Procedural Layout of a High-Speed Floating-Point Arithmetic Unit
Abstract
This thesis presents a case study in the procedural design of the layout of a complex digital circuit. This is the task of writing a program which constructs a VLSI circuit layout given a set of variable parameters which specify the desired functionality of the circuit. We present a set of techniques for guaranteeing that the constructed circuit obeys the geometric and electrical design rules imposed by the underlying circuit technology. These include a set of simple circuit forms and composition rules for building precharged combinatorial circuits which are free of critical race conditions and charge-sharing problems. As an example, we carry out the creation of a program for building a floating-point addition unit which has selectable number of bits of exponent and fraction in the floating-point representation. The high-level design of a companion floating-point multiplication unit is also discussed.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jun 01, 1985
- Accession Number
- ADA459612
Entities
People
- Robert C. Armstrong
Organizations
- Massachusetts Institute of Technology