23/09/26 18:25:57.43 jbwpiF7D.net
>>334 単精度アップデート(倍精度は手つかず)
今度はMKLをちゃんと動かせた
ST f32 1024x1024
my-version BLIS OpenBLAS MKL
17.344ms 123.816 GFLOPS 15.241ms 140.902 GFLOPS 15.166ms 141.598 GFLOPS 14.629ms 146.792 GFLOPS
17.338ms 123.860 GFLOPS 15.217ms 141.119 GFLOPS 15.167ms 141.592 GFLOPS 14.681ms 146.281 GFLOPS
17.805ms 120.610 GFLOPS 15.237ms 140.939 GFLOPS 15.258ms 140.741 GFLOPS 14.843ms 144.678 GFLOPS
17.808ms 120.590 GFLOPS 15.272ms 140.611 GFLOPS 15.186ms 141.410 GFLOPS 14.667ms 146.417 GFLOPS
17.478ms 122.871 GFLOPS 15.310ms 140.264 GFLOPS 15.195ms 141.328 GFLOPS 14.674ms 146.346 GFLOPS
MT 8 f32 1024x1024
2.401ms 894.478 GFLOPS 2.580ms 832.272 GFLOPS 2.514ms 854.201 GFLOPS 2.104ms 1020.428 GFLOPS
2.379ms 902.518 GFLOPS 2.483ms 864.862 GFLOPS 2.702ms 794.853 GFLOPS 2.159ms 994.802 GFLOPS
2.430ms 883.801 GFLOPS 2.539ms 845.796 GFLOPS 2.716ms 790.540 GFLOPS 2.187ms 982.003 GFLOPS
2.407ms 892.147 GFLOPS 2.438ms 880.740 GFLOPS 2.714ms 791.168 GFLOPS 2.160ms 994.223 GFLOPS
2.457ms 874.045 GFLOPS 2.558ms 839.510 GFLOPS 2.524ms 850.976 GFLOPS 2.178ms 985.895 GFLOPS
MT all f32 1024x1024
1.840ms 1167.088 GFLOPS 2.107ms 1019.108 GFLOPS 2.207ms 973.061 GFLOPS 1.647ms 1304.053 GFLOPS
1.926ms 1114.813 GFLOPS 2.133ms 1006.937 GFLOPS 2.145ms 1001.273 GFLOPS 1.603ms 1339.395 GFLOPS
1.793ms 1197.473 GFLOPS 2.092ms 1026.416 GFLOPS 2.250ms 954.622 GFLOPS 1.631ms 1316.506 GFLOPS
1.800ms 1192.852 GFLOPS 2.130ms 1008.175 GFLOPS 2.160ms 994.287 GFLOPS 1.557ms 1379.502 GFLOPS
1.920ms 1118.594 GFLOPS 2.111ms 1017.3