Mixed-signal AI accelerators offer the possibility of higher energy efficiency for moderate resolution computations compared to their digital counterparts. All-analog implementations, where all operations are performed in the analog domain, can further improve this energy advantage. An energy efficient multiply-accumulate cell for all-analog neural layer processing macros is presented. The proposed analog two-quadrant multiplier circuit consists of two complementary MOSFETs where the pulse width modulated input activation is applied to the gates and the weight signal to the isolated back-gate. The analog multi-bit resolution weight is dynamically stored on a memory capacitor. The multiply-accumulate operation result is represented by charge accumulated on a summation line and drawn from or put onto a computation capacitance. Simulation results based on a 22 nm FD-SOI CMOS technology show that the cell consumes about 0.67 fJ for a circuit-level multiply-accumulate operation. An area efficiency of 166 $\times$ 10 $^{12}$ MAC/s/mm $^2$ is achieved.
30 October 2025 in IEEE Solid-State Circuits Letters
Analog computing offers intrinsic energy and latency benefits that makes it attractive for real-time and edge applications. Conventional analog accelerators suffer from repeated conversions between analog and digital domain, which degrades efficiency and throughput. We propose an all-analog pipelined neural network accelerator architecture in 22 nm FD-SOI CMOS. Measurements of a demonstrator ASIC with analog I/Os and 6 bit weights are presented. The system energy efficiency is 290 TOPS/W or 558 TOPS/W if the energy for bias generation is neglected. The pipelined architecture achieves a throughput of 500M inferences/s and a latency of 1 ns/layer.