version | segs | threads | input pixel size | output pixel size | n params | ms per example | ns per pixel bit | ns per channel | ns per parameter |
---|---|---|---|---|---|---|---|---|---|
3 | 2 | 16 | 32 | 32 | 9216 | 32.938 | 1030182.004 | 1030182.004 | 3577.021 |
3 | 2 | 16 | 64 | 32 | 18432 | 64.750 | 1011854.566 | 2023709.133 | 3513.384 |
3 | 2 | 16 | 128 | 32 | 36864 | 129.938 | 1015413.960 | 4061655.842 | 3525.743 |
3 | 2 | 16 | 256 | 32 | 73728 | 277.000 | 1082303.786 | 8658430.289 | 3757.999 |
3 | 2 | 16 | 512 | 32 | 147456 | 568.000 | 1109743.197 | 17755891.156 | 3853.275 |
3 | 2 | 16 | 1024 | 32 | 294912 | 1125.500 | 1099288.490 | 35177231.688 | 3816.974 |
5 | 2 | 16 | 32 | 32 | 9216 | 10.312 | 323690.141 | 323690.141 | 1123.924 |
5 | 2 | 16 | 64 | 32 | 18432 | 15.875 | 248133.127 | 496266.254 | 861.573 |
5 | 2 | 16 | 128 | 32 | 36864 | 29.500 | 230856.398 | 923425.594 | 801.585 |
5 | 2 | 16 | 256 | 32 | 73728 | 57.750 | 225611.752 | 1804894.020 | 783.374 |
5 | 2 | 16 | 512 | 32 | 147456 | 149.250 | 291540.294 | 4664644.703 | 1012.293 |
5 | 2 | 16 | 1024 | 32 | 294912 | 492.500 | 481434.008 | 15405888.266 | 1671.646 |
version | segs | threads | input pixel size | output pixel size | n params | ms per example | ns per pixel bit | ns per channel | ns per parameter |
---|---|---|---|---|---|---|---|---|---|
3 | 2 | 16 | 32 | 32 | 9216 | 31.625 | 988877.023 | 988877.023 | 3433.601 |
3 | 2 | 16 | 64 | 32 | 18432 | 59.000 | 921982.258 | 1843964.516 | 3201.327 |
3 | 2 | 16 | 128 | 32 | 36864 | 117.250 | 916140.265 | 3664561.061 | 3181.043 |
3 | 2 | 16 | 256 | 32 | 73728 | 260.750 | 1019021.881 | 8152175.047 | 3538.270 |
3 | 2 | 16 | 512 | 32 | 147456 | 571.250 | 1115857.159 | 17853714.539 | 3874.504 |
3 | 2 | 16 | 1024 | 32 | 294912 | 1033.500 | 1009677.716 | 32309686.922 | 3505.825 |
5 | 2 | 16 | 32 | 32 | 9216 | 10.062 | 315686.988 | 315686.988 | 1096.135 |
5 | 2 | 16 | 64 | 32 | 18432 | 19.625 | 307469.552 | 614939.104 | 1067.603 |
5 | 2 | 16 | 128 | 32 | 36864 | 50.000 | 391013.694 | 1564054.775 | 1357.686 |
5 | 2 | 16 | 256 | 32 | 73728 | 231.000 | 902370.037 | 7218960.293 | 3133.229 |
5 | 2 | 16 | 512 | 32 | 147456 | 566.500 | 1106640.410 | 17706246.555 | 3842.501 |
5 | 2 | 16 | 1024 | 32 | 294912 | 4454.000 | 4349625.233 | 139188007.453 | 15102.865 |
version | segs | threads | input pixel size | output pixel size | n params | ms per example | ns per pixel bit | ns per channel | ns per parameter |
---|---|---|---|---|---|---|---|---|---|
3 | 2 | 16 | 32 | 32 | 9216 | 9.500 | 297792.469 | 297792.469 | 1034.002 |
3 | 2 | 16 | 64 | 32 | 18432 | 16.812 | 262990.287 | 525980.574 | 913.161 |
