Allgather is one of PMB's Collective Benchmarks. It is a benchmark of the MPI_Allgather function. Every process inputs X bytes and receives the gathered X*(#processes) bytes.
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 0.38 0.44 0.39 0.39 1 1000 588.59 866.55 1241.69 2300.20 2 1000 588.89 865.51 1243.23 2455.89 4 1000 589.71 869.18 1246.12 2489.85 8 1000 595.54 873.78 1253.26 2250.64 16 1000 607.20 885.08 1269.61 2624.24 32 1000 626.22 902.46 1297.43 2482.02 64 1000 673.43 955.05 1371.02 2799.97 128 1000 740.89 1068.95 1499.21 3004.73 256 1000 945.74 1268.90 1930.95 3312.19 512 1000 1193.19 1685.87 2250.80 14933.52 1024 1000 1694.42 2362.32 2701.97 28572.77 2048 1000 2834.54 3565.36 5548.47 44815.99 4096 1000 5018.00 6055.27 14818.61 61597.52 8192 1000 10181.83 11068.89 29511.51 106530.63 16384 1000 20659.84 22284.77 57310.04 132754.82 32768 1000 40280.40 41840.40 115403.28 205694.26 65536 640 166736.07 103170.90 205376.89 454635.08 131072 320 222877.78 233274.93 230034.09 1155136.75 262144 160 458425.04 469142.02 461162.80 1566099.62 524288 80 924689.93 932183.54 919721.54 1843950.38 1048576 40 1813785.78 1860086.11 1765071.19 2791439.14 2097152 20 3584000.77 3679077.29 3755035.53 6371308.64 4194304 10 6780246.95 7274533.28 7004422.09 12486541.02
