Numerous embodiments of input circuitry for an analog neural memory in a deep learning artificial neural network are disclosed.

According to some aspects, a low-leakage switch is provided. In some embodiments, the low-leakage switch includes a plurality of pass transistors in series that selectively couple two ports of the low-leakage switch and a node biasing circuit coupled to a node between the plurality of pass transistors. In these embodiments, the node biasing circuit may adjust a voltage at the node to change the gate-to-source voltage of the pass transistors and, thereby, reduce the leakage current through the pass transistors when the low-leakage switch is turned off. The node biasing circuit may also include circuitry to reduce the leakage current introduced by the node biasing circuit into the node when the low-leakage switch is turned on.

Systems, apparatuses and methods include technology that identifies whether a product of first and second digital numbers is associated with a positive value or a negative value. During a first clock phase, the technology sets a first reference voltage to have a first value or a second value based on whether the product is associated with the positive value or the negative value. During the first clock phase, the technology controls switches to supply the first reference voltage to first plates of capacitors. Each of the capacitors includes a respective first plate of the first plates and a second plate. Further, during the first clock phase, the technology controls the switches based on the first digital number to electrically connect at least one of the second plates to the first reference voltage and electrically connect at least one of the second plates to a second reference voltage.

A sampling switch circuit, including an input node, which receives an input voltage signal to be sampled, a sampling transistor having gate, source and drain terminals, the source terminal connected to the input node, a capacitor, a current source configured to cause a defined current to flow therethrough and switching circuitry configured to alternate between a precharge configuration and an output configuration depending upon a clock signal. In the precharge configuration, the switching circuitry connects the capacitor into a current path between said current source and a first voltage reference node to form a potential difference across the capacitor which is dependent on the defined current. In the output configuration, the switching circuitry connects the capacitor between a second voltage reference node and the gate terminal of the sampling transistor so that a voltage level applied at the gate terminal of the sampling transistor is dependent on the defined current.

A sampling switch circuit, comprising an input node, connected to receive an input voltage signal, a sampling transistor comprising a gate terminal, a source terminal and a drain terminal, the source terminal connected to the input node, a hold-control node connected to receive a hold-control voltage signal, an output node connected to the drain terminal of the sampling transistor, a buffer circuit having a buffer input connected to the input node and a buffer output connected to a track-control node, the buffer circuit configured to provide a track-control voltage signal at the track-control node dependent on the input voltage signal and switching circuitry configured to connect the gate terminal of the sampling transistor to the track-control node or to the hold-control node in dependence upon a clock signal.

Various embodiments provide apparatuses, systems, and methods for multibit analog representation, e.g., for in-memory computing. Embodiments may include a single-ended or differential ladder network to generate an analog value (e.g., a voltage or charge) based on a set of bits from a memory array. The ladder network may include a plurality of branches coupled to an output line, wherein individual branches include a capacitor with a first terminal coupled to the output line and a switch coupled to a second terminal of the capacitor. The switch may be controlled by a respective bit of the set of bits to selectively couple the second terminal of the capacitor to a first voltage node or a second voltage node based on a value of the respective bit. Other embodiments may be described and claimed.

To improve programming performance for a non-volatile memory , the verification of multiple programming levels can be performed based on a single discharge of a sensing capacitor through a selected memory cell by using different voltage levels on a second plate of the sensing capacitor: after discharging a first plate of the sensing capacitor through the selected memory cell, a result amount of charge is trapped on the first plate, which is then used to set first and second control gate voltages on a sensing transistor whose control gate is connected to the first place of the sensing capacitor based on respectively setting the second plate of the sensing capacitor to first and second voltage levels. To further improve programming performance, when the non-volatile memory stores in a multistate format, after the next to highest data state finishes programming, the next programming pulse can use a larger step size.

A memory unit with a multiply-accumulate assist scheme for a plurality of multi-bit convolutional neural network based computing-in-memory applications is controlled by a reference voltage, a word line and a multi-bit input voltage. The memory unit includes a non-volatile memory cell, a voltage divider and a voltage keeper. The non-volatile memory cell is controlled by the word line and stores a weight. The voltage divider includes a data line and generates a charge current on the data line according to the reference voltage, and a voltage level of the data line is generated by the non-volatile memory cell and the charge current. The voltage keeper generates an output current on an output node according to the multi-bit input voltage and the voltage level of the data line, and the output current is corresponding to the multi-bit input voltage multiplied by the weight.

A method for multiple copies of a set of multi-kernel set operations in a hardware accelerated neural network includes a word line for receiving a pixel value of an input image. A bit line communicates a modified pixel value. An analog memory cell including a first capacitor stores a first kernel weight of a first kernel in one of a plurality of kernel sets such that the pixel value is operated on by the first kernel weight to produce the modified pixel value. A charge connection connects the first capacitor to at least a second capacitor storing a second kernel weight of a related kernel of a second one of the plurality of kernel sets such that charge is shared between the first capacitor and at least the second capacitor to normalize the first kernel weight and the second kernel weight.

A signal sampling method comprising: sampling an electrical signal to be measured using a pre-determined sampling method to obtain a plurality of first sampling points, each of which is represented by a first amplitude and a corresponding first time; measuring a second amplitude of the electrical signal to be measured, wherein the second amplitude is different from the plurality of the first amplitudes; and delaying the electrical signal to be measured, and using the delayed electrical signal to determine a second time when the amplitude of the electrical signal to be measured reaches the second amplitude in order to obtain a second sampling point, which is represented by the second amplitude and the second time. A greater number of sampling points can be sampled, such that the precision of sampling and also the accuracy of subsequent signal restoration may be improved.