Alltoall is one of PMB's Collective Benchmarks. It is a benchmark of the MPI_Alltoall function. Every process inputs X*(#processes) bytes (X for each process) and receives X*(#processes) bytes (X from each process).
2-node Cluster Results
4-node Cluster Results
8-node Cluster Results
กก
#bytes #reps linux dom0 domU(noSMP) domU(SMP)
--------------------------------------------------------------------------
#bytes #reps linux dom0 domU(noSMP) domU(SMP)
--------------------------------------------------------------------------
#bytes #reps linux dom0 domU(noSMP) domU(SMP) -------------------------------------------------------------------------- 0 1000 0.39 0.41 0.39 0.36 1 1000 2678.25 2773.21 2686.01 14706.96 2 1000 2249.23 2575.34 2493.09 14008.09 4 1000 2159.37 2470.41 2266.44 14301.67 8 1000 2010.59 2116.34 2149.60 13649.23 16 1000 1606.12 1812.12 2108.82 12142.42 32 1000 1180.29 1514.41 2218.64 14258.53 64 1000 1355.38 1604.65 1865.77 12357.01 128 1000 1156.99 1527.08 1813.80 8134.88 256 1000 1453.60 1833.51 2068.89 7411.81 512 1000 1006.72 1331.20 44050.00 63847.75 1024 1000 1109.21 1497.47 9680.79 27108.28 2048 1000 1947.87 2673.41 101600.27 27313.09 4096 1000 3256.38 3834.41 35894.57 89693.63 8192 1000 9633.96 9693.02 109858.16 159793.37 16384 1000 16584.38 16113.34 126134.45 217652.17 32768 1000 66122.23 70757.14 107099.17 274720.67 65536 640 132776.10 128202.34 201898.90 504895.94 131072 320 213916.58 216455.56 244754.73 882699.44 262144 160 436046.73 444038.82 443075.76 1359254.52 524288 80 891764.05 906276.27 895841.75 2310571.51 1048576 40 1801052.91 1814068.68 1802805.14 4382027.32 2097152 20 3609545.99 3633866.13 3626608.94 8410867.65 4194304 10 7213814.30 7366660.86 7383321.82 16083507.46
