The performance of two second-generation supercomputers, the NEC SX-3 and the Fujitsu VP2600, is analyzed using the Standard Los Alamos Benchmark Set, the Mendez Fluid Dynamics Codes, and some highly vectorizable production-type codes from Los Alamos.For comparison, data are also given for a single processor of the CRAY Y-MP8/264.Factors affecting performance such as memory bandwidth, vector register organization, and the effects of multiple vector pipelines are examined.On a highly vectorizable code that can take advantage of multiple vector pipes, the SX-3 and VP2600 are faster than the single CRAY Y-MP processor by factors of seven to eight.

An experiment on vector processing and multitasking of the nuclear reactor transient analysis code RELAPS has been carried out. Vector processing and multitasking were achieved by using the independency of the spatial meshes. The vectorization ratio is 83%. The performance ratio in the vector mode to that in the scalar mode is about three on the FACOM VP-100. For multitasking the spatial meshes are halved, and each group of meshes is processed on a different processor. The effect of multitasking was estimated from the CPU time on the FACOM VP-100, which is a single-processor system. The improvement of wall clock time will be about 1.6 for parallelized subroutines. The total improvement of wall clock time will be about 1.2 because of parallelization ratio 45%.>