• DOI:10.1145/641845.641849
• Corpus ID: 15021135

VAX floating point: a solid foundation for numerical computation

@article{Payne1980VAXFP,
  title={VAX floating point: a solid foundation for numerical computation},
  author={Mary Payne and Dileep Bhandarkar},
  journal={SIGARCH Comput. Archit. News},
  year={1980},
  volume={8},
  pages={22-33}
}

• M. Payne, D. Bhandarkar
• Published 1 June 1980
• Engineering
• SIGARCH Comput. Archit. News

This paper describes the environment provided by the recently announced enhanced VAX floating point architecture and compares it with the features described in various proposals for an IEEE Floating Point Standard. The IEEE Standards Committee created a Working Group on Floating Point Arithmetic late in 1977. Four formal proposals for a floating point standard have been submitted to the Working Group. Of these, three have received extensive discussion. The fourth describes a floating point system in which range and precision are dynamically traded off against each other in order to avoid overflow and underflow; it was submitted after agreement had already been reached on the basic specifications for a floating point representation, and never received serious consideration. The other three proposals are in substantial agreement on the representation and arithmetic for ordinary floating point numbers. There has, however, been sharp disagreement on the handling of exceptions and on how extra precision should be made available for "critical" calculations.