Spatial Unfolding of Digit-Recurrence Division for Deterministic-Latency Field-Programmable Gate Array Architectures

2026

Full text Open access: Gold Metadata (CSL-JSON)
DOI 10.30919/es2585
Data confidence 70 medium
How this score is calculated

It is built from checkable properties of the record itself: whether the article has a DOI, how many independent sources confirmed its data, whether the required fields are filled in, and how recent the latest confirmation is. Disagreement between sources and a missing identifier lower it. A value of 80 or above means the record was published without further checking; 60 to 79 means it is published at reduced confidence, with a manual check advisable. A record verified by an editor by hand is never pushed below 90 by a recalculation. This scores our data about the article, not the article itself.

Abstract

Deterministic execution latency is a critical requirement for arithmetic units deployed in real-time and safety-critical digital systems.Conventional radix-4 digit-recurrence dividers generate successive quotient digits through dependent iterations; although pipelined tatations improve throughput, they retain multi-cycle latency, pipeline registers, and control logic, limiting their suitability for applications that demand predictable timing.This paper presents a fully combinational (arrayorganized) radix-4 digit-recurrence divider in which the recursive digit dependency is mapped onto a fixed-depth cascade of identical combinational stages, yielding single-cycle, operand-independent latency while preserving the radix-4 arithmetic formulation unchanged.We refer to this transformation as spatial unfolding.Three architectural features keep the fully spatial realization compact.The divisor multiples B, 2B and 3B and their complements are generated once and broadcast to all stages.Quotient selection is performed by parallel comparators, avoiding lookup tables.Subtraction is realized as complement-add that reuses the comparator outputs, removing dedicated subtractor circuitry.The architecture was described in parameterized Verilog HDL and implemented on a Xilinx Artix-7 XC7A100T-1CSG324C FPGA for operand widths from 8 to 32 bits.It attains operating frequencies from 57.14 MHz (8-bit) to 16.95 MHz (32-bit), occupies only 53-1080 LUTs and 24-98 flip-flops with no DSP blocks or embedded memory, and dissipates 0.010-0.020W