Substitute-and-simplify: a unified design paradigm for approximate and quality configurable circuits

摘要: Many applications are inherently resilient to inexactness or approximations in their underlying computations. Approximate circuit design is an emerging paradigm that exploits this inherent resilience realize hardware implementations highly efficient energy performance. In work, we propose Substitute-And-SIMplIfy (SASIMI), a new systematic approach the and synthesis of approximate circuits. The key insight behind SASIMI identify signal pairs assume same value with high probability, substitute one for other. While these substitutions introduce functional approximations, if performed judiciously, they result some logic be eliminated from while also enabling downsizing gates on critical paths (simplification), resulting significant power savings. We automatic framework performs substitution simplification iteratively, ensuring user-specified quality constraint satisfied. extend proposed perform configurable circuits can dynamically operate at different accuracy levels depending application requirements. used automatically synthesize wide range arithmetic units (Adders, Multipliers, MAC), complex data (SAD, FFT butterfly, Euclidean distance) ISCAS85 benchmarks, using various error metrics such as rate average magnitude. synthesized demonstrate improvements 10%–28% tight constraints, 30%–60% relaxed constraints. obtain between 14%–40% improvement mode, incurring no overheads accurate mode.