Design of a high performance 32×32-bit multiplier with a novel sign select Booth encoder
Kiwon Choi, Minkyu Song
Abstract
Kiwon Choi, Minkyu Song
Abstract
In this paper, a high performance 32/spl times/32-bit multiplier for a DSP core is proposed. The multiplier is composed of a novel sign select Booth encoder, an efficient data compressor block with a novel compound full-adder, and a 64-bit conditional sum adder with a separated carry generation block. The proposed 32/spl times/32-bit multiplier is designed by a full-custom method and there are about 28000 transistors in an active area of 1.59 mm/spl times/1.68 mm with 0.6 /spl mu/m CMOS technology. From the experimental results, the multiplication time of the 32/spl times/32-bit multiplier is about 9.8 ns at a 3.3 V power supply, and it consumes about 186 mW at 100 MHz.
OpenAlex reports 20 citations for this work. Citation counts describe recorded attention and do not establish research quality.
A contribution statement is not available in the OpenAlex record.
Method details are not available in the OpenAlex metadata.
Findings are not separately available in the OpenAlex metadata.
Limitations are not available in the OpenAlex metadata.
Application details are not available in the OpenAlex metadata.
In this paper, a high performance 32/spl times/32-bit multiplier for a DSP core is proposed. The multiplier is composed of a novel sign select Booth encoder, an efficient data compressor block with a novel compound full-adder, and a 64-bit conditional sum adder with a separated carry generation block. The proposed 32/spl times/32-bit multiplier is designed by a full-custom method and there are about 28000 transistors in an active area of 1.59 mm/spl times/1.68 mm with 0.6 /spl mu/m CMOS technology. From the experimental results, the multiplication time of the 32/spl times/32-bit multiplier is about 9.8 ns at a 3.3 V power supply, and it consumes about 186 mW at 100 MHz.
Key concepts: Adder, Multiplier (economics), Booth's multiplication algorithm, Encoder, CMOS, 8-bit, 4-bit, 16-bit