mm_23 0.0236 ms 92.7% triton_mm_26 0.0237 ms 92.2% triton_mm_20 0.0245 ms 89.0% triton_mm_30 0.0245 ms 89.0% bias_addmm 0.0251 ms 87.0% triton_mm_21 0.0252 ms 86.7% triton_mm_22 0.0252 ms 86.5% SingleProcess AUTOTUNE benchmarking takes 2.1972 seconds and 0.0009 seconds precompiling AUTOTUNE addmm(6864x3264, 6864x1088, 1088x3264) triton_mm_52 0.2188 ms 100.0% triton_mm_53 0.2245 ms 97.4% triton_mm_54 0.2525 ms 86.6% triton_mm_47 0.2624 ms 83.4% triton_mm_49 0.2705 ms 80.9% triton_mm_45 0.2730 ms 80.1% bias_addmm 0.2987 ms 73.2% triton_mm_46 0.3037 ms 72.0% triton_mm_50 0.3085 ms 70.9% addmm 0.3214 ms 68.1% SingleProcess AUTOTUNE benchmarking takes 2.5746 seconds and 0.0077 seconds precompiling AUTOTUNE addmm(6864x1088, 6864x1088, 1088x1088) triton_mm_72 0.0900 ms 100.0% triton_mm_71 0.0902 ms 99.8% bias_addmm 0.0963 ms 93.5% triton_mm_68 0.0980 ms 91.8% triton_mm_66 0.0984 ms 91.4% triton_mm_73 0.0989 ms 90.9% triton_mm_64 0.1014 ms 88.7% addmm 0.1094 ms 82.3% triton_mm_65 0.1112 ms 80.9% triton_mm_69 0.1120 ms 80.3% SingleProcess AUTOTUNE benchmarking takes 2.5117 seconds and 0.0009 seconds precompiling AUTOTUNE addmm(6864x4352, 6864x1088, 1088x4352) triton_mm_90 0.2895 ms 100.0% triton_mm_91 0.2937 ms 98.6% bias_addmm 0.3244 ms 89.2% triton_mm_92 0.3309 ms 87.5% triton_mm_85 0.3395 ms 85.3% triton_mm_83 0.3580 ms 80.9% triton_mm_87 0.3610 ms 80.2% addmm 0.3694 ms 78.4% triton_mm_84 0.3967 ms 73.0% triton_mm_88 0.4077 ms 71.0% SingleProcess AUTOTUNE benchmarking takes 2.6567 seconds and 0.0165 seconds precompiling AUTOTUNE addmm(6864x1088, 6864x4352, 4352x1088) bias_addmm 0.2937 ms 100.0% triton_mm_110 0.3052 ms 96.2% triton_mm_109 0.3140 ms 93.5% triton_mm_111 0.3255 ms 90.2% addmm 0.3285 ms 89.4% triton_mm_106 0.3545 ms 82.8% triton_mm_104 0.3566 ms 82.3% triton_mm_102 0.3690 ms 79.6% triton_mm_105 0.3704 ms 79.3% triton_mm_107 0.4046 ms 72.6% SingleProcess AUTOTUNE benchmarking takes 2.6584 seconds and 0.0010 seconds precompiling AUTOTUNE addmm(24x6528, 24x1088, 1088x6528) triton_mm_128 0.0214 ms 100.0% triton_mm_124 0.0218 ms 98.1% triton_mm_116 0.0232 ms 92.4% triton_mm_120 0.0232 ms 92.4% triton_mm_113 0.0236 ms 90.7% triton_mm_119 0.0236 ms 90.5% triton_mm_123 0.0237 ms 90.3% bias_addmm 0.0241 ms 88.8% triton_mm_114 0.0242 ms 88.4% triton_mm_115 0.0242 ms 88.4% SingleProcess AUTOTUNE benchmarking takes 2.1931 seconds and 0.0000 seconds precompiling AUTOTUNE addmm(24x6528, 24x1088, 1088x6528) triton_mm_221 0.0214 ms 100.0% triton_mm_217 0.0218 ms 98.4% triton_mm_213 0.0228 ms 93.7% triton_mm_209 0.0231 ms 92.7% triton_mm_206 0.0236 ms 90.7% triton_mm_212 0.0237 ms 90.4% triton_mm_207 0.0243 ms 88.3% triton_mm_208 0.0243 ms 88.3% triton_mm_216 0.0245 ms 87.3% bias_addmm 0.0260 ms 82.2% SingleProcess AUTOTUNE benchmarking takes 2.1738 seconds and 0.0000 seconds precompiling AUTOTUNE addmm(24x6528, 24x1088, 1088x6528) triton_mm_314 0.0224 ms 100.0% triton_mm_310 0.0229 ms 97.9% triton_mm_306 0.0230 ms 97.4% triton_mm_302 0.0236 ms 94.9% triton_mm_305 0.0236 ms 94.9% triton_mm_300 0.0242 ms 92.8% triton_mm_301 0.0242 ms 92.6% triton_mm_299 0.0244 ms 91.8% triton_mm_309 0.0245 ms 91.4% triton_mm_312 0.0262 ms 85.6% SingleProcess AUTOTUNE benchmarking takes 2.1769 seconds and 0.0000 seconds precompiling AUTOTUNE addmm(24x6528, 24x1088, 1088x6528) triton_mm_407 0.0214 ms 100.0% triton_mm_398 0.0227 ms 94.1% triton_mm_403 0.0229 ms 93.3% triton_mm_399 0.0232 ms 92.3% triton_mm_395 0.0234 ms 91.4% triton_mm_402 0.0236 ms 90.4% triton_mm_392 0.0244 ms 87.5% triton_mm_393 0.0252 ms 85.0% triton_mm_394 0.0252 ms 85.0% bias_addmm 0.0271 ms 79.0% SingleProcess AUTOTUNE benchmarking takes 2.1878 seconds and 0.0000 seconds precompiling AUTOTUNE addmm(24x6528, 24x1088, 1088x6528) triton_mm_500 0.0224 ms 100.0% triton_mm_491 0.0227 ms 98.7% triton_mm_496 0.0229 ms 98.0% triton_mm_492 0.0231 ms 97.2% triton_mm_485 0.0236 ms 95.0% triton_mm_488 0.0236 ms 95.0% triton_mm_495 0.0237 ms 94.7% triton_mm_487 0.0242 ms 92.6% triton_mm_486 0.0252 ms 89.2% bias_addmm 0.0260 ms 86.1% SingleProcess AUTOTUNE benchmarking takes 2.1798 seconds and 0.0000 seconds precompiling AUTOTUNE addmm(24x6528, 24x1088, 1088x6528) triton_mm_593 0.0214 ms 100.0% triton_mm_589 0.0218 ms 98.2% triton_mm_584 0.0227 ms 94.2% triton_mm_585 0.0229 ms 93.6% triton_mm_578 0.0236 ms 90.8% triton_mm_581 0.0236 ms 90.8% triton_mm_588 0.0236 ms 90.5% triton_mm_579 0.0243 ms 88.3% triton_mm_580 0.0243 ms 88.3% bias_addmm 0.0271 ms 78.9% SingleProcess AUTOTUNE benchmarking takes 2.2019 seconds and 0.0000 seconds precompiling AUTOTUNE addmm(24x6528, 24x1088, 1088x6528) triton_mm_686 0.0214 ms 100.0% triton_mm_677 0.0227 ms 94.1% triton_mm_682 0.0229 ms 93.3% triton_mm_678 0.0230 ms 92.9% triton_mm_674 0.0231 ms 92.5% triton_mm_681 0.0236 ms 90.4% triton_mm_672 0.0242 ms 88.2% triton_mm_673 0.0242 ms 88.2% triton_mm_671 0.0244 ms 87.4% bias_addmm 0.0260 ms 82.1% SingleProcess AUTOTUNE benchmarking takes 2.2025 seconds and 0.0000 seconds precompiling AUTOTUNE mm(24x1088, 1088x2176) triton_mm_1604 0.0144 ms 100.0% triton_mm_1608 0.0150 ms 95.9% triton_mm_1612 0.0155 ms 93.2% mm 0.0168 ms 86.0% triton_mm_1603 0.0168 ms 85.7% triton_mm_1616 0.0170 ms 84.7% triton_mm_1607 0.0176 ms 81.8% triton_mm_1602 0.0176 ms 81.7% triton_mm_1601 0.0186 ms 77.5% triton_mm_1614 0.0198 ms 72.8% SingleProcess AUTOTUNE benchmarking takes 7.7100 seconds and 0.0000 seconds precompiling AUTOTUNE addmm(6864x12288, 6864x1088, 1088x12288) bias_addmm 0.8139 ms 100.0% triton_mm_1634 0.8310 ms 97.9% triton_mm_1633 0.8342 ms 97.6% triton_mm_1635 0.9414 ms 86.5% triton_mm_1628 0.9851 ms 82.6% triton_mm_1630 1.0317 ms 78.9% addmm 1.0475 ms 77.7% triton_mm_1626 1.0657 ms 76.4% triton_mm_1631 1.1644 ms 69.9% triton_mm_1627 1.1732 ms 69.4% SingleProcess AUTOTUNE benchmarking takes 11.5766 seconds and 0.0000 seconds precompiling AUTOTUNE mm(12288x6864, 6864x1088) triton_mm_1671 0.7624 ms 