NERSCPowering Scientific Discovery Since 1974

NERSC GTC Benchmark Performance Data

The first table lists processing rates (in GFLOPS/s) and corresponding machine efficiences. Floating-point operation counts were obtained using the NERSC IPM utility on Seaborg. Benchmark runtimes are presented in the second table. Machine names are hyperlinks to machine descriptions.

GTC uses weak scaling.

All runs were conducted during normal production periods while a full workload was running. This can introduce a variability in the measurement that differs from machine to machine.

Detailed IPM Performance Counter Profiles

  • GTC Medium:  Bassi
  • GTC Large:  Bassi


  • MachineProcessing Rates (in GFLOPS/s) and Machine Efficiencies
    64 MPI Processes ("Medium" Case) 256 MPI Processes ("Large" Case)
    Seaborg 9.6 (10.0%) 32.5 (8.5%)
    Jacquard 33.9 (12.1%) 117.4 (10.4%)
    Bassi 42.6 (8.8%) 154.8 (8.0%)
    Jaguar XT3 Dual-Core 44.1 (13.2%) 178.4 (13.4%)
    Jaguar XT4 Dual-Core 45.4 (16.7%) 188.4 (18.7%)
    Franklin 52.2 (15.7%) 193.5 (14.5%)


    MachineBenchmark RunTime (seconds)
    64 MPI Processes
    ("Medium" Case)
    256 MPI Processes
    ("Large" Case)
    Seaborg 8019 s 9559 s
    Jacquard 2263 s 2648 s
    Bassi 1803 s 2008 s
    Jaguar XT3 Single-Core Small Pages 1472 s 1589 s
    Jaguar XT3 Single-Core Large Pages 2043 s 2158 s
    Jaguar XT3 Dual-Core Small Pages 1742 s
    Jaguar XT3 Dual-Core Large Pages 2100 s 2297 s
    Jaguar XT4 Single-Core Small Pages 1472 s 1518 s
    Jaguar XT4 Single-Core Large Pages 1995 s 2315 s
    Jaguar XT4 Dual-Core Small Pages 1690 s 1649 s
    Jaguar XT4 Dual-Core Large Pages 2045 s 2202 s
    Franklin 1472 s 1607 s