234results about "Electric analogue stores" patented technology

In an implementation, a device includes: a first inverter; a first sense amplifier connected to the first inverter; a first match line connected to the first sense amplifier; a first analog content addressable memory cell including: a first pull-down transistor; a second pull-down transistor, the first pull-down transistor and the second pull-down transistor being connected in series between the first match line and a reference node; a first memory circuit configured to store a first upper bound of a first range and to activate the first pull-down transistor when an analog input value is less than the first upper bound of the first range; and a second memory circuit configured to store a first lower bound of the first range and to activate the second pull-down transistor when the analog input value is greater than the first lower bound of the first range.

The basis of calculating memory capacity of modern computing is bit. Thus, the number of bits is the unit of information quantity of modern communication. The number of neurons (node number) in the neural networks in human brain is not the unit of the memory capacity of the human being. The complexity of neural network is much greater than the bit capacity. Hence, the current AI, which tries to imitate the human brain using computing with the basis on bits, performs inherently different processing of information from the human brain. In addition, computing based on bit number is always facing the limitation of integration. The present disclosure provides a system of information memory without relying on bits using three-dimensional neural networks. By replacing the electrical connection of non-volatile memory cells, which are distributed in a three-dimensional array, the mechanism of the information processing of the human brain can be imitated.

A method is disclosed for programming analog-valued weights in a non-volatile memory array using feedback-based estimation and adaptive pulse modulation. The method includes initializing a target weight, applying one or more programming pulses to a memory cell, estimating a weight state of the memory cell based on analog-to-digital conversion, determining a residual error between the estimated weight state and the target weight, and computing a subsequent programming pulse based on the residual error. The subsequent pulse may be adjusted by selecting a programming voltage from a quantized set of levels responsive to the recent weight response dynamics. The programming process may iterate until convergence may be achieved within a defined threshold. The technique supports precise control of conductance programming using closed-loop feedback and may be compatible with pulse-width, amplitude, and polarity modulation schemes.

A multi-reference-voltage sampled-bandgap system is provided in which a sample-and-hold circuit samples a bandgap reference voltage to produce a sampled bandgap reference voltage. The sample-and-hold circuit includes a voltage divider to divide the sampled bandgap reference into an at least one divided reference voltage.

The present disclosure relates to the field of memory technologies and provides a data sampling circuit, comprising a first sampling module configured to respond to a signal from the data signal terminal and a signal from the reference signal terminal and to act on the first node and the second node; a second sampling module configured to respond to a signal from the first node and a signal from the second node and to act on the third node and the fourth node; a latch module configured to input a high level to the first output terminal and input a low level to the second output terminal, or to input the low-level signal to the first output terminal and input the high-level signal to the second output terminal according to a signal from the third node and a signal from the fourth node; and an offset compensation module connected in parallel to the second sampling module and configured to compensate an offset voltage of the second sampling module. In this data sampling circuit, the second sampling module is additionally provided, and the offset compensation module is connected in parallel to the second sampling module, such that kickback noise is reduced.

In an aspect, a circuit includes a first bootstrap capacitor configured to sample a varying input voltage during a first sample period. In an aspect, the circuit further includes a second bootstrap capacitor configured to sample the varying input voltage during a second sample period. In an aspect, the circuit also includes a switching circuit configured to connect and disconnect the first bootstrap capacitor and the second bootstrap capacitor from the varying input voltage during at least one hold period. In an aspect, the circuit additionally includes an input circuit configured to provide the varying input voltage to the first bootstrap capacitor and the second bootstrap capacitor responsive to a connection configuration of the switching circuit. The input circuit includes three complementary switch pairs. At least two of the three switch pairs are configured to share a common node that is, in turn, configured to receive the varying input voltage.

A system includes a bootstrap circuit having an input and an output. The bootstrap circuit includes a boost capacitor having a first terminal and a second terminal, a first transistor coupled between the first terminal of the boost capacitor and the output of the bootstrap circuit, a second transistor, and a third transistor, wherein the second transistor and the third transistor are coupled in series between a gate of the first transistor and the second terminal of the boost capacitor. The system also includes a switch transistor, wherein a gate of the switch transistor is coupled to the output of the bootstrap circuit, and a terminal of the switch transistor is coupled to the input of the bootstrap circuit.