100.0% triton_mm_1672 0.7890 ms 96.6% triton_mm_1673 0.8253 ms 92.4% mm 0.8660 ms 88.0% triton_mm_1668 0.9469 ms 80.5% triton_mm_1666 0.9909 ms 76.9% triton_mm_1664 0.9992 ms 76.3% triton_mm_1667 1.0540 ms 72.3% triton_mm_1669 1.0642 ms 71.6% triton_mm_1665 1.0881 ms 70.1% SingleProcess AUTOTUNE benchmarking takes 9.8260 seconds and 0.0000 seconds precompiling AUTOTUNE mm(1088x6864, 6864x4352) mm 0.2932 ms 100.0% triton_mm_1744 0.2971 ms 98.7% triton_mm_1743 0.2983 ms 98.3% triton_mm_1745 0.3012 ms 97.3% triton_mm_1738 0.3544 ms 82.7% triton_mm_1736 0.3561 ms 82.3% triton_mm_1740 0.3655 ms 80.2% triton_mm_1739 0.3793 ms 77.3% triton_mm_1741 0.3994 ms 73.4% triton_mm_1737 0.4051 ms 72.4% SingleProcess AUTOTUNE benchmarking takes 9.8693 seconds and 0.0000 seconds precompiling AUTOTUNE mm(2176x24, 24x1088) triton_mm_1700 0.0111 ms 100.0% triton_mm_1703 0.0115 ms 96.4% triton_mm_1697 0.0118 ms 94.3% triton_mm_1706 0.0119 ms 93.0% triton_mm_1701 0.0122 ms 91.1% triton_mm_1704 0.0122 ms 90.8% triton_mm_1702 0.0123 ms 90.4% triton_mm_1707 0.0124 ms 89.7% triton_mm_1696 0.0124 ms 89.4% triton_mm_1698 0.0127 ms 87.3% SingleProcess AUTOTUNE benchmarking takes 6.0005 seconds and 0.0000 seconds precompiling AUTOTUNE mm(24x2176, 2176x1088) triton_mm_1682 0.0164 ms 100.0% triton_mm_1678 0.0166 ms 98.6% mm 0.0175 ms 93.2% triton_mm_1686 0.0213 ms 76.8% triton_mm_1690 0.0224 ms 72.9% triton_mm_1677 0.0231 ms 70.8% triton_mm_1676 0.0235 ms 69.6% triton_mm_1685 0.0255 ms 64.2% triton_mm_1681 0.0258 ms 63.5% triton_mm_1675 0.0270 ms 60.5% SingleProcess AUTOTUNE benchmarking takes 7.7163 seconds and 0.0000 seconds precompiling AUTOTUNE mm(6864x12288, 12288x1088) mm 0.8407 ms 100.0% triton_mm_1652 0.8517 ms 98.7% triton_mm_1653 0.8638 ms 97.3% triton_mm_1654 0.8841 ms 95.1% triton_mm_1649 0.9722 ms 86.5% triton_mm_1645 0.9835 ms 85.5% triton_mm_1647 0.9912 ms 84.8% triton_mm_1648 1.0190 ms 82.5% triton_mm_1646 1.0942 ms 76.8% triton_mm_1650 1.1125 ms 75.6% SingleProcess AUTOTUNE benchmarking takes 9.9393 seconds and 0.0000 seconds precompiling AUTOTUNE mm(6864x1088, 1088x4352) triton_mm_1724 0.2763 ms 100.0% triton_mm_1725 0.2784 ms 99.3% mm 0.2791 ms 99.0% triton_mm_1726 0.3153 ms 87.6% triton_mm_1717 0.3396 ms 81.4% triton_mm_1719 0.3419 ms 80.8% triton_mm_1721 0.3477 ms 79.5% triton_mm_1718 0.3630 ms 76.1% triton_mm_1722 0.3860 ms 71.6% triton_mm_1720 0.4229 ms 65.3% SingleProcess AUTOTUNE benchmarking takes 2.5222 seconds and 0.0010 seconds precompiling AUTOTUNE mm(6864x3264, 3264x1088) triton_mm_1839 0.2241 ms 100.0% triton_mm_1838 0.2284 ms 98.1% triton_mm_1840 0.2413 ms 92.9% mm 0.2564 ms 87.4% triton_mm_1835 0.2567 ms 87.3% triton_mm_1833 0.2613 ms 85.8% triton_mm_1831 0.2628 ms 85.3% triton_mm_1834 0.2866 ms 78.2% triton_mm_1832 0.2892 ms 77.5% triton_mm_1836 0.2951 ms 75.9% SingleProcess AUTOTUNE benchmarking takes 2.5090 seconds and 0.0009 seconds precompiling AUTOTUNE mm(4352x6864, 6864x1088) triton_mm_1781 0.2910 ms 100.0% mm 0.2926 ms 99.5% triton_mm_1782 0.2935 ms 99.2% triton_mm_1783 0.2997 ms 97.1% triton_mm_1778 0.3467 ms 84.0% triton_mm_1774 0.3541 ms 82.2% triton_mm_1776 0.3580 ms 81.3% triton_mm_1777 0.3817 ms 76.2% triton_mm_1779 0.3916 ms 74.3% triton_mm_1775 0.3993 ms 72.9% SingleProcess AUTOTUNE benchmarking takes 9.9375 seconds and 0.0000 seconds precompiling AUTOTUNE mm(3264x6864, 6864x1088) mm 0.2254 ms 100.0% triton_mm_1857 0.2779 ms 81.1% triton_mm_1854 0.2826 ms 79.8% triton_mm_1852 0.2829 ms 79.7% triton_mm_1858 0.2848 ms 79.1% triton_mm_1850 0.2858 ms 78.8% triton_mm_1859 0.2881 ms 78.2% triton_mm_1853 0.3110 ms 72.5% triton_mm_1855 0.3252 ms 69.3% triton_mm_1851 0.3292 ms 68.5% SingleProcess AUTOTUNE benchmarking takes 9.9602 seconds and 0.0000 seconds precompiling AUTOTUNE mm(1088x6864, 6864x1088) mm 0.0965 ms 100.0% triton_mm_1814 0.1052 ms 91.7% triton_mm_1820 0.1084 ms 89.1% triton_mm_1811 0.1180 ms 81.8% triton_mm_1812 0.1195 ms 80.8% triton_mm_1821 0.1195 ms 80.8% triton_mm_1816 0.1197 ms 80.6% triton_mm_1819 0.1214 ms 79.5% triton_mm_1815 0.1262 ms 76.5% triton_mm_1817 0.1309 ms 73.7% SingleProcess AUTOTUNE benchmarking takes 9.8478 seconds and 0.0000 seconds precompiling AUTOTUNE mm(6864x4352, 4352x1088) mm 0.3016 ms 100.0% triton_mm_1762 0.3022 ms 99.8% triton_mm_1763 0.3087 ms 97.7% triton_mm_1764 0.3191 ms 94.5% triton_mm_1759 0.3411 ms 88.4% triton_mm_1757 0.3490 ms 86.4% triton_mm_1755 0.3504 ms 86.1% triton_mm_1758 0.3659 ms 82.4% triton_mm_1756 0.3851 ms 78.3% triton_mm_1760 0.3929 ms 76.7% SingleProcess AUTOTUNE benchmarking takes 9.7538 seconds and 0.0000 seconds precompiling AUTOTUNE mm(6864x1088, 1088x1088) triton_mm_1801 0.0833 ms 100.0% mm 0.0845 ms 98.6% triton_mm_1800 0.0856 ms 97.3% triton_mm_1797 0.0920 ms 90.5% triton_mm_1802 0.0943 ms 88.4% triton_mm_1793 0.0951 ms 87.6% triton_mm_1795 0.0953 ms 87.4% triton_mm_1794 0.1016 ms 82.0% triton_mm_1798 0.1049 ms 79.4% triton_mm_1796 0.1234 ms 67.5% SingleProcess AUTOTUNE benchmarking takes 2.4162 seconds and 0.0011 seconds precompiling AUTOTUNE mm(6864x1088, 1088x4352) triton_mm_1910 0.2770 ms 100.0% triton_mm_1911 0.2777 ms 99.8% triton_mm_1912 0.3154 ms 87.8% mm 0.3185 ms 87.0% triton_mm_1903 0.3379 ms 82.0% triton_mm_1905 0.3410 ms 81.2% triton_mm_1907 0.3454 ms 80.2% triton_mm_1904 0.3618 ms 76.6% triton_mm_1908 0.3835 ms 72.2% triton_mm_1906 0.4225 ms 65.6% SingleProcess AUTOTUNE benchmarking takes 2.5347 seconds and 0.0000 seconds precompiling AUTOTUNE mm(6864x4352, 4352x1088) mm 0.3011 ms 100.0% triton_mm_1948 0.3013 ms 99.9% triton_mm_1949 0.3089 ms 97.5% triton_mm_1950 0.3188 ms 94.4% triton_mm_1945 0.3415 ms 88.2% triton_mm_1943 0.3483 ms 86.4% triton_mm_1941 0.3494 ms 86.2% triton_mm_1944 0.3652 ms 82.4% triton_mm_1942 0.3847 ms 78.3% triton_mm_1946 0.3925 ms 76.7% SingleProcess AUTOTUNE benchmarking takes 2.5382 seconds and 0.0000 seconds precompiling AUTOTUNE mm(6528x24, 24x1088) triton_mm_1888 