Scheduled System Maintenance:
Some services will be unavailable Sunday, March 29th through Monday, March 30th. We apologize for the inconvenience.
By Topic

IBM Journal of Research and Development

Issue 2 • Date March 1986

Filter Results

Displaying Results 1 - 9 of 9
  • New scalar and vector elementary functions for the IBM System/370

    Publication Year: 1986 , Page(s): 126 - 144
    Cited by:  Papers (10)  |  Patents (11)
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (1658 KB)  

    Algorithms have been developed to compute short- and long-precision elementary functions: SIN, COS, TAN, COTAN, LOG, LOG10, EXP, POWER, SQRT, ATAN, ASIN, ACOS, ATAN2, and CABS, in scalar (28 functions) and vector (22 functions) mode. They have been implemented as part of the new VS FORTRAN library recently announced along with the IBM 3090 Vector Facility. These algorithms are essentially table-based algorithms. An important feature of these algorithms is that they produce bitwise-identical results on scalar and vector System/370 machines. Of these, for five functions the computed value result is always the correctly rounded value of the infinite-precision result. For the rest of the functions, the value returned is one of the two floating-point neighbors bordering the infinite-precision result, which implies exact results if they are machine-representable. For the five correctly rounded elementary functions, scalar and vector algorithms are designed independently so as to optimize performance in each case. For other functions, the bitwise-identical constraint leads to algorithms which compromise between scalar and vector performance. We have been able to design algorithms where this compromise is minimal and thus achieve very good performance on both scalar and vector implementations. For our test measurements on high-end System/370 machines, our scalar functions are always faster (sometimes by as much as a factor of 2.5) as compared to the old VS FORTRAN library. Our vector functions are usually two or three times faster than our scalar functions. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Fourier transform and convolution subroutines for the IBM 3090 Vector Facility

    Publication Year: 1986 , Page(s): 145 - 161
    Cited by:  Papers (10)  |  Patents (4)
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (1312 KB)  

    A set of highly optimized subroutines for digital signal processing has been included in the Engineering and Scientific Subroutine Library (ESSL) for the IBM 3090 Vector Facility. These include FORTRAN-callable subroutines for Fourier transforms, convolution, and correlation. The subroutines are carefully designed and tuned for optimal vector and cache performance. Speedups of up to 9½ times over scalar performance on the 3090 have been obtained. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • A vectorizing Fortran compiler

    Publication Year: 1986 , Page(s): 163 - 171
    Cited by:  Papers (9)  |  Patents (11)
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (809 KB)  

    This paper describes a vectorizing Fortran compiler for the IBM 3090 Vector Facility. Opportunities for vectorization are investigated for eight levels of DO-loop nesting. Recurrences in inner loops do not prevent vectorization of outer loops. A least-cost analysis determines from the opportunities identified which specific vectorization will result in the fastest execution. The normal optimization phases of the compiler produce much of the information needed for the vectorization analysis. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Seismic migration on the IBM 3090 Vector Facility

    Publication Year: 1986 , Page(s): 172 - 183
    Cited by:  Papers (2)  |  Patents (1)
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (1035 KB)  

    Seismic prospecting aims at determining the structure of the earth from indirect measurements. Acoustic wave fields are generated at the surface, penetrate the earth, and are backscattered by the earth's inhomogeneities. The data recorded at the surface are processed in a complex sequence of steps among which seismic migration plays an important role. This is a wave depropagation process that permits the localization in depth of the origin of the diffraction events measured (in time) at the surface. This paper presents an overview of the major wave-equation migration methods. The most frequently executed algorithms or kernels on which the execution speed depends most crucially are given particular attention. The speedup resulting from scalar-to-vector formulation is presented over wide ranges of dimensionality for linear tridiagonal equation solvers, Fourier Transforms, and convolution operations. The vectorizability and resulting speedup are also addressed in the case of migration schemes known as the Phase-Shift Method and the Phase Shift Plus Interpolation (PSPI) Method. It is shown that Fourier domain migration based on the phase-shift concept lends itself conveniently to multilevel parallelism on the 3090 Vector Facility (VF): vectorization of the innermost loops and concurrent processing in the outer loops by means of the VS FORTRAN Version 2 Multitasking Facility. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Parallel iterative linear solvers for oil reservoir models

    Publication Year: 1986 , Page(s): 184 - 192
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (695 KB)  

    This paper suggests a new algorithm for solving sparse linear systems which is readily parallelized and is very efficient for matrices arising in such domains as reservoir modeling. The algorithm is, in fact, a variant of the incomplete block factorization technique, accelerated by Biconjugate Gradient iterations or by another acceleration method. Implementation on a vector computer such as the IBM 3090 Vector Facility is described. We also suggest some refinements of known preconditioning methods enabling their parallel computation. Numerical experiments are presented to display the performance of the suggested methods. Special attention is given to fully implicit, multiphase models which yield asymmetrical systems. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Monte Carlo photon transport on a vector supercomputer

    Publication Year: 1986 , Page(s): 193 - 202
    Cited by:  Papers (4)
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (797 KB)  

    The use of “event-based” algorithms for particle transport Monte Carlo methods has allowed the successful adaptation of these methods to vector supercomputers. An alternative algorithm for the specific application of photon transport in an axisymmetric inertially confined fusion plasma has been developed and implemented on a vector supercomputer. The new algorithm is described; its unique features are discussed and compared with existing vectorized algorithms for Monte Carlo. Numerical results are presented illustrating its efficiency on a vector supercomputer, relative to an optimized scalar Monte Carlo algorithm that was developed for this purpose. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Chemical and mechanical performance of flexible magnetic tape containing chromium dioxide

    Publication Year: 1986 , Page(s): 203 - 216
    Cited by:  Papers (6)
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (1156 KB)  

    The development of a magnetic-recording tape using chromium dioxide suitable for use with the tape drive of the IBM 3480 Magnetic Tape Subsystem was found to require careful optimization of the chemical and mechanical properties of the coating. This paper discusses the role of chromium dioxide in the oxidative and hydrolytic degradation of the polyester-polyurethane binders used in most flexible tape coatings, and the subsequent necessity for careful binder selection to eliminate (for all practical purposes) these degradative effects. In addition to the chemical behavior, the paper discusses the role of the interaction of the chromium dioxide with the binder necessary to obtain mechanical performance suitable for the 3480 tape drive. A high modulus and a relatively high glass-transition temperature were found to be required to avoid changes in the frictional properties of the tape. The incorporation of a rough-textured back coat was found to control the change in the surface topography of the recording surface near the spool hub which arises as a result of compressive forces. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Recent papers by IBM authors

    Publication Year: 1986 , Page(s): 217 - 226
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (871 KB)  

    The information listed heries supplied by the Institute for Scientific Information. Reprints ofthe papers may be obtained by writing dirctly to the first author cited. Papers are listed alphabetically by author. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Recent IBM patents

    Publication Year: 1986 , Page(s): 227 - 228
    Save to Project icon | PDF file iconPDF (173 KB)  
    Freely Available from IEEE

Aims & Scope

The IBM Journal of Research and Development is a peer-reviewed technical journal, published bimonthly, which features the work of authors in the science, technology and engineering of information systems.

Full Aims & Scope

Meet Our Editors

Editor-in-Chief
Clifford A. Pickover
IBM T. J. Watson Research Center