3 | 2 | 16 | 128 | 32 | 36864 | 33.812 | 264188.154 | 1056752.617 | 917.320 |
3 | 2 | 16 | 256 | 32 | 73728 | 79.250 | 309573.160 | 2476585.281 | 1074.907 |
3 | 2 | 16 | 512 | 32 | 147456 | 163.000 | 318630.125 | 5098082.000 | 1106.355 |
3 | 2 | 16 | 1024 | 32 | 294912 | 333.000 | 325208.355 | 10406667.359 | 1129.196 |
5 | 2 | 16 | 32 | 32 | 9216 | 7.938 | 249043.172 | 249043.172 | 864.733 |
5 | 2 | 16 | 64 | 32 | 18432 | 12.250 | 192092.760 | 384185.520 | 666.989 |
5 | 2 | 16 | 128 | 32 | 36864 | 20.812 | 163036.229 | 652144.916 | 566.098 |
5 | 2 | 16 | 256 | 32 | 73728 | 38.500 | 150683.047 | 1205464.375 | 523.205 |
5 | 2 | 16 | 512 | 32 | 147456 | 111.000 | 216949.456 | 3471191.297 | 753.297 |
5 | 2 | 16 | 1024 | 32 | 294912 | 449.500 | 439156.565 | 14053010.094 | 1524.849 |
version | segs | threads | input pixel size | output pixel size | n params | ms per example | ns per pixel bit | ns per channel | ns per parameter |
---|---|---|---|---|---|---|---|---|---|
3 | 2 | 16 | 32 | 32 | 9216 | 8.000 | 251794.223 | 251794.223 | 874.285 |
3 | 2 | 16 | 64 | 32 | 18432 | 14.250 | 222984.823 | 445969.646 | 774.253 |
3 | 2 | 16 | 128 | 32 | 36864 | 29.062 | 227464.340 | 909857.361 | 789.807 |
3 | 2 | 16 | 256 | 32 | 73728 | 61.625 | 241207.314 | 1929658.512 | 837.525 |
3 | 2 | 16 | 512 | 32 | 147456 | 142.000 | 277431.813 | 4438909.008 | 963.305 |
3 | 2 | 16 | 1024 | 32 | 294912 | 269.000 | 262829.659 | 8410549.078 | 912.603 |
5 | 2 | 16 | 32 | 32 | 9216 | 6.875 | 215699.184 | 215699.184 | 748.955 |
5 | 2 | 16 | 64 | 32 | 18432 | 15.500 | 242780.401 | 485560.803 | 842.988 |
5 | 2 | 16 | 128 | 32 | 36864 | 40.938 | 320035.783 | 1280143.131 | 1111.235 |
5 | 2 | 16 | 256 | 32 | 73728 | 208.625 | 815322.578 | 6522580.621 | 2830.981 |
5 | 2 | 16 | 512 | 32 | 147456 | 517.000 | 1010128.052 | 16162048.828 | 3507.389 |
5 | 2 | 16 | 1024 | 32 | 294912 | 4033.000 | 3938599.417 | 126035181.328 | 13675.692 |
version | segs | threads | input pixel size | output pixel size | n params | ms per example | ns per pixel bit | ns per channel | ns per parameter |
---|---|---|---|---|---|---|---|---|---|
3 | 2 | 16 | 32 | 32 | 9216 | 32.688 | 1023065.971 | 1023065.971 | 3552.312 |
3 | 2 | 16 | 64 | 32 | 18432 | 61.938 | 968498.838 | 1936997.676 | 3362.843 |
3 | 2 | 16 | 128 | 32 | 36864 | 129.312 | 1010397.063 | 4041588.252 | 3508.323 |
3 | 2 | 16 | 256 | 32 | 73728 | 261.500 | 1021857.097 | 8174856.777 | 3548.115 |
3 | 2 | 16 | 512 | 32 | 147456 | 527.750 | 1031084.930 | 16497358.875 | 3580.156 |
3 | 2 | 16 | 1024 | 32 | 294912 | 1043.500 | 1019264.069 | 32616450.219 | 3539.111 |
5 | 2 | 16 | 32 | 32 | 9216 | 10.875 | 341049.057 | 341049.057 | 1184.198 |