0.0174 ms 100.0% triton_mm_1883 0.0183 ms 95.3% triton_mm_1887 0.0183 ms 95.3% triton_mm_1885 0.0184 ms 94.8% triton_mm_1889 0.0184 ms 94.8% triton_mm_1884 0.0184 ms 94.4% triton_mm_1886 0.0185 ms 94.3% triton_mm_1882 0.0187 ms 93.3% triton_mm_1892 0.0192 ms 90.5% triton_mm_1891 0.0193 ms 90.1% SingleProcess AUTOTUNE benchmarking takes 5.8602 seconds and 0.0000 seconds precompiling AUTOTUNE mm(1088x6864, 6864x4352) mm 0.2915 ms 100.0% triton_mm_1930 0.2956 ms 98.6% triton_mm_1929 0.2958 ms 98.5% triton_mm_1931 0.3009 ms 96.9% triton_mm_1924 0.3529 ms 82.6% triton_mm_1922 0.3549 ms 82.1% triton_mm_1926 0.3632 ms 80.3% triton_mm_1925 0.3775 ms 77.2% triton_mm_1927 0.3982 ms 73.2% triton_mm_1923 0.4043 ms 72.1% SingleProcess AUTOTUNE benchmarking takes 2.5391 seconds and 0.0000 seconds precompiling AUTOTUNE mm(4352x6864, 6864x1088) triton_mm_1967 0.2891 ms 100.0% mm 0.2905 ms 99.5% triton_mm_1968 0.2936 ms 98.4% triton_mm_1969 0.3004 ms 96.2% triton_mm_1964 0.3468 ms 83.4% triton_mm_1960 0.3542 ms 81.6% triton_mm_1962 0.3575 ms 80.8% triton_mm_1963 0.3833 ms 75.4% triton_mm_1965 0.3929 ms 73.6% triton_mm_1961 0.3991 ms 72.4% SingleProcess AUTOTUNE benchmarking takes 2.5345 seconds and 0.0000 seconds precompiling AUTOTUNE mm(24x6528, 6528x1088) triton_mm_1864 0.0265 ms 100.0% mm 0.0269 ms 98.6% triton_mm_1868 0.0308 ms 86.3% triton_mm_1872 0.0409 ms 64.9% triton_mm_1863 0.0516 ms 51.4% triton_mm_1876 0.0519 ms 51.1% triton_mm_1862 0.0553 ms 48.0% triton_mm_1867 0.0596 ms 44.5% triton_mm_1871 0.0620 ms 42.8% triton_mm_1861 0.0650 ms 40.8% SingleProcess AUTOTUNE benchmarking takes 2.0890 seconds and 0.0007 seconds precompiling AUTOTUNE mm(3264x6864, 6864x1088) mm 0.2264 ms 100.0% triton_mm_2043 0.2771 ms 81.7% triton_mm_2038 0.2828 ms 80.1% triton_mm_2044 0.2835 ms 79.9% triton_mm_2040 0.2841 ms 79.7% triton_mm_2036 0.2869 ms 78.9% triton_mm_2045 0.2874 ms 78.8% triton_mm_2039 0.3115 ms 72.7% triton_mm_2041 0.3240 ms 69.9% triton_mm_2037 0.3300 ms 68.6% SingleProcess AUTOTUNE benchmarking takes 2.5174 seconds and 0.0000 seconds precompiling AUTOTUNE mm(1088x6864, 6864x1088) mm 0.0975 ms 100.0% triton_mm_2007 0.0995 ms 98.0% triton_mm_2000 0.1053 ms 92.6% triton_mm_2006 0.1083 ms 90.0% triton_mm_1997 0.1179 ms 82.6% triton_mm_1998 0.1196 ms 81.5% triton_mm_2002 0.1197 ms 81.4% triton_mm_2005 0.1217 ms 80.1% triton_mm_2001 0.1270 ms 76.7% triton_mm_2003 0.1310 ms 74.4% SingleProcess AUTOTUNE benchmarking takes 2.4128 seconds and 0.0000 seconds precompiling AUTOTUNE mm(6864x1088, 1088x1088) triton_mm_1987 0.0834 ms 100.0% mm 0.0845 ms 98.6% triton_mm_1986 0.0850 ms 98.1% triton_mm_1983 0.0929 ms 89.7% triton_mm_1988 0.0943 ms 88.4% triton_mm_1979 0.0953 ms 87.5% triton_mm_1981 0.0953 ms 87.5% triton_mm_1980 0.1017 ms 82.0% triton_mm_1984 0.1051 ms 79.3% triton_mm_1977 0.1237 ms 67.4% SingleProcess AUTOTUNE benchmarking takes 2.4077 seconds and 0.0000 seconds precompiling AUTOTUNE mm(6864x3264, 3264x1088) mm 0.2236 ms 100.0% triton_mm_2025 0.2244 ms 99.6% triton_mm_2024 0.2290 ms 97.6% triton_mm_2026 0.2426 ms 92.1% triton_mm_2021 0.2577 ms 86.8% triton_mm_2019 0.2626 ms 85.1% triton_mm_2017 0.2632 ms 84.9% triton_mm_2020 0.2879 ms 77.7% triton_mm_2018 0.2907 ms 76.9% triton_mm_2022 0.2965 ms 75.4% SingleProcess AUTOTUNE benchmarking takes 2.5150 seconds and 0.0000 seconds precompiling AUTOTUNE mm(6864x1088, 1088x4352) triton_mm_2096 0.2772 ms 100.0% triton_mm_2097 0.2783 ms 99.6% mm 0.2797 ms 99.1% triton_mm_2098 0.3164 ms 87.6% triton_mm_2089 0.3391 ms 81.7% triton_mm_2091 0.3420 ms 81.0% triton_mm_2093 0.3475 ms 79.8% triton_mm_2090 0.3628 ms 76.4% triton_mm_2094 0.3850 ms 72.0% triton_mm_2092 0.4227 ms 65.6% SingleProcess AUTOTUNE benchmarking takes 2.5221 seconds and 0.0000 seconds precompiling AUTOTUNE mm(6864x4352, 4352x1088) mm 0.3014 ms 100.0% triton_mm_2134 0.3031 ms 99.4% triton_mm_2135 0.3101 ms 97.2% triton_mm_2136 0.3199 ms 94.2% triton_mm_2131 0.3432 ms 87.8% triton_mm_2129 0.3495 ms 86.2% triton_mm_2127 0.3511 ms 85.8% triton_mm_2130 0.3663 ms 82.3% triton_mm_2128 0.3857 ms 78.1% triton_mm_2132 0.3946 ms 76.4% SingleProcess AUTOTUNE benchmarking takes 2.5327 seconds and 0.0000 seconds precompiling AUTOTUNE mm(1088x6864, 6864x4352) mm 0.2914 ms 100.0% triton_mm_2116 0.2968 ms 98.2% triton_mm_2115 0.2973 ms 98.0% triton_mm_2117 0.3025 ms 96.3% triton_mm_2110 0.3536 ms 82.4% triton_mm_2108 0.3556 ms 82.0% triton_mm_2112 0.3655 ms 79.7% triton_mm_2111 0.3794 ms 76.8% triton_mm_2113 0.3989 ms 73.0% triton_mm_2109 0.4051 ms 71.9% SingleProcess AUTOTUNE benchmarking takes 2.5269 seconds and 0.0000 seconds precompiling AUTOTUNE mm(4352x6864, 6864x1088) triton_mm_2153 0.2888 ms 100.0% mm 0.2924 ms 98.8% triton_mm_2154 0.2947 ms 98.0% triton_mm_2155 0.3011 ms 95.9% triton_mm_2150 0.3468 ms 83.3% triton_mm_2146 0.3553 ms 81.3% triton_mm_2148 0.3581 ms 80.6% triton_mm_2149 0.3836 ms 75.3% triton_mm_2151 0.3924 ms 73.6% triton_mm_2147 0.3999 ms 72.2% SingleProcess AUTOTUNE benchmarking takes 2.5280 seconds and 0.0000 seconds precompiling AUTOTUNE mm(24x6528, 6528x1088) triton_mm_2050 0.0265 ms 100.0% mm 0.0270 ms 98.3% triton_mm_2054 0.0308 ms 86.3% triton_mm_2058 0.0411 ms 64.6% triton_mm_2062 0.0508 ms 52.2% triton_mm_2049 0.0516 ms 51.5% triton_mm_2048 0.0561 ms 47.3% triton_mm_2053 0.0596 ms 44.5% triton_mm_2057 0.0620 ms 42.8% triton_mm_2047 0.0650 ms 40.8% SingleProcess AUTOTUNE benchmarking takes 2.0849 seconds and 0.0000 seconds precompiling AUTOTUNE mm(3264x6864, 6864x1088) mm 0.2267 ms 100.0% triton_mm_2229 0.2776 ms 81.7% triton_mm_2230 0.2838 ms 79.9% triton_mm_2224 0.2840 ms 79.8% triton_mm_2226 0.2854 ms 79.4% triton_mm_2222 0.2858 ms 79.3% triton_mm_2231 0.2890 ms 78.4% triton_mm_2225 0.3117 ms 72.7% triton_mm_2227 0.3254 ms 69.7% triton_mm_2223 0.3298 ms 68.7% SingleProcess AUTOTUNE benchmarking takes 2.5179 seconds and 0.0000 seconds precompiling AUTOTUNE mm(1088x6864, 6864x1088) mm 0.0966 