Performance of the Lawrence Livermore National Lab benchmark on the Alliant FX/8
Allen D. Malony, Walid Abu-Sufah · OSTI OAI (U.S. Department of Energy Office of Scientific and Technical Information) · 1986
The benchmark timing of the twenty-four Lawrence Livermore National Laboratory (LLNL) kernels was done on the Alliant FX/8 and the results are presented in this report. These kernels measure Fortran numerical execution rates for a spectrum of CPU-bond computations representing the LLNL workload. Following the instructions and guidelines for executing the LLNL benchmark, several types of experiments were performed. First, the benchmark was compiled for vector optimizations and executed on one CE of the FX/8. Experiments were then conducted on two, four, and eight CEs with concurrent optimizations also enabled. All experiments were performed in single-user mode on the FX/8. Each experiment was executed three times and the results were averaged.