05-05-2017, 09:35 AM
Radix-4 Modified Booth Multiplier and this implementation is compared to Radix-2 Booth Multiplier. The modified Booth algorithm employs both addition and subtraction and also treats positive and negative operands evenly. No special actions are required for negative numbers. In this work, we investigate the implementation method of Parallel MAC with the least possible delay. Parallel MAC is often used in digital signal processing and video / graphics applications.
A new multiplier and accumulator (MAC) architecture for high-speed arithmetic combining multiplication with accumulation and the design of a rear view aggregator (CLA) improves performance. The modified cabin multiplication algorithm is designed using a high speed adder. The high-speed adder is used to speed up the multiplication operation. The design of this algorithm is performed using VHDL and is simulated using Xilinx ISE 9.1i software has been used and implemented in FPGA xc3s50-5pq208.
By combining multiplication with accumulation and the creation of a hybrid type of carry save adder (CSA), performance has been improved. Since the accumulator having the largest MAC delay was merged into CSA, the overall performance was high. The proposed CSA tree uses the radix-2-based Booth algorithm based on the complement of 1 (MBA) and has the modified matrix for the sign extension in order to increase the bit density of the operands. The CSA propagates the transport to the least significant bits of the partial products and generates the least significant bits in advance to decrease the number of input bits of the final adder.
In addition, the proposed MAC accumulates the intermediate results in the sum and transport bit type instead of the final adder output, which allowed to optimize the channelization scheme to improve performance. The proposed architecture was synthesized with 250, 180 and 130 / xm, and a standard CMOS library of 90 nm. Based on the theoretical and experimental estimation, we analyze the results such as the amount of hardware resources, the delay, and the pipeline scheme. We use the Sakurai alpha power law for delay modeling. The proposed MAC showed superior properties to the standard design in many ways and operating twice as much as previous research on the similar clock frequency. We are experts in that the proposed MAC can adapt to various fields that require high performance, such as signal processing areas.
A new multiplier and accumulator (MAC) architecture for high-speed arithmetic combining multiplication with accumulation and the design of a rear view aggregator (CLA) improves performance. The modified cabin multiplication algorithm is designed using a high speed adder. The high-speed adder is used to speed up the multiplication operation. The design of this algorithm is performed using VHDL and is simulated using Xilinx ISE 9.1i software has been used and implemented in FPGA xc3s50-5pq208.
By combining multiplication with accumulation and the creation of a hybrid type of carry save adder (CSA), performance has been improved. Since the accumulator having the largest MAC delay was merged into CSA, the overall performance was high. The proposed CSA tree uses the radix-2-based Booth algorithm based on the complement of 1 (MBA) and has the modified matrix for the sign extension in order to increase the bit density of the operands. The CSA propagates the transport to the least significant bits of the partial products and generates the least significant bits in advance to decrease the number of input bits of the final adder.
In addition, the proposed MAC accumulates the intermediate results in the sum and transport bit type instead of the final adder output, which allowed to optimize the channelization scheme to improve performance. The proposed architecture was synthesized with 250, 180 and 130 / xm, and a standard CMOS library of 90 nm. Based on the theoretical and experimental estimation, we analyze the results such as the amount of hardware resources, the delay, and the pipeline scheme. We use the Sakurai alpha power law for delay modeling. The proposed MAC showed superior properties to the standard design in many ways and operating twice as much as previous research on the similar clock frequency. We are experts in that the proposed MAC can adapt to various fields that require high performance, such as signal processing areas.