ms 100.0% triton_mm_2193 0.0985 ms 98.1% triton_mm_2186 0.1045 ms 92.4% triton_mm_2192 0.1074 ms 89.9% triton_mm_2183 0.1180 ms 81.8% triton_mm_2184 0.1196 ms 80.8% triton_mm_2188 0.1204 ms 80.2% triton_mm_2191 0.1208 ms 79.9% triton_mm_2187 0.1262 ms 76.5% triton_mm_2189 0.1302 ms 74.2% SingleProcess AUTOTUNE benchmarking takes 2.4241 seconds and 0.0000 seconds precompiling AUTOTUNE mm(6864x1088, 1088x1088) mm 0.0841 ms 100.0% triton_mm_2173 0.0842 ms 99.9% triton_mm_2172 0.0850 ms 99.0% triton_mm_2169 0.0929 ms 90.5% triton_mm_2174 0.0943 ms 89.2% triton_mm_2167 0.0953 ms 88.2% triton_mm_2165 0.0954 ms 88.2% triton_mm_2166 0.1017 ms 82.7% triton_mm_2170 0.1051 ms 80.0% triton_mm_2168 0.1241 ms 67.8% SingleProcess AUTOTUNE benchmarking takes 2.4165 seconds and 0.0000 seconds precompiling AUTOTUNE mm(6864x3264, 3264x1088) mm 0.2237 ms 100.0% triton_mm_2211 0.2242 ms 99.8% triton_mm_2210 0.2294 ms 97.5% triton_mm_2212 0.2416 ms 92.6% triton_mm_2207 0.2582 ms 86.7% triton_mm_2205 0.2632 ms 85.0% triton_mm_2203 0.2632 ms 85.0% triton_mm_2206 0.2882 ms 77.6% triton_mm_2204 0.2911 ms 76.9% triton_mm_2208 0.2975 ms 75.2% SingleProcess AUTOTUNE benchmarking takes 2.5131 seconds and 0.0000 seconds precompiling AUTOTUNE mm(6864x1088, 1088x4352) triton_mm_2282 0.2772 ms 100.0% triton_mm_2283 0.2784 ms 99.6% mm 0.2793 ms 99.3% triton_mm_2284 0.3165 ms 87.6% triton_mm_2275 0.3410 ms 81.3% triton_mm_2277 0.3427 ms 80.9% triton_mm_2279 0.3470 ms 79.9% triton_mm_2276 0.3623 ms 76.5% triton_mm_2280 0.3844 ms 72.1% triton_mm_2278 0.4231 ms 65.5% SingleProcess AUTOTUNE benchmarking takes 2.5336 seconds and 0.0000 seconds precompiling AUTOTUNE mm(6864x4352, 4352x1088) mm 0.3015 ms 100.0% triton_mm_2320 0.3034 ms 99.4% triton_mm_2321 0.3103 ms 97.2% triton_mm_2322 0.3197 ms 94.3% triton_mm_2317 0.3432 ms 87.8% triton_mm_2315 0.3497 ms 86.2% triton_mm_2313 0.3509 ms 85.9% triton_mm_2316 0.3671 ms 82.1% triton_mm_2314 0.3866 ms 78.0% triton_mm_2318 0.3948 ms 76.4% SingleProcess AUTOTUNE benchmarking takes 2.5405 seconds and 0.0000 seconds precompiling AUTOTUNE mm(1088x6864, 6864x4352) mm 0.2921 ms 100.0% triton_mm_2302 0.2968 ms 98.4% triton_mm_2301 0.2986 ms 97.8% triton_mm_2303 0.3014 ms 96.9% triton_mm_2296 0.3547 ms 82.3% triton_mm_2294 0.3564 ms 82.0% triton_mm_2298 0.3654 ms 79.9% triton_mm_2297 0.3799 ms 76.9% triton_mm_2299 0.3999 ms 73.0% triton_mm_2295 0.4054 ms 72.1% SingleProcess AUTOTUNE benchmarking takes 2.5470 seconds and 0.0000 seconds precompiling AUTOTUNE mm(4352x6864, 6864x1088) triton_mm_2339 0.2901 ms 100.0% mm 0.2927 ms 99.1% triton_mm_2340 0.2945 ms 98.5% triton_mm_2341 0.3007 ms 96.5% triton_mm_2336 0.3478 ms 83.4% triton_mm_2332 0.3553 ms 81.7% triton_mm_2334 0.3582 ms 81.0% triton_mm_2335 0.3842 ms 75.5% triton_mm_2337 0.3944 ms 73.6% triton_mm_2333 0.4008 ms 72.4% SingleProcess AUTOTUNE benchmarking takes 2.5406 seconds and 0.0000 seconds precompiling AUTOTUNE mm(24x6528, 6528x1088) triton_mm_2236 0.0266 ms 100.0% mm 0.0272 ms 97.9% triton_mm_2240 0.0318 ms 83.5% triton_mm_2244 0.0410 ms 64.8% triton_mm_2248 0.0515 ms 51.6% triton_mm_2235 0.0516 ms 51.6% triton_mm_2234 0.0561 ms 47.4% triton_mm_2239 0.0603 ms 44.1% triton_mm_2243 0.0620 ms 42.9% triton_mm_2233 0.0651 ms 40.9% SingleProcess AUTOTUNE benchmarking takes 2.1044 seconds and 0.0000 seconds precompiling AUTOTUNE mm(3264x6864, 6864x1088) mm 0.2262 ms 100.0% triton_mm_2415 0.2776 ms 81.5% triton_mm_2410 0.2831 ms 79.9% triton_mm_2416 0.2839 ms 79.7% triton_mm_2412 0.2850 ms 79.4% triton_mm_2408 0.2871 ms 78.8% triton_mm_2417 0.2879 ms 78.6% triton_mm_2411 0.3125 ms 72.4% triton_mm_2413 0.3239 ms 69.8% triton_mm_2409 0.3305 ms 68.4% SingleProcess AUTOTUNE benchmarking takes 2.5231 seconds and 0.0000 seconds precompiling AUTOTUNE mm(1088x6864, 6864x1088) mm 0.0966 ms 100.0% triton_mm_2379 0.0995 ms 97.1% triton_mm_2372 0.1053 ms 91.7% triton_mm_2378 0.1083 ms 89.2% triton_mm_2369 0.1180 ms 81.9% triton_mm_2370 0.1195 ms 80.9% triton_mm_2374 0.1197 ms 80.7% triton_mm_2377 0.1216 ms 79.4% triton_mm_2373 0.1270 ms 76.1% triton_mm_2375 0.1311 ms 73.7% SingleProcess AUTOTUNE benchmarking takes 2.4295 seconds and 0.0000 seconds precompiling AUTOTUNE mm(24x6528, 6528x1088) triton_mm_2422 0.0265 ms 100.0% mm 0.0269 ms 98.3% triton_mm_2426 0.0308 ms 86.2% triton_mm_2430 0.0411 ms 64.5% triton_mm_2421 0.0516 ms 51.4% triton_mm_2434 0.0516 ms 51.3% triton_mm_2420 0.0554 ms 47.8% triton_mm_2425 0.0596 ms 44.5% triton_mm_2429 0.0620 ms 42.7% triton_mm_2419 0.0658 ms 40.3% SingleProcess AUTOTUNE benchmarking takes 2.0960 seconds and 0.0000 seconds precompiling AUTOTUNE mm(1088x6864, 6864x1088) mm 0.0967 ms 100.0% triton_mm_2565 0.0993 ms 97.4% triton_mm_2558 0.1052 ms 92.0% triton_mm_2564 0.1081 ms 89.5% triton_mm_2555 0.1180 ms 82.0% triton_mm_2556 0.1196 ms 80.9% triton_mm_2560 0.1197 ms 80.8% triton_mm_2563 0.1216 ms 79.5% triton_mm_2559 0.1271 ms 76.1% triton_mm_2561 0.1311 ms 73.8% SingleProcess AUTOTUNE benchmarking takes 2.4220 seconds and 0.0000 seconds precompiling AUTOTUNE mm(24x6528, 6528x1088) triton_mm_2608 0.0265 ms 100.0% mm 0.0269 ms 98.7% triton_mm_2612 0.0307 ms 86.4% triton_mm_2616 0.0411 ms 64.6% triton_mm_2607 0.0516 ms 51.5% triton_mm_2620 0.0518 ms 51.2% triton_mm_2606 0.0553 ms 48.0% triton_mm_2611 0.0603 ms 44.0% triton_mm_2615 0.0627 ms 42.3% triton_mm_2605 0.0657 ms 40.4% SingleProcess AUTOTUNE benchmarking takes 2.0993 seconds and 0.0000 seconds precompiling AUTOTUNE mm(1088x6864, 6864x1088) mm 0.0967 ms 100.0% triton_mm_2751 0.0994 ms 97.3% triton_mm_2744 0.1044 ms 92.6% triton_mm_2750 0.1074 ms 90.0% triton_mm_2741 0.1186 ms 81.5% triton_mm_2742 0.1189 ms 81.3% triton_mm_2746 0.1205 ms 80.2% triton_mm_2749 0.1208 ms 80.1% triton_mm_2745 0.1262 ms 76.6% triton_mm_2747 0.1302 ms 74.3% SingleProcess AUTOTUNE benchmarking takes 2.4262 seconds and 0.0000 seconds precompiling AUTOTUNE mm(24x6528, 6528x1088) triton_mm_2794 