Abstract: Disclosed are methods, apparatus and systems for a binary neural network accelerator engine. One example circuit is designed to perform a multiply-and-accumulate (MAC) operation using logic circuits that include a first set of exclusive nor (XNOR) gates to generate a product vector based on a bit-wise XNOR operation two vectors. The result is folded and operated on by another set of logic circuits that provide an output for a series of adder circuits. The MAC circuit can be implemented as part of binary neural network at a small footprint to effect power and cost savings.

Abstract: The disclosed technology can be used to convert direct-current voltage and current from an input to a different or the same voltage and current at an output. One example direct-current to direct-current (DC-DC) power converter includes a first switch connected between a source voltage and a first side of an inductor, a second switch connected between the first side of the inductor and a ground, a third switch connected between a second side of the inductor and the ground, and a fourth switch connected between the second side of the inductor and a capacitor. The power converter may further include a comparator configured to compare an output voltage at the capacitor to a threshold voltage and based on the result of the comparison selectively activate or deactivate the first, second, third, and fourth switches in a power cycle.

Abstract: Disclosed are methods, devices and systems for all-in-one signal processing, linear and non-linear vector arithmetic accelerator. The accelerator, which in some implementations can operate as a companion co-processor and accelerator to a main system, can be configured to perform various linear and non-linear arithmetic operations, and is customized to provide shorter execution times and fewer task operations for corresponding arithmetic vector operation, thereby providing an overall energy saving. The compact accelerator can be implemented in devices in which energy consumption and footprint of the electronic circuits are important, such as in Internet of Things (IoT) devices, in sensors and as part of artificial intelligence systems.

Abstract: Devices, methods and systems provide improved performance for voltage converters and enable more efficient operations by reducing energy loss in a head switch of the voltage converter. These improvements are achieved in-part by reducing one or both of an ohmic conduction loss and a switching loss of the head switch. One example floating head switch includes an n-type metal oxide semiconductor (NMOS) transistor, a capacitor, one or more drivers and an active low switch. The capacitor ends are connected to the supply voltages of the drivers and the active low switch is coupled to, and controlled by, the output of the one or more drivers and turns on or off in response to a change in the input voltage.

Abstract: Devices, methods, and systems are described that generate process, voltage and temperature tolerant clock generators, which can be used in low power and low cost applications. The clock generators eliminate the need for a crystal oscillator, are simple to implement, and can use a single frequency calibration step to initially tune the frequency to a reference frequency value, and to allow the clock generator to operate in the presence of process, voltage or temperature variations. One example clock circuit includes a voltage-controlled oscillator that provides a clock output, a gain circuit to receive a reference voltage as one input and a changeable voltage on another input. The clock circuit also includes a frequency-to-voltage convertor circuit that receives a reference current and produces the changeable voltage provided to gain circuit, while a ratio of the reference voltage to the reference current is constant.

Abstract: Disclosed are techniques that can be used in a semiconductor chip to determine performance such as timing performance. Among other features, supply voltages and clock rates may be adjusted to accommodate the operating temperature and to compensate for the processing variations that occurred when that chip was produced, or may occur as the chip is used. The techniques include determining a series of variables that affect performance, determining the sensitivity of timing paths in the circuit to each variable, duplicating the most sensitive paths. A novel sensor circuit is produced that includes the sensitive paths, which can be used to determine when the chip is performing as required and when it is not, and adjusting one or more supply voltages and/or clock rates in a static or real time manner when the circuit is not performing as required.

Abstract: Disclosed are low power (e.g., nanowatt) voltage regulator circuits and devices, systems and methods using the same for ultra-low power applications, such as, but not limited to, Internet of things (IoT) applications. The disclosed devices and systems relate to low-dropout (LDO) circuits and methods for constructing and using the same. The disclosed LDOs operate with uniform output frequency characteristics over a wide range of load currents.

Abstract: Methods, devices and systems are described that relate to an energy optimal computing control system, where clock rates and supply voltage are control knobs to adjust the total energy consumption of an electronic circuit. An ultra-wide voltage range, such as from near-threshold voltage to the device maximum voltage, is used to maximize the performance and to minimize the energy consumption. One example device includes a processor that receives or determines the temperature of the electronic circuit, and determines the optimum voltage levels and clock rates for the electronic circuit at the operating temperature. This information is provided to a voltage regulator and a clock generator to adjust the supply voltage and clock frequency accordingly.