Take a look at the new version of this page: [ beta version ]. Tell us what you think.

ABSTRACT

Much of the performance analysis done in designing a computer is based on fundamental operation rates, like cycle time and number of pipe stages in an operation pipeline. This kind of analysis yields peak computation rates which, in fact, may never be realized. Resource contention between different units, each of which has a fundamental operation rate adequate to support a given overall peak, may cause the actual obtainable rate to be much less. In order to determine the effects of interactions between different requestors and common resources, micro-analysis, or detailed modelling of the part of the system in question, is necessary. In this paper, we report on the results of such an analysis on the operation pipe of the Titan graphics supercomputer. The Titan is a new class of machine with a supercomputer-style architecture implemented in a technology appropriate for a single-user machine. The micro-analysis reported here resulted in enhancing the actual obtainable computation rate from 10.8 Mflops to 14.6 Mflops for a particular real application, while the fundamental operation rate is 16 Mflops.

REFERENCES

• 1: Boris Beizer, Micro-Analysis of Computer System Performance, John Wiley & Sons, Inc., New York, NY, 1978
• 2: Tony Cheung , James E. Smith, A simulation study of the CRAY X-MP memory system, IEEE Transactions on Computers, v.35 n.7, p.613-622, July 1986 [doi>10.1109/TC.1986.1676802]
• 3: Dongarra, Jack J. "Performance of Various Computers Using Standard Linear Equations Software in a Fortran Environment" Technical Memorandum No. 23. Argonne National Laboratory, Feb. 1988.
• 4: Gateway Design Automation Corp. Hardware Description Language and Simulator. Gateway Design Automation Corp., Westford, Mass., March 1987.
• 5: Kogge, Peter M. The Architecture of Pipelined Computers. McGraw-Hill Co., New York, 1981.
• 6: Zurcher, Frank and Brian Randell. "'Iterative, Multi- Level Modelling---a Methodology for Computer System Design" in Proceedings IFIP Congress 68, 1968.

INDEX TERMS

Primary Classification:
  C. Computer Systems Organization
  C.4 PERFORMANCE OF SYSTEMS
      Subjects: Design studies

Additional Classification:
  C. Computer Systems Organization
  C.4 PERFORMANCE OF SYSTEMS
      Subjects: Performance attributes; Modeling techniques
  C.5 COMPUTER SYSTEM IMPLEMENTATION
      C.5.1 Large and Medium ("Mainframe") Computers
          Subjects: Super (very large) computers

  I. Computing Methodologies
  I.3 COMPUTER GRAPHICS
      I.3.7 Three-Dimensional Graphics and Realism
          Subjects: Color, shading, shadowing, and texture

General Terms:
Design, Performance