Supercomputers and Their Performance in Computational Fluid by Toshihiro Hongo (auth.), Kozo Fujii (eds.)

By Toshihiro Hongo (auth.), Kozo Fujii (eds.)

Supercomputer applied sciences have developed swiftly because the first commercial-based supercomputer, CRAY-1 was once brought in 1976. In early 1980's 3 eastern large­ pcs seemed, and Cray study added the X-MP sequence. those machines together with the later-announced CRAY-2 and NEC SX sequence created one new release of supercomputers, and the marketplace was once unfold dramatically. the height functionality used to be greater than 1 GFLOPS and the compiler development was once extraordinary. There seemed many articles and books that defined their structure and their functionality at the past due 1980's observed a brand new new release of supercomputers. a number of benchmark difficulties. Following CRAY Y-MP and Hitachi S-820 brought in 1988, NEC introduced SX-3 and Fujitsu introduced the VP2000 sequence in 1990. furthermore, Cray examine introduced the Y-MP C-90 past due in 1991. the height functionality of those machines reached numerous to a couple ten's GFLOPS. The features of those machines are identified, yet their useful functionality has no longer been good documented thus far. Computational Fluid Dynamics (CFD) is among the vital learn fields which have been progressing with the expansion of supercomputers. brand new fluid dynamic re­ seek can't be mentioned with no supercomputers and because CFD is without doubt one of the im­ portant clients of supercomputers, destiny improvement of supercomputers has to take the necessities of CFD into consideration. there are various benchmark stories on hand at the present time. in spite of the fact that, they usually use so known as kernels. For fluid dynamics researchers, benchmark try on actual fluid dynamic codes are necessary.

Show description

Read Online or Download Supercomputers and Their Performance in Computational Fluid Dynamics PDF

Best dynamics books

Supercomputers and Their Performance in Computational Fluid Dynamics

Supercomputer applied sciences have developed speedily because the first commercial-based supercomputer, CRAY-1 was once brought in 1976. In early 1980's 3 eastern large­ pcs seemed, and Cray study brought the X-MP sequence. those machines together with the later-announced CRAY-2 and NEC SX sequence created one new release of supercomputers, and the industry used to be unfold dramatically.

Scale Invariance, Interfaces, and Non-Equilibrium Dynamics

The NATO complex research Institute on "Scale Invariance, Interfaces and Non­ Equilibrium Dynamics" was once held on the Isaac Newton Institute for Mathematical Sciences in Cambridge, united kingdom from 20-30 June 1994. the themes mentioned on the Institute have been all interested in the beginning and nature of advanced buildings chanced on faraway from equilibrium.

Additional info for Supercomputers and Their Performance in Computational Fluid Dynamics

Sample text

A variety of software products has also been developed to fully exploit the hardware capabilities. 1. INTRODUCTION As science and technology advance in scale and complexity, demand for sheer numerical calculation power is accelerating. Furthermore, as society becomes more sophisticated, the use of the supercomputer is spreading into a variety of fields thus far unthought-of. To meet increasing power demand and diversifying application needs, Hitachi has developed the S-820 supercomputer, with the design objectives: (1) world-class performance, (2) large system capacity, 43 (3) ease of use, (4) high reliability, (5) ease of installation.

Invisible to the user, the operating system is provided with an asynchronous move instruction, which is similar in format and operation to an input/output instruction. 6 Fig. 3 Vector Register During execution, vector operands reside in the Main Storage and are fetched into the Vector Processor as necessary. Likewise, the results of computation are stored into the Main Storage. In many instances, operands tend to be used more than once, and the result of computation tend to be referenced later in the same program.

Two mask pipelines can be operated in parallel. Two identical multiply & add/logical pipelines which are called universal pipelines are adopted in the VP2000 series. ) Each pipeline can execute either of the multiply, add, multiply & add, first order recurrence or logical operations. These two flexible pipelines can be usefully operated by executing operations concurrently under various combinations. The divide pipeline is provided for executing especially the divide operation. The divide operation is realized with its throughput being 1/3 of the add or multiply operation.

Download PDF sample

Rated 4.50 of 5 – based on 3 votes
 

Author: admin