Browse Prior Art Database

N-bit Multiply Step using Twos Complement within Iteration

IP.com Disclosure Number: IPCOM000108513D
Original Publication Date: 1992-Jun-01
Included in the Prior Art Database: 2005-Mar-22
Document File: 2 page(s) / 71K

Publishing Venue

IBM

Related People

Muhich, JS: AUTHOR

Abstract

Disclosed is a step-wise algorithm for binary multiplication comprised of precomputation of one operand before and during the iterative multiply step in a RISC base microprocessor. By precomputing the operand, the amount of computation required for the partial multiply step is reduced, thus reducing the amount of hardware path length and cycle time required to perform the function.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 70% of the total text.

N-bit Multiply Step using Twos Complement within Iteration

       Disclosed is a step-wise algorithm for binary
multiplication comprised of precomputation of one operand before and
during the iterative multiply step in a RISC base microprocessor.  By
precomputing the operand, the amount of computation required for the
partial multiply step is reduced, thus reducing the amount of
hardware path length and cycle time required to perform the function.

      Below is a program flow of the N-bit multiply step algorithm
disclosed.  Generally, the B operand is conditionally twos
complemented as shown in lines 6 and 9 of the program below. By
presetting the B operand to this state before and during the main
loop, the range of operation performed is split in half. Line 8 of
the program is reduced such that the partial product of (TB * M) is
always added to TA rather than added to or subtracted from as in the
case where the B operand had not been precomputed for complement.

      With proper pipelining and parallel execution of both the
partial multiply and the two complement functions, higher speed logic
is obtained.  When implementing a highly parallel/pipeline RISC base
processor, the designer cannot afford to slow down the cycle time of
the processor due to multicycle instructions such as the multiply
instruction. This disclosed method for executing the multiply
function helps aid the designer to keep the multiply function off the
critical path of the processor.
...