The present disclosure relates to a data sampling circuit and a data sampling device. The sampling circuit includes: a first sampling module configured to respond to a signal from a data signal terminal and a signal from a reference signal terminal and to act on a first node and a second node; a second sampling module configured to respond to a signal from the first node and a signal from the second node and to act on a third node and a fourth node; a latch module configured to, according to a signal from the third node and a signal from the fourth node, input high level and low level signals to a first output terminal and a second output terminal; and a decision feedback equalization module configured to reduce intersymbol interference. The data sampling circuit provided in the present disclosure can reduce intersymbol interference and have lower power consumption.

The present disclosure provides for crossbar circuits with minimized write disturbance. A crossbar circuit may include a plurality of bit lines intersecting with a plurality of word lines, a plurality of select lines, a plurality of cross-point devices, and a ramp-rate adjustable DAC that comprises a control circuit and an operational amplifier. An input of the operational amplifier is connected to a capacitor. The control circuit may generate, based on a digital input, a control signal. To program a cross-point device of the crossbar circuit, the capacitor may be charged using a reference current. As the charging rate of the capacitor is limited by the reference current, and the charging duration is controlled by the control signal, thus the output of the operational amplifier corresponds to the digital input and may be applied to the cross-point device as a programming signal with limited slew-rate adjustable by the reference current.

The present invention discloses various embodiments of a word line decoder, a control gate decoder, a bit line decoder, a low voltage row decoder, and a high voltage row decoder, as well as various types of physical layout designs for simulating a non-volatile flash memory array in a neural system. The present invention discloses shared and segmented embodiments of a high voltage row decoder.

In one example, a system comprises an array of non-volatile memory cells arranged into rows and columns, each non-volatile memory cell comprising a first bit line terminal, a first erase gate terminal, a first control gate terminal, a first floating gate, a word line, a second floating gate, a second control gate terminal, a second erase gate terminal, and a second bit line terminal, wherein the first floating gate can store a first digital or analog value and the second floating gate can store a second digital or analog value; and a bit line decoder for a column to selectively provide a first voltage to the first bit line terminals of non-volatile memory cells in the column and a second voltage to the second bit line terminals of the non-volatile memory cells in the column.

A top-plate sampling circuitry (10) for sampling a voltage input signal (11), the circuitry comprising: a sampling transistor (110) and a sampling capacitor (120); the sampling transistor being operable as a switch by a gate node (14) according to a tracking phase and a holding phase; a bootstrapper sub-circuitry (140) configured to bootstrap the voltage input signal with a fixed surplus voltage to a bootstrapped voltage (155), and to, during the tracking phase, provide the bootstrapped voltage to the gate node; and a gate switch (190) between the gate node and a fixed reference voltage (160), wherein the gate switch is configured to transition from open to closed during the switching, thereby opening the sampling transistor at a sampling time; wherein the circuitry is further configured to, during the switching from the tracking phase to the holding phase, provide a resistive coupling (180) between the bootstrapped voltage and the gate node.

A recursive Analog Content Addressable Memory (ACAM) device includes a first comparison stage and a second comparison stage. The first comparison stage includes a first match line, a first base segment comparator connected between the first match line and a reference node, a first incremented segment comparator, and a first pass transistor. The first pass transistor and the first incremented segment comparator are connected in series between the first match line and the reference node. The second comparison stage includes a first inverter connected to the first pass transistor, a second match line connected to the first inverter, and a second base segment comparator connected between the second match line and the reference node. The device enables efficient comparison of multi-bit values.

