We propose a design optimization flow for a high-speed and low-power operational transconductance amplifier (OTA) using a gm/ID lookup table design methodology in scaled CMOS. This methodology advantages from using gm/ID as a primary design parameter to consider all operation regions including strong, moderate, and weak inversion regions, and enables the lowest power design. SPICE-based lookup table approach is employed to optimize the operation region specified by the gm/ID with sufficient accuracy for short-channel transistors. The optimized design flow features 1) a proposal of the worst-case design scenario for specification and gm/ID lookup table generations from worst-case SPICE simulations, 2) an optimization procedure accomplished by the combination of analytical and simulation-based approaches in order to eliminate tweaking of circuit parameters, and 3) an additional use of gm/ID subplots to take second-order effects into account. A gain-boosted folded-cascode OTA for a switched capacitor circuit is adopted as a target topology to explore the effectiveness of the proposed design methodology for a circuit with complex topology. Analytical expressions of the gain-boosted folded-cascode OTA in terms of DC gain, frequency response and output noise are presented, and detailed optimization of gm/IDs as well as circuit parameters are illustrated. The optimization flow is verified for the application to a residue amplifier in a 10-bit 125 MS/s pipeline A/D converter implemented in a 0.18 µm CMOS technology. The optimized circuit satisfies the required specification for all corner simulations without additional tweaking of circuit parameters. We finally explore the possibility of applying this design methodology as a technology migration tool, and illustrate the failure analysis by comparing the differences in the gm/ID characteristics.

The development of energy-efficient neural network hardware using magnetic tunnel junction (MTJ) devices has been widely investigated. One of the issues in the use of MTJ devices is large write energy. Since MTJ devices show stochastic behaviors, a large write current with enough time length is required to guarantee the certainty of the information held in MTJ devices. This paper demonstrates that quantized neural networks (QNNs) exhibit high tolerance to bit errors in weights and an output feature map. Since probabilistic switching errors in MTJ devices do not have always a serious effect on the performance of QNNs, large write energy is not required for reliable switching operations of MTJ devices. Based on the evaluation results, we achieve about 80% write-energy reduction on buffer memory compared to the conventional method. In addition, it is demonstrated that binary representation exhibits higher bit-error tolerance than the other data representations in the range of large error rates.

This letter presents an efficient graph-based evolutionary optimization technique, and its application to the transistor-level design of multiple-valued arithmetic circuits. The key idea is to introduce "circuit graphs with colored terminals" for modeling heterogeneous networks of various components. The potential of the proposed approach is demonstrated through experimental synthesis of a radix-4 signed-digit (SD) full adder circuit.

This paper presents an efficient graph-based evolutionary optimization technique called Evolutionary Graph Generation (EGG) and its extension to a parallel version. A new version of parallel EGG system is based on a coarse-grained model of parallel processing and can synthesize heterogeneous networks of various different components efficiently. The potential capability of parallel EGG system is demonstrated through the design of current-mode logic circuits.

A multiple-valued current-mode (MVCM) circuit using current-flow control is proposed for a power-greedy sequential linear-array system. Whenever operation is completed in processing element (PE) at the present stage, every possible current source in the PE at the previous stage is cut off, which greatly reduces the wasted power dissipation due to steady current flows during standby states. The completion of the operation can be easily detected using "operation monitor" that observes input and output signals at latches, and that generates control signal immediately at the time completed. Since the wires of data and control signals are shared in the proposed MVCM circuit, no additional wires are required for current-flow control. In fact, it is demonstrated that the power consumption of the MVCM circuit using the proposed method is reduced to 53 percent in comparison with that without current-source control.