Abstract:Multiplications have been commonly conducted in quantized CNNs, filters, and reconfigurable cores, and so on, which are widely deployed in mobile and embedded applications. Most multipliers are designed to perform multiplications with symmetric bit-widths, i.e., n - by n -bit multiplication. Such features would cause extra area overhead and performance loss when m - by n -bit multiplications ( m > n ) are deployed in the same hardware design, resulting in inefficient multiplication operations. It is highly desired and challenging to propose a reconfigurable multiplier design to accommodate operands with both symmetric and asymmetric bit-widths. In this work, we propose a reconfigurable approximate multiplier to support multiplications at various precisions, i.e., bit-widths. Unlike prior works of approximate adders assuming a uniform weight distribution with bit-wise independence, scenarios like a quantized CNN may have a centralized weight distribution and hence follow a Gaussian-like distribution with correlated adjacent bits. Thus, a new block-based approximate adder is also proposed as part of the multiplier to ensure energy-efficient operation with an awareness of the bit-wise correlation. Our experimental results show that the proposed approximate adder significantly reduces the error rate by 76% to 98% over a state-of-the-art approximate adder for Gaussian-like distribution scenarios. Evaluation results show that the proposed multiplier is 19% faster and 22% more power saving than a Xilinx multiplier IP at the same bit precision and achieves a 23.94-dB peak signal-to-noise ratio, which is comparable to the accurate one of 24.10 dB when deployed in a Gaussian filter for image processing tasks.

A Survey of Some Recent Bit-Parallel GF(2n) Multipliers

Low complexity bit parallel multiplier for GF(2m) generated by equally-spaced trinomials

Research of Fast Modular Multiplier for a Class of Finite Fields

Implementation of Bit-Parallel Multiplier over Finite Field

A New Approach to Subquadratic Space Complexity Parallel Multipliers for Extended Binary Fields

Efficient bit-parallel multiplier for finite field of trinomial using SPB and its weakly dual basis.

High Quality Multiplier Design Based on Improved Circuits

Fast Bit Parallel-Shifted Polynomial Basis Multipliers in $Gf(2^{n})$

Low Complexity Bit-Parallel Normal Bases Multipliers for Gf(2(N))

A Chinese Remainder Theorem Approach to Bit-Parallel GF(2^n) Polynomial Basis Multipliers for Irreducible Trinomials

A Reconfigurable Multiplier for Signed Multiplications with Asymmetric Bit-Widths.

Fast Bit-Parallel GF(2^n) Multiplier for All Trinomials

Scalable Karatsuba multiplier over finite field GF (2m )

Normal Basis Multiplication Algorithms for GF(2n) (full Version).

Low space complexity CRT-based bit-parallel (2) polynomial basis multipliers for irreducible trinomials.

Alternative to the Karatsuba Algorithm for Software Implementations of Gf(2(N)) Multiplications

Performance Comparison of Finite Field Multipliers for SM2 Algorithm Based on FPGA Implementation

Towards Low Space Complexity Design of Gaussian Normal Basis Multiplication

Parallel Integer Polynomial Multiplication

New GF(2n) Parallel Multiplier Using Redundant Representation

Relationship Between GF(2^m) Montgomery and Shifted Polynomial Basis Multiplication Algorithms