0.0265 ms 100.0% mm 0.0269 ms 98.7% triton_mm_2798 0.0308 ms 86.3% triton_mm_2802 0.0408 ms 65.0% triton_mm_2806 0.0508 ms 52.2% triton_mm_2793 0.0516 ms 51.5% triton_mm_2792 0.0553 ms 47.9% triton_mm_2797 0.0596 ms 44.5% triton_mm_2801 0.0620 ms 42.8% triton_mm_2791 0.0656 ms 40.4% SingleProcess AUTOTUNE benchmarking takes 2.1080 seconds and 0.0000 seconds precompiling AUTOTUNE mm(1088x6864, 6864x1088) mm 0.0975 ms 100.0% triton_mm_2937 0.0985 ms 99.0% triton_mm_2930 0.1044 ms 93.4% triton_mm_2936 0.1075 ms 90.7% triton_mm_2927 0.1187 ms 82.2% triton_mm_2928 0.1189 ms 82.0% triton_mm_2932 0.1204 ms 81.0% triton_mm_2935 0.1209 ms 80.7% triton_mm_2931 0.1262 ms 77.3% triton_mm_2933 0.1302 ms 74.9% SingleProcess AUTOTUNE benchmarking takes 2.4125 seconds and 0.0000 seconds precompiling AUTOTUNE mm(24x6528, 6528x1088) mm 0.0269 ms 100.0% triton_mm_2980 0.0275 ms 98.0% triton_mm_2984 0.0308 ms 87.6% triton_mm_2988 0.0400 ms 67.4% triton_mm_2992 0.0508 ms 53.0% triton_mm_2979 0.0522 ms 51.6% triton_mm_2978 0.0560 ms 48.1% triton_mm_2983 0.0596 ms 45.2% triton_mm_2987 0.0627 ms 43.0% triton_mm_2977 0.0650 ms 41.4% SingleProcess AUTOTUNE benchmarking takes 2.0806 seconds and 0.0000 seconds precompiling AUTOTUNE mm(1088x6840, 6840x96) mm 0.0316 ms 100.0% triton_mm_4874 0.0453 ms 69.8% triton_mm_4878 0.0527 ms 60.0% triton_mm_4872 0.0606 ms 52.2% triton_mm_4873 0.0625 ms 50.6% triton_mm_4871 0.0722 ms 43.8% triton_mm_4877 0.0731 ms 43.3% triton_mm_4882 0.0797 ms 39.7% triton_mm_4881 0.0811 ms 39.0% triton_mm_4880 0.0880 ms 35.9% SingleProcess AUTOTUNE benchmarking takes 9.5923 seconds and 0.0000 seconds precompiling /home/user/VAR/tokenizer/tokenizer_image/dino_enc/dinov2.py:122: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. with torch.cuda.amp.autocast(enabled=False): /home/user/VAR/tokenizer/tokenizer_image/dino_enc/dinov2.py:122: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. with torch.cuda.amp.autocast(enabled=False): /home/user/VAR/tokenizer/tokenizer_image/dino_enc/dinov2.py:122: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. with torch.cuda.amp.autocast(enabled=False): AUTOTUNE addmm(6840x1088, 6840x96, 96x1088) triton_mm_13 0.0364 ms 100.0% triton_mm_16 0.0378 ms 96.3% triton_mm_9 0.0386 ms 94.4% triton_mm_6 0.0420 ms 86.8% triton_mm_5 0.0423 ms 86.1% triton_mm_14 0.0427 ms 85.3% triton_mm_10 0.0430 ms 84.7% bias_addmm 0.0485 ms 75.1% triton_mm_15 0.0502 ms 72.6% triton_mm_11 0.0560 ms 65.0% SingleProcess AUTOTUNE benchmarking takes 2.2716 seconds and 0.0060 seconds precompiling AUTOTUNE addmm(24x6528, 24x1088, 1088x6528) triton_mm_31 0.0221 ms 100.0% triton_mm_35 0.0224 ms 98.6% triton_mm_26 0.0228 ms 96.6% triton_mm_27 0.0232 ms 95.4% triton_mm_20 0.0234 ms 94.5% triton_mm_23 0.0235 ms 93.9% triton_mm_30 0.0237 ms 93.1% bias_addmm 0.0249 ms 88.8% triton_mm_22 0.0251 ms 88.0% triton_mm_21 0.0252 ms 87.5% SingleProcess AUTOTUNE benchmarking takes 2.1919 seconds and 0.0008 seconds precompiling AUTOTUNE addmm(6864x3264, 6864x1088, 1088x3264) triton_mm_52 0.2238 ms 100.0% triton_mm_53 0.2305 ms 97.1% triton_mm_54 0.2580 ms 86.8% triton_mm_47 0.2686 ms 83.3% triton_mm_49 0.2764 ms 81.0% triton_mm_45 0.2796 ms 80.0% bias_addmm 0.2980 ms 75.1% triton_mm_46 0.3110 ms 72.0% triton_mm_50 0.3142 ms 71.2% addmm 0.3216 ms 69.6% SingleProcess AUTOTUNE benchmarking takes 2.6285 seconds and 0.0147 seconds precompiling AUTOTUNE addmm(6864x1088, 6864x1088, 1088x1088) triton_mm_72 0.0901 ms 100.0% triton_mm_71 0.0906 ms 99.5% bias_addmm 0.0963 ms 93.6% triton_mm_68 0.0984 ms 91.6% triton_mm_66 0.0987 ms 91.3% triton_mm_73 0.0991 ms 91.0% triton_mm_64 0.1021 ms 88.2% addmm 0.1107 ms 81.4% triton_mm_65 0.1116 ms 80.8% triton_mm_69 0.1122 ms 80.3% SingleProcess AUTOTUNE benchmarking takes 2.5132 seconds and 0.0010 seconds precompiling AUTOTUNE addmm(6864x4352, 6864x1088, 1088x4352) triton_mm_90 0.2934 ms 100.0% triton_mm_91 0.2980 ms 98.5% bias_addmm 0.3244 ms 90.5% triton_mm_92 0.3351 ms 87.6% triton_mm_85 0.3452 ms 85.0% triton_mm_83 0.3640 ms 80.6% triton_mm_87 0.3663 ms 80.1% triton_mm_84 0.4027 ms 72.9% triton_mm_88 0.4121 ms 71.2% addmm 0.4164 ms 70.5% SingleProcess AUTOTUNE benchmarking takes 2.6505 seconds and 0.0042 seconds precompiling AUTOTUNE addmm(6864x1088, 6864x4352, 4352x1088) bias_addmm 0.2986 ms 100.0% triton_mm_110 0.3096 ms 96.5% triton_mm_109 0.3178 ms 94.0% triton_mm_111 0.3282 ms 91.0% addmm 0.3348 ms 89.2% triton_mm_106 0.3607 ms 82.8% triton_mm_104 0.3632 ms 82.2% triton_mm_105 0.3739 ms 79.9% triton_mm_102 0.3777 ms 79.1% triton_mm_107 0.4109 ms 72.7% SingleProcess AUTOTUNE benchmarking takes 2.6537 seconds and 0.0010 seconds precompiling AUTOTUNE mm(24x1088, 1088x2176) triton_mm_1604 0.0133 ms 100.0% triton_mm_1608 0.0148 ms 90.2% triton_mm_1612 0.0149 ms 89.5% triton_mm_1616 0.0169 ms 78.9% triton_mm_1607 0.0174 ms 76.6% mm 0.0174 ms 76.5% triton_mm_1602 0.0175 ms 76.1% triton_mm_1603 0.0177 ms 75.2% triton_mm_1601 0.0184 ms 72.2% triton_mm_1611 0.0189 ms 70.3% SingleProcess AUTOTUNE benchmarking takes 7.5863 seconds and 0.0000 seconds precompiling AUTOTUNE addmm(6864x12288, 6864x1088, 1088x12288) bias_addmm 0.8293 ms 100.0% triton_mm_1634 0.8460 ms 98.0% triton_mm_1633 0.8524 ms 97.3% triton_mm_1635 0.9528 ms 87.0% triton_mm_1628 1.0055 ms 82.5% triton_mm_1630 1.0467 ms 79.2% addmm 1.0694 ms 77.6% triton_mm_1626 1.0816 ms 76.7% triton_mm_1631 1.1788 ms 70.4% triton_mm_1627 1.1929 ms 69.5% SingleProcess AUTOTUNE benchmarking takes 11.5155 seconds and 0.0000 seconds precompiling AUTOTUNE mm(12288x6864, 6864x1088) triton_mm_1671 0.7744 ms 100.0% triton_mm_1672 0.7999 ms 96.8% triton_mm_1673 0.8391 ms 92.3% mm 0.8669 ms 89.3% triton_mm_1668 0.9621 ms 80.5% triton_mm_1666 1.0054 ms 77.0% triton_mm_1664 1.0181 ms 76.1% triton_mm_1667 1.0712 ms 72.3% triton_mm_1669 1.0775 ms 71.9% triton_mm_1665 1.1059 ms 70.0% SingleProcess AUTOTUNE benchmarking takes 9.6100 seconds