Skip to content

Results: Runtime Performance of Graph Generations

Hiroki Kanezashi edited this page Jan 15, 2019 · 2 revisions

Performance results of large-scale transaction generation

Execution Environments

  • GPU: NVIDIA Tesla V100-PCIE, 8 devices
  • CPU: Intel(R) Xeon(R) CPU E5-2697 v4 @ 2.30GHz, total 72 cores (18 cores, 4 sockets)
  • RAM: 1TB DDR4 2400MHz
  • OS: Ubuntu 16.04.5 LTS
  • Software
    • GCC: version 5.4.0 20160609 (Ubuntu 5.4.0-6ubuntu1~16.04.10)
    • CUDA: V9.0.176
    • Python: 2.7.12
      • NetworkX: 1.11
      • NumPy: 1.15.2
    • Java: OpenJDK version "1.8.0_191" (64 bit)
    • TensorFlow: 1.11.0

10 million transactions

100 million transactions

1 billion transactions