Performance of scientific computing platforms running MCNP4B

摘要

A performance study has been made for the MCNP4B Monte Carlo radiation transport code on a wide variety of scientific computing platforms ranging from personal computers to Cray mainframes. We present the timing study results using MCNP4B and its new test set and libraries. This timing study is unlike other timing studies because of its widespread reproducibility, its direct comparability to the predecessor study in 1993, and its focus upon a nuclear engineering code. Our results, derived from using the new 29-problem test set for MCNP4B, (1) use a highly versatile and readily available physics code; (2) show that timing studies are very problem dependent; (3) present the results as raw data allowing comparisons of performance to other computing platforms not included in this study to those platforms that were included; (4) are reproducible; and (5) provide a measure of improvement in performance with the MCNP code due to advancements in software and hardware over the past 4 years. In the 1993 predecessor study using MCNP4A, the performances were based on a 25 problem test set. We present our data based on MCNP4B's new 29 problem test set which cover 97% of all the FORTRAN physics code lines in MCNP4B. Like the previous study the new test problems and the test data library are available from the Radiation Shielding and Information Computational Center (RSICC) in Oak Ridge, Tennessee. Our results are reproducible because anyone with the same workstation, compiler, and operating system can duplicate the results presented here. The computing platforms included in this study are the Sun Sparc2, Sun Sparc5, Cray YMP 8/128, HP C180,SGI origin 2000, DBC 3000/600, DBC AiphaStation 500(300 MHz), IBM RS/6000-590, HP /9000-735, Micron Milienia Pro 200 MHz PC, and the Cray T94/128.