and 0.0000 seconds precompiling AUTOTUNE mm(1088x6864, 6864x4352) mm 0.2984 ms 100.0% triton_mm_1744 0.3007 ms 99.2% triton_mm_1743 0.3014 ms 99.0% triton_mm_1745 0.3071 ms 97.2% triton_mm_1738 0.3593 ms 83.0% triton_mm_1736 0.3614 ms 82.5% triton_mm_1740 0.3698 ms 80.7% triton_mm_1739 0.3854 ms 77.4% triton_mm_1741 0.4051 ms 73.7% triton_mm_1737 0.4103 ms 72.7% SingleProcess AUTOTUNE benchmarking takes 9.6825 seconds and 0.0000 seconds precompiling AUTOTUNE mm(2176x24, 24x1088) triton_mm_1700 0.0110 ms 100.0% triton_mm_1697 0.0113 ms 97.2% triton_mm_1703 0.0114 ms 96.3% triton_mm_1696 0.0114 ms 96.1% triton_mm_1698 0.0115 ms 95.8% triton_mm_1704 0.0115 ms 95.3% triton_mm_1702 0.0120 ms 91.2% triton_mm_1699 0.0121 ms 91.0% triton_mm_1701 0.0122 ms 89.8% triton_mm_1695 0.0123 ms 89.6% SingleProcess AUTOTUNE benchmarking takes 5.8981 seconds and 0.0000 seconds precompiling AUTOTUNE mm(24x2176, 2176x1088) triton_mm_1678 0.0165 ms 100.0% triton_mm_1682 0.0174 ms 94.9% mm 0.0176 ms 93.6% triton_mm_1686 0.0212 ms 77.8% triton_mm_1677 0.0223 ms 73.9% triton_mm_1690 0.0236 ms 69.9% triton_mm_1676 0.0241 ms 68.5% triton_mm_1681 0.0262 ms 63.1% triton_mm_1675 0.0264 ms 62.5% triton_mm_1685 0.0267 ms 61.9% SingleProcess AUTOTUNE benchmarking takes 2.1321 seconds and 0.0009 seconds precompiling AUTOTUNE mm(6864x12288, 12288x1088) mm 0.8572 ms 100.0% triton_mm_1652 0.8613 ms 99.5% triton_mm_1653 0.8776 ms 97.7% triton_mm_1654 0.9016 ms 95.1% triton_mm_1649 0.9885 ms 86.7% triton_mm_1645 1.0012 ms 85.6% triton_mm_1647 1.0091 ms 84.9% triton_mm_1648 1.0391 ms 82.5% triton_mm_1646 1.1056 ms 77.5% triton_mm_1650 1.1269 ms 76.1% SingleProcess AUTOTUNE benchmarking takes 2.6639 seconds and 0.0009 seconds precompiling AUTOTUNE mm(6864x1088, 1088x4352) triton_mm_1724 0.2811 ms 100.0% mm 0.2833 ms 99.2% triton_mm_1725 0.2836 ms 99.1% triton_mm_1726 0.3196 ms 88.0% triton_mm_1717 0.3434 ms 81.9% triton_mm_1719 0.3448 ms 81.5% triton_mm_1721 0.3516 ms 80.0% triton_mm_1718 0.3665 ms 76.7% triton_mm_1722 0.3897 ms 72.1% triton_mm_1720 0.4238 ms 66.3% SingleProcess AUTOTUNE benchmarking takes 2.5240 seconds and 0.0009 seconds precompiling AUTOTUNE mm(6864x3264, 3264x1088) triton_mm_1839 0.2268 ms 100.0% triton_mm_1838 0.2321 ms 97.7% triton_mm_1840 0.2436 ms 93.1% mm 0.2561 ms 88.5% triton_mm_1835 0.2600 ms 87.2% triton_mm_1833 0.2682 ms 84.6% triton_mm_1831 0.2700 ms 84.0% triton_mm_1834 0.2927 ms 77.5% triton_mm_1832 0.2958 ms 76.7% triton_mm_1836 0.2995 ms 75.7% SingleProcess AUTOTUNE benchmarking takes 2.4930 seconds and 0.0009 seconds precompiling AUTOTUNE mm(4352x6864, 6864x1088) triton_mm_1781 0.2929 ms 100.0% mm 0.2969 ms 98.6% triton_mm_1782 0.2985 ms 98.1% triton_mm_1783 0.3050 ms 96.0% triton_mm_1778 0.3519 ms 83.2% triton_mm_1774 0.3591 ms 81.6% triton_mm_1776 0.3633 ms 80.6% triton_mm_1777 0.3876 ms 75.5% triton_mm_1779 0.3980 ms 73.6% triton_mm_1775 0.4049 ms 72.3% SingleProcess AUTOTUNE benchmarking takes 10.2623 seconds and 0.0000 seconds precompiling AUTOTUNE mm(3264x6864, 6864x1088) mm 0.2289 ms 100.0% triton_mm_1857 0.2786 ms 82.2% triton_mm_1858 0.2855 ms 80.2% triton_mm_1852 0.2863 ms 79.9% triton_mm_1854 0.2891 ms 79.2% triton_mm_1850 0.2903 ms 78.9% triton_mm_1859 0.2907 ms 78.7% triton_mm_1853 0.3145 ms 72.8% triton_mm_1855 0.3271 ms 70.0% triton_mm_1851 0.3326 ms 68.8% SingleProcess AUTOTUNE benchmarking takes 2.5174 seconds and 0.0010 seconds precompiling AUTOTUNE mm(1088x6864, 6864x1088) mm 0.1007 ms 100.0% triton_mm_1814 0.1054 ms 95.5% triton_mm_1820 0.1073 ms 93.8% triton_mm_1812 0.1185 ms 85.0% triton_mm_1811 0.1191 ms 84.5% triton_mm_1821 0.1200 ms 83.9% triton_mm_1816 0.1201 ms 83.8% triton_mm_1819 0.1202 ms 83.8% triton_mm_1815 0.1260 ms 79.9% triton_mm_1817 0.1302 ms 77.3% SingleProcess AUTOTUNE benchmarking takes 9.8177 seconds and 0.0000 seconds precompiling AUTOTUNE mm(6864x4352, 4352x1088) triton_mm_1762 0.3055 ms 100.0% mm 0.3066 ms 99.6% triton_mm_1763 0.3134 ms 97.5% triton_mm_1764 0.3240 ms 94.3% triton_mm_1759 0.3496 ms 87.4% triton_mm_1757 0.3529 ms 86.6% triton_mm_1755 0.3566 ms 85.7% triton_mm_1758 0.3690 ms 82.8% triton_mm_1756 0.3892 ms 78.5% triton_mm_1760 0.3990 ms 76.6% SingleProcess AUTOTUNE benchmarking takes 2.5395 seconds and 0.0010 seconds precompiling AUTOTUNE mm(6864x1088, 1088x1088) triton_mm_1801 0.0843 ms 100.0% mm 0.0852 ms 98.9% triton_mm_1800 0.0858 ms 98.2% triton_mm_1797 0.0933 ms 90.3% triton_mm_1802 0.0942 ms 89.5% triton_mm_1793 0.0958 ms 87.9% triton_mm_1795 0.0964 ms 87.4% triton_mm_1794 0.1020 ms 82.6% triton_mm_1798 0.1052 ms 80.1% triton_mm_1796 0.1231 ms 68.5% SingleProcess AUTOTUNE benchmarking takes 2.4144 seconds and 0.0011 seconds precompiling AUTOTUNE mm(6864x1088, 1088x4352) triton_mm_1910 0.2805 ms 100.0% triton_mm_1911 0.2820 ms 99.5% mm 0.3179 ms 88.2% triton_mm_1912 0.3205 ms 87.5% triton_mm_1903 0.3434 ms 81.7% triton_mm_1905 0.3459 ms 81.1% triton_mm_1907 0.3520 ms 79.7% triton_mm_1904 0.3674 ms 76.3% triton_mm_1908 0.3893 ms 72.1% triton_mm_1906 0.4242 ms 66.1% SingleProcess AUTOTUNE benchmarking takes 2.5239 seconds and 0.0000 seconds precompiling AUTOTUNE mm(6528x24, 24x1088) triton_mm_1888 0.0173 ms 100.0% triton_mm_1883 0.0173 ms 99.8% triton_mm_1887 0.0180 ms 95.9% triton_mm_1889 0.0181 ms 95.5% triton_mm_1885 0.0182 ms 95.1% triton_mm_1886 0.0183 ms 94.4% triton_mm_1884 0.0184 ms 94.3% triton_mm_1882 0.0185 ms 93.4% triton_mm_1890 0.0190 ms 91.2% triton_mm_1891 0.0191 ms 90.5% SingleProcess AUTOTUNE benchmarking takes 2.1463 seconds and 0.0008 seconds precompiling AUTOTUNE mm(1088x6864, 6864x4352) mm 0.2982 ms 100.0% triton_mm_1930 0.3005 ms 99.2% triton_mm_1929 0.3013 ms 99.0% triton_mm_1931 0.3056 ms 97.6% triton_mm_1924 0.3576 ms 83.4% triton_mm_1922 0.3610 ms 82.6% triton_mm_1926 0.3697 ms 80.7% triton_mm_1925 0.3846 ms 77.6% triton_mm_1927 0.4044 ms 73.8% triton_mm_1923 0.4098 ms 72.8% SingleProcess AUTOTUNE benchmarking