Embodiments provide a storage device that can improve characteristics of the storage device. According to one embodiment, a device includes a sense amplifier that senses a first signal based on first data in a cell and a second signal based on second data in the cell. The sense amplifier includes a current mirror that causes a first current to flow into a first node connected to the cell and causes a second current to flow into a second node based on a potential of the first node, a first switch connected to the second node and a third node, a transistor including a terminal connected to the second node and a gate connected to the third node, a second switch connected to the second node and a fourth node, and a circuit connected to the second node and the third node and causing a third current to flow into the second node based on a potential of the third node.

To provide an information storage system which simulates a neural network on a silicon chip without using a bit.SOLUTION: A neural network has: a plurality of islands periodically disposed in a first axis direction and a second axis direction on a surface of a semiconductor device; a first link and a second link disposed between two islands adjacent to each other in the first direction and the second direction among the plurality of islands; and a first word line in a third axis direction; and a second word line in a fourth axis direction. Each of the plurality of islands has a diffusion layer formed on the surface of the semiconductor device. The first link is a first selection gate for bridging the two islands adjacent to each other in the first direction among the plurality of islands. The second link is a second selection gate for bridging the two islands adjacent to each other in the second direction among the plurality of islands. The first selection gate and the second selection gate have respective selection gate contacts. The selection gate contact is selected by the first word line and the second word line.SELECTED DRAWING: Figure 7

An analog-to-digital converter (ADC) circuit includes a signal input terminal (101), a sample-and-hold circuit (110), and a successive approximation register (SAR) ADC (112). The sample-and-hold circuit (110) includes an input terminal (110A) coupled to the signal input terminal (101). The SAR ADC (112) includes a comparator, a first capacitive digital-to-analog converter (CDAC), and a second CDAC. The first CDAC includes a first input terminal coupled to the signal input terminal, a second input terminal coupled to an output terminal of the sample-and-hold circuit, and an output terminal coupled to a first input terminal of the comparator. The second CDAC includes a first input terminal coupled to the signal input terminal, an output terminal coupled to a second input terminal of the comparator.

A sample and hold circuit includes a sampling circuit including a first amplifier configured to amplify an input voltage to generate an amplification voltage, the sampling circuit configured to perform a sampling operation of sampling the amplification voltage. The sample and hold circuit also includes a holding circuit configured to perform a holding operation of setting an output voltage to a voltage level of the input voltage, based on the sampling operation and an amplification operation of a second amplifier.

The invention relates to a memcapacitor element (1) for operating under a non-linear capacitance-voltage curve (2). The memcapacitor element (1) comprises a plurality of electrical contacts for making electrical contact with the memcapacitor element, a first electrode (3a) and a second electrode (3b). The memcapacitor element (1) further comprises at least one dielectric layer (5a, 5b) arranged between the first electrode and the second electrode, where the electrical contacts, the electrodes (3a, 3b) and / or the at least one dielectric layer (5a, 5b) have charge trapping sites forming potential wells of different depths for trapping different numbers of charges. In addition, the invention also relates to a method for operating the memcapacitor element.

A reference buffer is disclosed. The reference buffer includes an amplifier configured to amplify a difference between a reference output voltage and a reference input voltage during a closed-loop phase of the reference buffer. The reference buffer also includes a sample circuit configured to sample an amplifier output voltage during the closed-loop phase. The reference buffer further includes a buffer circuit having a buffer input coupled to the sample circuit and a buffer output configured to provide the reference output voltage. The reference buffer also includes first and second switch circuits. The first switch circuit is configured to isolate the sample circuit from the output stage of the amplifier during an open-loop phase of the reference buffer. The second switch circuit is configured to isolate the input stage from the output of the reference buffer during the open-loop phase.

A successive approximation register analog-to-digital converter may include a reference digital-to-analog converter, a successive approximation register, and a top-plate sampled comparator configured to compare a reference signal generated by the reference digital-to-analog converter to an analog input signal in order to generate a comparator output to the successive approximation register. The top-plate sampled comparator may include sampling transistors, sampling switches, non-linear capacitors, and circuitry configured to pre-condition the comparator by maintaining a constant voltage across non-linear capacitors of the top-plate sampled comparator during a sampling phase of the comparator in order to maintain sampling linearity of the top-plate sampled comparator.