5 | 2 | 16 | 64 | 32 | 18432 | 16.812 | 263036.051 | 526072.102 | 913.320 |
5 | 2 | 16 | 128 | 32 | 36864 | 28.562 | 223549.749 | 894198.996 | 776.214 |
5 | 2 | 16 | 256 | 32 | 73728 | 57.250 | 224051.503 | 1792412.027 | 777.957 |
5 | 2 | 16 | 512 | 32 | 147456 | 148.750 | 290771.451 | 4652343.219 | 1009.623 |
5 | 2 | 16 | 1024 | 32 | 294912 | 466.500 | 455779.345 | 14584939.031 | 1582.567 |
version | segs | threads | input pixel size | output pixel size | n params | ms per example | ns per pixel bit | ns per channel | ns per parameter |
---|---|---|---|---|---|---|---|---|---|
3 | 2 | 16 | 32 | 32 | 9216 | 8.688 | 272522.389 | 272522.389 | 946.258 |
3 | 2 | 16 | 64 | 32 | 18432 | 14.000 | 219625.833 | 439251.666 | 762.590 |
3 | 2 | 16 | 128 | 32 | 36864 | 28.000 | 219228.379 | 876913.518 | 761.210 |
3 | 2 | 16 | 256 | 32 | 73728 | 61.750 | 241528.711 | 1932229.691 | 838.641 |
3 | 2 | 16 | 512 | 32 | 147456 | 128.250 | 250879.322 | 4014069.156 | 871.109 |
3 | 2 | 16 | 1024 | 32 | 294912 | 306.500 | 299331.826 | 9578618.422 | 1039.347 |
5 | 2 | 16 | 32 | 32 | 9216 | 7.625 | 239311.895 | 239311.895 | 830.944 |
5 | 2 | 16 | 64 | 32 | 18432 | 16.625 | 260267.513 | 520535.025 | 903.707 |
5 | 2 | 16 | 128 | 32 | 36864 | 43.625 | 341027.564 | 1364110.256 | 1184.123 |
5 | 2 | 16 | 256 | 32 | 73728 | 211.625 | 827053.178 | 6616425.422 | 2871.712 |
5 | 2 | 16 | 512 | 32 | 147456 | 478.750 | 935470.216 | 14967523.461 | 3248.160 |
5 | 2 | 16 | 1024 | 32 | 294912 | 3804.500 | 3715350.809 | 118891225.891 | 12900.524 |
version | segs | threads | input pixel size | output pixel size | n params | ms per example | ns per pixel bit | ns per channel | ns per parameter |
---|---|---|---|---|---|---|---|---|---|
3 | 2 | 1 | 32 | 32 | 9216 | 28.441 | 888814.643 | 888814.643 | 3086.162 |
3 | 2 | 1 | 64 | 32 | 18432 | 54.285 | 848228.163 | 1696456.327 | 2945.237 |
3 | 2 | 1 | 128 | 32 | 36864 | 107.711 | 841520.823 | 3366083.293 | 2921.947 |
3 | 2 | 1 | 256 | 32 | 73728 | 215.102 | 840262.276 | 6722098.208 | 2917.577 |
3 | 2 | 1 | 512 | 32 | 147456 | 432.703 | 845142.628 | 13522282.044 | 2934.523 |
3 | 2 | 1 | 1024 | 32 | 294912 | 857.906 | 837800.704 | 26809622.517 | 2909.030 |
5 | 2 | 1 | 32 | 32 | 9216 | 8.453 | 264163.800 | 264163.800 | 917.235 |
5 | 2 | 1 | 64 | 32 | 18432 | 14.371 | 224579.332 | 449158.665 | 779.789 |
5 | 2 | 1 | 128 | 32 | 36864 | 25.227 | 197086.169 | 788344.677 | 684.327 |
5 | 2 | 1 | 256 | 32 | 73728 | 46.281 | 180809.993 | 1446479.948 | 627.812 |
5 | 2 | 1 | 512 | 32 | 147456 | 95.156 | 185866.345 | 2973861.518 | 645.369 |
5 | 2 | 1 | 1024 | 32 | 294912 | 185.500 | 181181.513 | 5797808.405 | 629.102 |