Skip to main content
. 2014 Mar 27;8:30. doi: 10.3389/fninf.2014.00030

Table 1.

Technical specifications of the MASSIVE high performance computing system.

M1 AT THE AUSTRALIAN SYNCHROTRON
42 nodes (504 CPU-cores total) in one configuration:
    42 nodes with 12 cores per node running at 2.66 GHz
       48 GB RAM per node (2016 GB RAM total)
       2 NVIDIA M2070 GPUs with 6GB GDDR5 per node (84 GPUs total)
153 TB of fast access parallel file system
4x QDR Infiniband Interconnect
M2 AT MONASH UNIVERSITY
118 nodes (1720 CPU-cores total) in four configurations:
    32 nodes with 12 cores per node running at 2.66 GHz
       48 GB RAM per node (1536 GB RAM total)
       2 × NVIDIA M2070 GPUs with 6 GB GDDR5 per node (64 GPUs total)
    10 nodes with 12 cores per node (visualization/high memory configuration)
       192 GB RAM per node (1920 GB RAM total)
       2 × NVIDIA M2070Q GPUs with 6 GB GDDR5 per node (20 GPUs total)
    56 nodes with 16 cores per node running at 2.66 GHz
       64 GB RAM per node (3584 GB RAM total)
       2 × NVIDIA K20 (9 nodes—18 GPUs total)
       2 × Intel PHI (10 nodes—20 coprocessors total)
    20 nodes with 16 cores per node running at 2.66 GHz
       128 GB RAM per node (2560 GB RAM total)
       2 × NVIDIA K20 (40 GPUs total)
345 TB of fast access parallel file system
4 × QDR Infiniband Interconnect
Combined the M1 and M2 have 2,224 CPU-cores.