To improve an operation speed of a circuit.SOLUTION: There are provided: a first transistor; a second transistor in which a first terminal is connected to a gate of the first transistor, and has a function for setting a value of a potential of the gate of the first transistor to a value where the first transistor is ON; a third transistor that sets a value of the potential of the gate of the second transistor to a value where the second transistor is ON, and has a function for setting the gate of the second transistor to be in a floating state; and a fourth transistor that has a function for setting a value of the potential of the gate of the second transistor to a value where the second transistor is OFF. By such a construction, a potential difference between the gate and a source of the second transistor can be maintained so as to be a value that is larger than a threshold value voltage of the second transistor, and can improve an operation speed.SELECTED DRAWING: Figure 1

A compute-in-memory (CIM) circuit includes a processing circuit. The processing circuit includes a data-selection circuit and a charge-domain passive summation circuit. The data-selection circuit includes a memory array and a selection circuit. The memory array stores a plurality of candidate weights. The selection circuit selects a target weight from the plurality of candidate weights stored in the memory array. The charge-domain passive summation circuit generates an analog computation result of an input received by the processing circuit and the target weight stored in the memory array through a weighted capacitor array integrated with the memory array.

A memory device includes a memory array storing weights; a pre-charging circuit coupled to the memory array through data lines and charging, in response to a pre-charge signal, at least one data line in the data lines to a read voltage in a read operation to one in the weights; and a calibration circuit generating the pre-charge signal according to an address of the one in the weights.

A memristor-based circuit is described in which a voltage generator is arranged to apply a series of voltage pulses to a memristor to progressively change the resistance of the memristor. A comparator is arranged: to receive an input electrical value; to receive an electrical value based on the resistance of the memristor; to compare the received values; and, based on the comparison, to enable the application of the voltage pulses to the memristor by the voltage generator until a defined condition is satisfied. This circuit can be used to enable the memristor to be programmed to a desired resistance value, such as for use as a non-volatile memory. It can also enable the resistance of one memristor to be replicated to another memristor. By counting the number of applied voltage pulses, the circuit can be used as an encoder or analog-to- digital converter. Other variants of the circuit enable construction of a decoder or digital-to-analog converter, and an authentication circuit.

A satisfiability modulo theories (SMT) solver accelerator is implemented with analog content addressable memory (CAM) cells. A job-shop scheduling problem (JSSP) may be mapped to expression clauses that are stored in an analog CAM array. A test cycle may be performed by searching for a test vector of input variables in the analog CAM array, selecting a candidate variable of the test vector that is violating a largest number of the expression clauses, and changing the candidate variable of the test vector. The test cycle may be repeated until none of the expression clauses are violated by the test vector of the input variables. The resulting test vector of the input variables that violates no expression clauses is a solution for the JSSP.

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

To provide configurable input blocks and output blocks and a physical layout for analog neural memory systems that utilize non-volatile memory cells.SOLUTION: In a vector by matrix multiplication (VMM) system 3400, input blocks 3409, 3410 can be configured to support different numbers of arrays 3401 to 3404 arranged in a horizontal direction, and output blocks 3411, 3412 can be configured to support different numbers of arrays arranged in a vertical direction. Adjustable components are disclosed for use in the configurable input blocks and output blocks.SELECTED DRAWING: Figure 34

The present disclosure provides for crossbar circuits with minimized write disturbance. A crossbar circuit may include a plurality of bit lines intersecting with a plurality of word lines, a plurality of select lines, a plurality of cross-point devices, and a ramp-rate adjustable DAC that comprises a control circuit and an operational amplifier. An input of the operational amplifier is connected to a capacitor. The control circuit may generate, based on a digital input, a control signal. To program a cross-point device of the crossbar circuit, the capacitor may be charged using a reference current. As the charging rate of the capacitor is limited by the reference current, and the charging duration is controlled by the control signal, thus the output of the operational amplifier corresponds to the digital input and may be applied to the cross-point device as a programming signal with limited slew-rate adjustable by the reference current.