takes 2.5304 seconds and 0.0000 seconds precompiling AUTOTUNE mm(4352x6864, 6864x1088) triton_mm_1967 0.2925 ms 100.0% mm 0.2961 ms 98.8% triton_mm_1968 0.2980 ms 98.2% triton_mm_1969 0.3053 ms 95.8% triton_mm_1964 0.3513 ms 83.3% triton_mm_1960 0.3603 ms 81.2% triton_mm_1962 0.3612 ms 81.0% triton_mm_1963 0.3877 ms 75.4% triton_mm_1965 0.3980 ms 73.5% triton_mm_1961 0.4047 ms 72.3% SingleProcess AUTOTUNE benchmarking takes 2.5349 seconds and 0.0000 seconds precompiling AUTOTUNE mm(24x6528, 6528x1088) mm 0.0266 ms 100.0% triton_mm_1864 0.0267 ms 99.9% triton_mm_1868 0.0307 ms 86.8% triton_mm_1872 0.0409 ms 65.1% triton_mm_1876 0.0507 ms 52.5% triton_mm_1863 0.0520 ms 51.2% triton_mm_1862 0.0560 ms 47.5% triton_mm_1867 0.0602 ms 44.2% triton_mm_1871 0.0619 ms 43.0% triton_mm_1861 0.0650 ms 41.0% SingleProcess AUTOTUNE benchmarking takes 6.5877 seconds and 0.0000 seconds precompiling AUTOTUNE mm(3264x6864, 6864x1088) mm 0.2281 ms 100.0% triton_mm_2043 0.2791 ms 81.7% triton_mm_2044 0.2863 ms 79.7% triton_mm_2038 0.2866 ms 79.6% triton_mm_2040 0.2866 ms 79.6% triton_mm_2045 0.2896 ms 78.8% triton_mm_2036 0.2900 ms 78.6% triton_mm_2039 0.3146 ms 72.5% triton_mm_2041 0.3275 ms 69.6% triton_mm_2037 0.3329 ms 68.5% SingleProcess AUTOTUNE benchmarking takes 2.5117 seconds and 0.0000 seconds precompiling AUTOTUNE mm(1088x6864, 6864x1088) triton_mm_2007 0.0991 ms 100.0% mm 0.1008 ms 98.3% triton_mm_2000 0.1044 ms 94.9% triton_mm_2006 0.1080 ms 91.8% triton_mm_1998 0.1180 ms 84.0% triton_mm_1997 0.1189 ms 83.3% triton_mm_2002 0.1202 ms 82.5% triton_mm_2005 0.1208 ms 82.0% triton_mm_2001 0.1260 ms 78.6% triton_mm_2003 0.1301 ms 76.2% SingleProcess AUTOTUNE benchmarking takes 2.4176 seconds and 0.0000 seconds precompiling AUTOTUNE mm(6864x1088, 1088x1088) triton_mm_1987 0.0836 ms 100.0% mm 0.0836 ms 99.9% triton_mm_1986 0.0852 ms 98.0% triton_mm_1983 0.0926 ms 90.3% triton_mm_1988 0.0950 ms 88.0% triton_mm_1981 0.0956 ms 87.4% triton_mm_1979 0.0959 ms 87.2% triton_mm_1980 0.1021 ms 81.9% triton_mm_1984 0.1059 ms 78.9% triton_mm_1982 0.1238 ms 67.5% SingleProcess AUTOTUNE benchmarking takes 2.4127 seconds and 0.0000 seconds precompiling AUTOTUNE mm(6864x3264, 3264x1088) triton_mm_2025 0.2262 ms 100.0% mm 0.2267 ms 99.8% triton_mm_2024 0.2322 ms 97.4% triton_mm_2026 0.2450 ms 92.4% triton_mm_2021 0.2612 ms 86.6% triton_mm_2019 0.2660 ms 85.1% triton_mm_2017 0.2673 ms 84.7% triton_mm_2020 0.2904 ms 77.9% triton_mm_2018 0.2940 ms 76.9% triton_mm_2022 0.3006 ms 75.3% SingleProcess AUTOTUNE benchmarking takes 2.5092 seconds and 0.0000 seconds precompiling AUTOTUNE mm(6864x1088, 1088x4352) triton_mm_2096 0.2814 ms 100.0% triton_mm_2097 0.2838 ms 99.2% mm 0.2846 ms 98.9% triton_mm_2098 0.3200 ms 87.9% triton_mm_2089 0.3416 ms 82.4% triton_mm_2091 0.3458 ms 81.4% triton_mm_2093 0.3518 ms 80.0% triton_mm_2090 0.3680 ms 76.5% triton_mm_2094 0.3900 ms 72.2% triton_mm_2092 0.4249 ms 66.2% SingleProcess AUTOTUNE benchmarking takes 2.5382 seconds and 0.0000 seconds precompiling AUTOTUNE mm(6528x24, 24x1088) triton_mm_2069 0.0174 ms 100.0% triton_mm_2073 0.0181 ms 96.1% triton_mm_2071 0.0182 ms 95.8% triton_mm_2075 0.0182 ms 95.6% triton_mm_2072 0.0183 ms 94.9% triton_mm_2074 0.0184 ms 94.8% triton_mm_2070 0.0184 ms 94.6% triton_mm_2068 0.0186 ms 93.8% triton_mm_2078 0.0186 ms 93.6% triton_mm_2076 0.0191 ms 91.1% SingleProcess AUTOTUNE benchmarking takes 2.1474 seconds and 0.0000 seconds precompiling AUTOTUNE mm(1088x6864, 6864x4352) mm 0.2951 ms 100.0% triton_mm_2116 0.3003 ms 98.3% triton_mm_2115 0.3011 ms 98.0% triton_mm_2117 0.3059 ms 96.5% triton_mm_2110 0.3582 ms 82.4% triton_mm_2108 0.3607 ms 81.8% triton_mm_2112 0.3695 ms 79.9% triton_mm_2111 0.3848 ms 76.7% triton_mm_2113 0.4050 ms 72.9% triton_mm_2109 0.4092 ms 72.1% SingleProcess AUTOTUNE benchmarking takes 2.5389 seconds and 0.0000 seconds precompiling AUTOTUNE mm(4352x6864, 6864x1088) triton_mm_2153 0.2925 ms 100.0% mm 0.2955 ms 99.0% triton_mm_2154 0.2975 ms 98.3% triton_mm_2155 0.3044 ms 96.1% triton_mm_2150 0.3527 ms 82.9% triton_mm_2146 0.3600 ms 81.3% triton_mm_2148 0.3627 ms 80.6% triton_mm_2149 0.3912 ms 74.8% triton_mm_2151 0.3980 ms 73.5% triton_mm_2147 0.4046 ms 72.3% SingleProcess AUTOTUNE benchmarking takes 2.5376 seconds and 0.0000 seconds precompiling AUTOTUNE mm(24x6528, 6528x1088) triton_mm_2050 0.0267 ms 100.0% mm 0.0268 ms 99.4% triton_mm_2054 0.0307 ms 87.0% triton_mm_2058 0.0399 ms 66.9% triton_mm_2062 0.0508 ms 52.6% triton_mm_2049 0.0515 ms 51.9% triton_mm_2048 0.0554 ms 48.2% triton_mm_2053 0.0602 ms 44.3% triton_mm_2057 0.0626 ms 42.7% triton_mm_2047 0.0656 ms 40.7% SingleProcess AUTOTUNE benchmarking takes 2.0969 seconds and 0.0000 seconds precompiling AUTOTUNE mm(3264x6864, 6864x1088) mm 0.2285 ms 100.0% triton_mm_2229 0.2788 ms 81.9% triton_mm_2230 0.2869 ms 79.6% triton_mm_2224 0.2879 ms 79.3% triton_mm_2226 0.2888 ms 79.1% triton_mm_2222 0.2902 ms 78.7% triton_mm_2231 0.2912 ms 78.4% triton_mm_2225 0.3146 ms 72.6% triton_mm_2227 0.3275 ms 69.8% triton_mm_2223 0.3325 ms 68.7% SingleProcess AUTOTUNE benchmarking takes 2.5152 seconds and 0.0000 seconds precompiling AUTOTUNE mm(1088x6864, 6864x1088) triton_mm_2193 0.0982 ms 100.0% mm 0.1005 ms 97.8% triton_mm_2186 0.1045 ms 94.0% triton_mm_2192 0.1074 ms 91.5% triton_mm_2184 0.1180 ms 83.3% triton_mm_2183 0.1190 ms 82.5% triton_mm_2188 0.1201 ms 81.8% triton_mm_2191 0.1203 ms 81.7% triton_mm_2187 0.1261 ms 77.9% triton_mm_2189 0.1301 ms 75.5% SingleProcess AUTOTUNE benchmarking takes 2.4245 seconds and 0.0000 seconds precompiling AUTOTUNE mm(6864x1088, 1088x1088) triton_mm_2173 0.0836 ms 100.0% mm 0.0844 ms 99.0% triton_mm_2172 0.0852 ms 98.0% triton_mm_2169 0.0927 ms 90.2% triton_mm_2174 0.0949 ms 88.1% triton_mm_2167 0.0956 ms 87.4% triton_mm_2165 0.0958 ms 87.2% triton_mm_2166 0.1021 ms 81.9% triton_mm_2170 0.1060 ms 78.9% triton_mm_2168 0.1237 ms 67.5% SingleProcess AUTOTUNE benchmarking