Browse Prior Art Database

High Speed Multiply Operations

IP.com Disclosure Number: IPCOM000090220D
Original Publication Date: 1969-Mar-01
Included in the Prior Art Database: 2005-Mar-05
Document File: 2 page(s) / 32K

Publishing Venue

IBM

Related People

Frye, HE: AUTHOR [+2]

Abstract

In a data processing system having a processing unit 3 for multiplying which has a maximum data width less than that of the operands to be multiplied, the multiply can be carried out at greater speed with essentially the same accuracy if some of the multiplication steps are performed only with the higher order bits of the multiplier operand.

This text was extracted from a PDF file.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 56% of the total text.

Page 1 of 2

High Speed Multiply Operations

In a data processing system having a processing unit 3 for multiplying which has a maximum data width less than that of the operands to be multiplied, the multiply can be carried out at greater speed with essentially the same accuracy if some of the multiplication steps are performed only with the higher order bits of the multiplier operand.

A 56-bit multiplier operand is placed in register 4 which has two equal 28-bit sections A and AX. A 56-bit multiplicand operand is placed in the register 5 which also has two equal 28-bit sections B and BX. Multiplication of the two operands is performed in the processing unit 3. Assuming unit 3 has a capability of performing multiplication of two 28-bit portions, 28-by-28 multiply, of the operands in register 4 and 5, the normal multiplication forms the partial products AX BX, A BX, B AX, and :AB and adds them together in a weighted fashion. More particularly, the high half of AX BX is added to the product A BX while the low order half is discarded. The result of that add is added to the product B AX and the low order half of this result is discarded. Thereafter the high order half of that last addition is added to the final product AB. In this normal operation, all multiplications are 28-by-28 multiplies.

Since the lower order portion of all the partial products, except AB, is truncated, the multiplication is simplified if the lower order multiplications are never performed in the first instan...