Memory Bandwidth Bytes, GB/s per node 6291456, 379.297050 100663296, 3754.674992 509607936, 6521.472413 1610612736, 8513.456479 3932160000, 9018.901766 GEMM M, N, K, BATCH, GF/s per rank 16, 8, 16, 256, 0.564958 16, 16, 16, 256, 243.148058 16, 32, 16, 256, 440.346877 32, 8, 32, 256, 439.194136 32, 16, 32, 256, 847.334141 32, 32, 32, 256, 1430.892623 64, 8, 64, 256, 1242.756741 64, 16, 64, 256, 2196.689493 64, 32, 64, 256, 3697.458072 16, 8, 256, 256, 899.582627 16, 16, 256, 256, 1673.537756 16, 32, 256, 256, 2959.597089 32, 8, 256, 256, 1558.858630 32, 16, 256, 256, 2864.839445 32, 32, 256, 256, 4810.671254 64, 8, 256, 256, 2386.092942 64, 16, 256, 256, 4451.665937 64, 32, 256, 256, 5942.124095 8, 256, 16, 256, 799.867271 16, 256, 16, 256, 1584.624888 32, 256, 16, 256, 1949.422338 8, 256, 32, 256, 1389.417474 16, 256, 32, 256, 2668.344493 32, 256, 32, 256, 3234.162120 8, 256, 64, 256, 2150.925128 16, 256, 64, 256, 4012.488132 32, 256, 64, 256, 5154.785521 Communications Packet bytes, direction, GB/s per node 4718592, 1, 245.026198 4718592, 2, 251.180996 4718592, 3, 361.110977 4718592, 5, 247.898447 4718592, 6, 249.867523 4718592, 7, 359.033061 15925248, 1, 255.030946 15925248, 2, 264.453890 15925248, 3, 392.949183 15925248, 5, 256.040644 15925248, 6, 264.681896 15925248, 7, 392.102622 37748736, 1, 258.823333 37748736, 2, 268.181577 37748736, 3, 401.478191 37748736, 5, 258.995363 37748736, 6, 268.206586 37748736, 7, 400.397611 Per node summary table L , Wilson, DWF4, Staggered, GF/s per node 8 , 155, 1386, 50 12 , 694, 4208, 230 16 , 1841, 6675, 609 24 , 3934, 8573, 1641 32 , 5083, 9771, 3086