Sensing devices might include a first voltage node configured to receive a first voltage level, a second voltage node configured to receive a second voltage level lower than the first voltage level, a p-type field-effect transistor (pFET) selectively connected to a data line, and a sense node selectively connected to the pFET. The pFET might be connected between the first voltage node and the data line, between the second voltage node and the data line, and between the first voltage node and the data line. Memories might have controllers configured to cause the memories to determine whether a memory cell has an intended threshold voltage using similar sensing devices.

A nonvolatile memory device is provided. The device comprises an active region, a floating gate over the active region and a wordline next to the floating gate. The floating gate has at least two narrow tips adjacent to the wordline and a portion of the floating gate between the narrow tips has a concave profile.

Numerous embodiments are disclosed for accessing redundant non-volatile memory cells in place of one or more rows or columns containing one or more faulty non-volatile memory cells during a program, erase, read, or neural read operation in an analog neural memory system used in a deep learning artificial neural network.

A neural network device with synapses having memory cells each having a floating gate and a first gate over first and second portions of a channel region disposed between source and drain regions, and a second gate over the floating gate or the source region. First lines each electrically connect the first gates in one of the memory cell rows, second lines each electrically connect the second gates in one of the memory cell rows, third lines each electrically connect the source regions in one of the memory cell columns, and fourth lines each electrically connect the drain regions in one of the memory cell columns. The synapses receive a first plurality of inputs as electrical voltages on the first or second lines, and provide a first plurality of outputs as electrical currents on the third or fourth lines.

A parallel-to-serial conversion circuit includes: parallel branches, each including first input end, second input end, control ends and output end, where the first input end is configured to receive high level signal, the second input end is configured to receive low level signal, the control ends are connected to selection unit and the output end is connected to a serial wire, and the selection unit is configured to receive selection signal and at least two branch signals, and is configured to select, based on the selection signal, one of the branch signals and transmit a selected branch signal to the parallel branch; the serial wire, configured to organize signals output by the parallel branches into a serial signal; and a drive unit, connected to the serial wire for enhancing drive capability of the serial wire, where an output end of the drive unit is configured to output the serial signal.

A memory device and method for a non-volatile memory cell having a gate that includes programming the memory cell to an initial program state corresponding to a target read current and a threshold voltage, including applying a program voltage having a first value to the gate, storing the first value in a memory, reading the memory cell in a first read operation using a read voltage applied to the gate that is less than the target threshold voltage to generate a first read current, and subjecting the memory cell to additional programming in response to determining that the first read current is greater than the target read current. The additional programming includes retrieving the first value from the memory, determining a second value greater than the first value, and programming the selected non-volatile memory cell that includes applying a program voltage having the second value to the gate.

A method of programing a memory device having a plurality of memory cell groups where each of the memory cell group includes N non-volatile memory cells, where N is an integer greater than or equal to 2. For each memory cell group, the method includes programming each of the non-volatile memory cells in the memory cell group to a particular program state, performing multiple read operations on each of the non-volatile memory cells in the memory cell group, identifying one of the non-volatile memory cells in the memory cell group that exhibits a lowest read variance during the multiple read operations, deeply programming all of the non-volatile memory cells in the memory cell group except the identified non-volatile memory cell, and programming the identified non-volatile memory cell in the memory cell group with user data.

Numerous embodiments are disclosed for providing temperature compensation in an analog memory array. A method and related system are disclosed for compensating for temperature changes in an array of memory cells by measuring an operating temperature within the array of memory cells and changing a threshold voltage of a selected memory cell in the array of memory cells to compensate for a change in the operating temperature.

A multiple input device is disclosed. The multiple input device includes a semiconductor structure extending in a first direction, a first dielectric material surrounding a portion of the semiconductor structure, a floating gate on the first dielectric material and surrounding the portion of the semiconductor structure, and a second dielectric material on the floating gate and surrounding the portion of the semiconductor structure. The multiple input device also includes a plurality of control gates on the second dielectric material. At least one of the control gates extends vertically away from the semiconductor structure in a second direction and at least one of the control gates extends vertically away from the semiconductor structure in a third direction.

Numerous examples of a precision programming apparatus are disclosed for precisely and quickly depositing the correct amount of charge on the floating gate of a non-volatile memory cell within a vector-by-matrix multiplication (VMM) array in an artificial neural network. In one example, a neuron output circuit for providing a current to program as a weight value in a selected memory cell in a vector-by-matrix multiplication array is disclosed, the neuron output circuit comprising a first adjustable current source to generate a scaled current in response to a neuron current to implement a positive weight, and a second adjustable current source to generate a scaled current in response to a neuron current to implement a negative weight.