Donald H. Gibson

Considerations in block-oriented systems design

The feasibility of transmitting blocks of words between memory and CPU is the subject of this study. The question is pertinent to the design of very fast computing systems where the nanoseconds to traverse a few feet become significant. There is intuitive advantage to transmitting blocks of words, rather than a word at a time. The initial access time due to physical distance, effective address mapping, and priority is a few hundred nanoseconds. If this time could be prorated against several words, then the effective access time could be reduced to a few tens of nanoseconds. The question is, of course, can the extra words be useful to the CPU? This question was explored in a simulation model driven by customer-based IBM 7000 series data. The simulation results indicate that blocks of 4, 8, or 16 words, transmitted to a local storage of 2K to 4K words, will adequately prorate memory access time. With this configuration, block transfer is seen to be an efficient memory access method which can provide high performance, superior to single-word access.

Engineering and scientific processing on the IBM 3090

The IBM 3090 processor implementation of the System/370 Vector Architecture represents a major new system design for engineering and scientific processing, featuring both scalar and vector capability in a uniprocessor and in a dyadic and four-way parallel processing environment. The history of large-scale scientific processing is reviewed, leading to a statement of current requirements. The design objectives for scalar, parallel, and vector capabilities are identified, followed by a summary of the resulting 3090 features. Selected highlights of the vector hardware are given, followed by a summary of the supporting software. The paper concludes with a discussion of performance, beginning with the identification of suitable applications. An example is given of one application utilizing each of the three capabilities: scalar, parallel, and vector. Several of the most important performance parameters are identified.

Structural aspects of the system/360 model 85: I general organization

A basic design objective for the Model 85 was to add a computer to the SYSTEM/360 line that offers high performance over a wide range of job types. Simulation studies indicate that the Model 85 will provide an average three- to five-fold increase in internal performance with main storage capacities of up to four million bytes. This part of the paper discusses the major elements of the Model 85 within the architectural context of SYSTEM/360, including the addition of a high-speed buffer, called a cache. Also summarized are the simulation studies that led to use of the cache, selection of its parameters, and verification of internal performance of the system.

Design of the IBM System/390 computer family for numerically intensive applications: an overview for engineers and scientists

The IBM System/390® (S/390) computer family provides a two-order-of-magnitude performance range for numerically intensive applications. The engineer or scientist can use the same operating system, compiler, and run-time environment commonly across the family. This paper provides an overview of primary S/390 hardware and software products of interest for numerically intensive applications, including MVS/ESA™, VM/ESA®, AIX/ESA™, and the extension of FORTRAN for very large applications and parallel applications. The primary portion of the paper is focused on details of design interest in three specific hardware products within the S/390 family, with emphasis on the Enterprise System/9000™ (ES/9000™) Model 900. Also described is a potential parallel-computing configuration using the ESCON Director™. The paper concludes with a discussion of the generic system environments within which S/390 products can support the technical user.