Browse Prior Art Database

Pipelined Carry-Lookahead Adder for Fixed-Point Arithmetic

IP.com Disclosure Number: IPCOM000060026D
Original Publication Date: 1986-Feb-01
Included in the Prior Art Database: 2005-Mar-08
Document File: 2 page(s) / 44K

Publishing Venue

IBM

Related People

Sinha, B: AUTHOR

Abstract

A technique is described whereby pipelined architecture for a carry- look-ahead (CLA) adder achieves high throughput rates for fixed-point add operations. A method to pipeline a sixteen-bit fixed-point CLA adder is described. By following the same procedure, this method can be extended to thirty-two, sixty-four or higher number of strings. A standard sixteen-bit fixed point CLA adder using carry-propagate/generate units and a carry-look-ahead unit is shown in Fig. 1. The eight inputs to the look-ahead unit are generated from block carry propagate/generate units. Four-bit blocks 11 generate the block outputs P1j 12 and G1j 13 for j=0, 1, 2 and 3. The carry outputs 14 are connected to the block carry inputs in the first level, which finally calculates the sum.

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 76% of the total text.

Page 1 of 2

Pipelined Carry-Lookahead Adder for Fixed-Point Arithmetic

A technique is described whereby pipelined architecture for a carry- look-ahead (CLA) adder achieves high throughput rates for fixed-point add operations. A method to pipeline a sixteen-bit fixed-point CLA adder is described. By following the same procedure, this method can be extended to thirty-two, sixty-four or higher number of strings. A standard sixteen-bit fixed point CLA adder using carry-propagate/generate units and a carry-look-ahead unit is shown in Fig. 1. The eight inputs to the look-ahead unit are generated from block carry propagate/generate units. Four-bit blocks 11 generate the block outputs P1j 12 and G1j 13 for j=0, 1, 2 and 3. The carry outputs 14 are connected to the block carry inputs in the first level, which finally calculates the sum. Pipelining such a dual-level design requires separating each subfunction in the time domain and storing the intermediate results into separate memory devices. The five subfunctions of adding two sixteen- bit fixed-point numbers are: Subfunction 1: Generating P0i and G0i:

(Image Omitted)

Subfunction 2: Generating the block propagate/generate outputs for the CLA unit:

(Image Omitted)

Subfunction 3: Generating the block carry for the first level carry propagate/generate units:

(Image Omitted)

Subfunction 4: Generating the other bit positions - carry bits

(Image Omitted)

Subfunction 5: Production of the sum bits for each bit position:

(Image Omitted)

With...