bartekkuncer opened a new pull request #19561: URL: https://github.com/apache/incubator-mxnet/pull/19561
Opperf results: lhs | rhs | no change | change | time difference in % -- | -- | -- | -- | -- 1000x3000 | 3000x2000 | 3,7838 | 3,906 | 3,229557588 2560x1000 | 1000x1000 | 1,5147 | 1,6312 | 7,691292005 2048x2048 | 2048x4000 | 10,0919 | 9,6068 | -4,806825276 1500x20000 | 20000x3000 | 56,8936 | 58,9078 | 3,540292757 6900x2000 | 2000x3000 | 24,1915 | 21,7022 | -10,28997788 1040x3000 | 3000x3000 | 6,7088 | 6,5443 | -2,452003339 2048x1000 | 1000x3000 | 3,6074 | 3,5855 | -0,607085435 7020x4500 | 4500x300 | 5,9606 | 6,3557 | 6,628527329 3000x1000 | 1000x3000 | 5,407 | 5,7826 | 6,946550768 400x3000 | 3000x300 | 0,3275 | 0,4734 | 44,54961832 100x2000 | 2000x300 | 0,0949 | 0,0975 | 2,739726027 3120x300 | 300x18720 | 11,1002 | 11,72 | 5,583683177 2400x1000 | 1000x18720 | 26,6274 | 24,5511 | -7,797606976 5400x2000 | 2000x1872 | 10,908 | 10,4899 | -3,83296663 2496x4500 | 4500x1800 | 13,2535 | 12,1826 | -8,080129777 1000x3 | 3x200 | 0,0345 | 0,0361 | 4,637681159 500x512x10 | 10x100x10 | 14,8204 | 14,58 | -1,622088473 100x2048x3 | 3x20x200 | 46,6874 | 40,927 | -12,33823259 1000x5x30x200 | 200x5x30x20 | 47,6729 | 46,3023 | -2,875008653 30x512x45x20 | 20x5x30x20 | 114,2006 | 114,5709 | 0,324253988 50x52x40x30 | 30x5x30x20 | 13,0779 | 13,0464 | -0,240864359 10x248x10x10 | 10x5x30x20 | 5,7018 | 6,083 | 6,685608054 45x52x30x45 | 45x5x30x20 | 9,1216 | 9,0936 | -0,306963691 10x5x30x20x10 | 10x5x30x20x10 | 62,8782 | 62,1395 | -1,174810984 10x4x10x10x30 | 30x5x30x20x10 | 5,3065 | 5,4481 | 2,668425516 3x52x2x45x20 | 20x5x30x20x10 | 31,8084 | 30,4628 | -4,230329095 50x52x3x4x30 | 30x52x3x4x30 | 25,0933 | 24,8697 | -0,89107451 10x8x3x10x10 | 10x52x3x4x30 | 3,066 | 3,1714 | 3,437703849 10x3x10x18x20 | 20x52x3x4x30 | 7,7454 | 7,992 | 3,183825238 4x52x3x4x45 | 45x5x3x40x30 | 3,2964 | 3,2558 | -1,231646645 1000x3 | 3x200 | 0,0331 | 0,0397 | 19,93957704 256000x10 | 10x1000 | 19,9643 | 14,6538 | -26,59998097 204800x3 | 3x4000 | 58,8355 | 41,2864 | -29,82740012 150000x200 | 200x3000 | 47,306 | 46,1426 | -2,459307487 691200x20 | 20x3000 | 114,2128 | 114,6147 | 0,351887004 104000x30 | 30x3000 | 13,244 | 12,9124 | -2,503775294 24800x10 | 10x3000 | 5,7193 | 6,1124 | 6,873218751 70200x45 | 45x3000 | 9,1538 | 9,0278 | -1,376477528 30000x10 | 10x30000 | 62,7901 | 62,5743 | -0,343684753 4000x30 | 30x30000 | 5,322 | 5,4248 | 1,93160466 14040x20 | 20x30000 | 31,7733 | 30,5213 | -3,94041538 31200x30 | 30x18720 | 25,1541 | 24,7054 | -1,783804628 2400x10 | 10x18720 | 3,0625 | 3,1734 | 3,62122449 5400x20 | 20x18720 | 7,7275 | 7,9511 | 2,893561954 2496x45 | 45x18000 | 3,2888 | 3,2345 | -1,651058137 ---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: [email protected]
