Architectural vs. delivered performance of the IBM RS/6000 and the Astronautics ZS-1

The paper presents a comparative performance evaluation of two novel superscalar architectures, the IBM RS/6000 and the Astronautics ZS-1. The machine architectures are presented to lay the groundwork for discussing performance issues, with a RISC architecture, the Decstation 3100, serving as a baseline architecture. Important qualitative observations are made to highlight similarities between the RS/6000 and the ZS-1. Performance measurements from running systems are presented to examine delivered performance. An analytic method of bounding machine performance is introduced to help strip out implementation related details. By comparing these performance bounds, and their component terms, to the achieved performance, the authors assess how well an application code does and how well it might do, in capitalizing on the available performance. Common sources of performance losses are revealed, areas for improvement are highlighted, and specific improvements are suggested.<<ETX>>

[1]  David A. Patterson,et al.  Computer Architecture: A Quantitative Approach , 1969 .

[2]  Peter Y.-T. Hsu,et al.  Overlapped loop support in the Cydra 5 , 1989, ASPLOS III.

[3]  Yale N. Patt,et al.  HPS, a new microarchitecture: rationale and introduction , 1985, MICRO 18.

[4]  David W. Anderson,et al.  The IBM System/360 model 91: machine philosophy and instruction-handling , 1967 .

[5]  Edward S. Davidson,et al.  Polycyclic vector scheduling vs. chaining on 1-port vector supercomputers , 1988, Proceedings. SUPERCOMPUTING '88.

[6]  James C. Dehnert,et al.  Overlapped loop support in the Cydra 5 , 1989, ASPLOS 1989.

[7]  Andrew R. Pleszkun,et al.  Structured Memory Access Architecture , 1983, ICPP.

[8]  Geoffrey C. Fox,et al.  The Perfect Club Benchmarks: Effective Performance Evaluation of Supercomputers , 1989, Int. J. High Perform. Comput. Appl..

[9]  F. H. Mcmahon,et al.  The Livermore Fortran Kernels: A Computer Test of the Numerical Performance Range , 1986 .

[10]  E.S. Davidson,et al.  The effects of memory latency and fine-grain parallelism on Astronautics ZS-1 performance , 1990, Twenty-Third Annual Hawaii International Conference on System Sciences.

[11]  Andrew R. Pleszkun,et al.  Features of the Structured Memory Access (SMA) Architecture , 1986, IEEE Computer Society International Conference.

[12]  Gerry Kane,et al.  MIPS RISC Architecture , 1987 .

[13]  James E. Smith,et al.  The ZS-1 central processor , 1987, ASPLOS 1987.

[14]  James E. Smith,et al.  Decoupled access/execute computer architectures , 1984, TOCS.