takes 2.4199 seconds and 0.0000 seconds precompiling AUTOTUNE mm(6864x3264, 3264x1088) mm 0.2271 ms 100.0% triton_mm_2211 0.2273 ms 99.9% triton_mm_2210 0.2326 ms 97.6% triton_mm_2212 0.2447 ms 92.8% triton_mm_2207 0.2620 ms 86.7% triton_mm_2205 0.2663 ms 85.3% triton_mm_2203 0.2669 ms 85.1% triton_mm_2206 0.2906 ms 78.2% triton_mm_2204 0.2938 ms 77.3% triton_mm_2208 0.3007 ms 75.5% SingleProcess AUTOTUNE benchmarking takes 2.5080 seconds and 0.0000 seconds precompiling AUTOTUNE mm(6864x1088, 1088x4352) triton_mm_2282 0.2806 ms 100.0% triton_mm_2283 0.2832 ms 99.1% mm 0.2839 ms 98.8% triton_mm_2284 0.3216 ms 87.2% triton_mm_2275 0.3449 ms 81.3% triton_mm_2277 0.3461 ms 81.1% triton_mm_2279 0.3527 ms 79.6% triton_mm_2276 0.3680 ms 76.3% triton_mm_2280 0.3894 ms 72.0% triton_mm_2278 0.4245 ms 66.1% SingleProcess AUTOTUNE benchmarking takes 2.5314 seconds and 0.0000 seconds precompiling AUTOTUNE mm(6528x24, 24x1088) triton_mm_2260 0.0173 ms 100.0% triton_mm_2255 0.0174 ms 99.8% triton_mm_2257 0.0174 ms 99.8% triton_mm_2254 0.0179 ms 97.1% triton_mm_2259 0.0181 ms 95.8% triton_mm_2261 0.0182 ms 95.4% triton_mm_2262 0.0182 ms 95.1% triton_mm_2258 0.0184 ms 94.4% triton_mm_2264 0.0186 ms 93.3% triton_mm_2256 0.0192 ms 90.5% SingleProcess AUTOTUNE benchmarking takes 2.1495 seconds and 0.0000 seconds precompiling AUTOTUNE mm(1088x6864, 6864x4352) mm 0.2944 ms 100.0% triton_mm_2302 0.2996 ms 98.2% triton_mm_2301 0.3013 ms 97.7% triton_mm_2303 0.3062 ms 96.1% triton_mm_2296 0.3589 ms 82.0% triton_mm_2294 0.3606 ms 81.6% triton_mm_2298 0.3692 ms 79.7% triton_mm_2297 0.3851 ms 76.4% triton_mm_2299 0.4042 ms 72.8% triton_mm_2295 0.4104 ms 71.7% SingleProcess AUTOTUNE benchmarking takes 2.5384 seconds and 0.0000 seconds precompiling AUTOTUNE mm(4352x6864, 6864x1088) triton_mm_2339 0.2932 ms 100.0% mm 0.2964 ms 99.0% triton_mm_2340 0.2976 ms 98.6% triton_mm_2341 0.3041 ms 96.4% triton_mm_2336 0.3513 ms 83.5% triton_mm_2332 0.3597 ms 81.5% triton_mm_2334 0.3625 ms 80.9% triton_mm_2335 0.3882 ms 75.5% triton_mm_2337 0.3984 ms 73.6% triton_mm_2333 0.4062 ms 72.2% SingleProcess AUTOTUNE benchmarking takes 2.5342 seconds and 0.0000 seconds precompiling AUTOTUNE mm(24x6528, 6528x1088) triton_mm_2236 0.0267 ms 100.0% mm 0.0268 ms 99.6% triton_mm_2240 0.0317 ms 84.2% triton_mm_2244 0.0399 ms 66.8% triton_mm_2235 0.0515 ms 51.8% triton_mm_2248 0.0517 ms 51.6% triton_mm_2234 0.0554 ms 48.2% triton_mm_2239 0.0603 ms 44.3% triton_mm_2243 0.0619 ms 43.1% triton_mm_2233 0.0656 ms 40.7% SingleProcess AUTOTUNE benchmarking takes 2.0959 seconds and 0.0003 seconds precompiling AUTOTUNE mm(24x6528, 6528x1088) mm 0.0266 ms 100.0% triton_mm_2422 0.0267 ms 99.8% triton_mm_2426 0.0307 ms 86.8% triton_mm_2430 0.0399 ms 66.7% triton_mm_2421 0.0514 ms 51.8% triton_mm_2434 0.0515 ms 51.7% triton_mm_2420 0.0554 ms 48.1% triton_mm_2425 0.0596 ms 44.7% triton_mm_2429 0.0619 ms 43.0% triton_mm_2419 0.0656 ms 40.6% SingleProcess AUTOTUNE benchmarking takes 2.0927 seconds and 0.0000 seconds precompiling AUTOTUNE mm(24x6528, 6528x1088) mm 0.0267 ms 100.0% triton_mm_2608 0.0267 ms 100.0% triton_mm_2612 0.0307 ms 87.1% triton_mm_2616 0.0399 ms 66.9% triton_mm_2607 0.0515 ms 51.9% triton_mm_2620 0.0517 ms 51.6% triton_mm_2606 0.0554 ms 48.2% triton_mm_2611 0.0601 ms 44.4% triton_mm_2615 0.0625 ms 42.7% triton_mm_2605 0.0655 ms 40.7% SingleProcess AUTOTUNE benchmarking takes 2.0849 seconds and 0.0000 seconds precompiling AUTOTUNE mm(24x6528, 6528x1088) mm 0.0266 ms 100.0% triton_mm_2794 0.0267 ms 99.6% triton_mm_2798 0.0307 ms 86.7% triton_mm_2802 0.0399 ms 66.7% triton_mm_2793 0.0515 ms 51.7% triton_mm_2806 0.0516 ms 51.6% triton_mm_2792 0.0560 ms 47.5% triton_mm_2797 0.0596 ms 44.7% triton_mm_2801 0.0619 ms 43.0% triton_mm_2791 0.0650 ms 40.9% SingleProcess AUTOTUNE benchmarking takes 2.0847 seconds and 0.0000 seconds precompiling AUTOTUNE mm(24x6528, 6528x1088) mm 0.0265 ms 100.0% triton_mm_2980 0.0276 ms 96.2% triton_mm_2984 0.0317 ms 83.6% triton_mm_2988 0.0399 ms 66.5% triton_mm_2992 0.0515 ms 51.4% triton_mm_2979 0.0521 ms 50.9% triton_mm_2978 0.0561 ms 47.3% triton_mm_2983 0.0595 ms 44.5% triton_mm_2987 0.0625 ms 42.4% triton_mm_2977 0.0656 ms 40.4% SingleProcess AUTOTUNE benchmarking takes 2.0943 seconds and 0.0000 seconds precompiling AUTOTUNE mm(24x6528, 6528x1088) triton_mm_3166 0.0267 ms 100.0% mm 0.0267 ms 99.8% triton_mm_3170 0.0316 ms 84.3% triton_mm_3174 0.0399 ms 66.7% triton_mm_3178 0.0517 ms 51.5% triton_mm_3165 0.0522 ms 51.1% triton_mm_3164 0.0553 ms 48.2% triton_mm_3169 0.0596 ms 44.8% triton_mm_3173 0.0619 ms 43.1% triton_mm_3163 0.0649 ms 41.1% SingleProcess AUTOTUNE benchmarking takes 2.0923 seconds and 0.0000 seconds precompiling AUTOTUNE mm(24x6528, 6528x1088) mm 0.0265 ms 100.0% triton_mm_3538 0.0276 ms 96.2% triton_mm_3542 0.0317 ms 83.6% triton_mm_3546 0.0409 ms 64.8% triton_mm_3550 0.0512 ms 51.8% triton_mm_3537 0.0521 ms 50.9% triton_mm_3536 0.0561 ms 47.3% triton_mm_3541 0.0602 ms 44.0% triton_mm_3545 0.0626 ms 42.4% triton_mm_3535 0.0649 ms 40.8% SingleProcess AUTOTUNE benchmarking takes 2.0841 seconds and 0.0000 seconds precompiling AUTOTUNE mm(1088x6840, 6840x96) mm 0.0328 ms 100.0% triton_mm_4874 0.0452 ms 72.4% triton_mm_4878 0.0537 ms 61.0% triton_mm_4872 0.0607 ms 54.0% triton_mm_4873 0.0623 ms 52.6% triton_mm_4871 0.0718 ms 45.6% triton_mm_4877 0.0737 ms 44.5% triton_mm_4882 0.0793 ms 41.3% triton_mm_4881 0.0817 ms 40.1% triton_mm_4880 0.0885 ms 37.0% SingleProcess AUTOTUNE benchmarking takes 2.3468 seconds and 0.1488 seconds precompiling /home/user/VAR/tokenizer/tokenizer_image/dino_enc/dinov2.py:122: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. with torch.cuda.amp.autocast(enabled=False): /home/user/VAR/tokenizer/tokenizer_image/dino_enc/dinov2.py:122: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead. with torch.cuda.amp.autocast(enabled=False): Validation Progress: 0%| | 0/44 [00:00