Methods, systems, and devices for dynamic power distribution for stacked memory are described. A stacked memory device may include switching components that support dynamic coupling between a shared power source of the memory device and circuitry associated with operating memory arrays of respective memory dies. In some examples, such techniques include coupling a power source with array circuitry based on an access activity or a degree of access activity for the array circuitry. In some examples, such techniques include isolating a power source from array circuitry based on a lack of access activity or a degree of access activity for the array circuitry. The dynamic coupling or isolation may be supported by various signaling of the memory device, such as signaling between memory dies, signaling between a memory die and a central controller, or signaling between the memory device and a host device.

Novel connection between neurons of a neural network is provided. A perceptron included in the neural network includes a plurality of neurons; the neuron includes a synapse circuit and an activation function circuit; and the synapse circuit includes a plurality of memory cells. A bit line selected by address information for selecting a memory cell is shared by a plurality of perceptrons. The memory cell is supplied with a weight coefficient of an analog signal, and the synapse circuit is supplied with an input signal. The memory cell multiplies the input signal by the weight coefficient and converts the multiplied result into a first current. The synapse circuit generates a second current by adding a plurality of first currents and converts the second current into a first potential. The activation function circuit is a semiconductor device that converts the first potential into a second potential by a ramp function and supplies the second potential as an input signal of the synapse circuit included in the perceptron in a next stage.

Novel connection between neurons of a neural network is provided. A perceptron included in the neural network includes a plurality of neurons; the neuron includes a synapse circuit and an activation function circuit; and the synapse circuit includes a plurality of memory cells. A bit line selected by address information for selecting a memory cell is shared by a plurality of perceptrons. The memory cell is supplied with a weight coefficient of an analog signal, and the synapse circuit is supplied with an input signal. The memory cell multiplies the input signal by the weight coefficient and converts the multiplied result into a first current. The synapse circuit generates a second current by adding a plurality of first currents and converts the second current into a first potential. The activation function circuit is a semiconductor device that converts the first potential into a second potential by a ramp function and supplies the second potential as an input signal of the synapse circuit included in the perceptron in a next stage.

According to one embodiment, a memory system includes a non-volatile memory and a memory controller. The non-volatile memory includes a plurality of groups, each including a plurality of memory cells. The memory controller is configured to determine whether to execute a refresh process for a first group based on whether a first temperature in a write process for the first group and a second temperature after the write process for the first group satisfy a first condition.

A computer-implemented method includes an act of configuring hardware to cause at least a part of the hardware to operate as a double data rate (DDR) memory controller, and to produce a capture clock to time a read data path, where a timing of the capture clock is based on a first clock signal of a first clock, delay the first clock signal to produce a delayed first clock signal, adjust the delay such that at least one clock edge of the delayed first clock signal is placed nearer to at least one clock edge of at least one data strobe (DQS), or at least one signal dependent on a DQS timing, and produce a modified timing of the capture clock based on the delay of the first clock signal.

An object is to shorten the time for rewriting data in memory cells. A memory module includes a first memory cell, a second memory cell, a selection transistor, and a wiring WBL1. The first memory cell includes a first memory node. The second memory cell includes a second memory node. One end of the first memory cell is electrically connected to the wiring WBL1 through the selection transistor. The other end of the first memory cell is electrically connected to one end of the second memory cell. The other end of the second memory cell is electrically connected to the wiring WBL1. When the selection transistor is on, data in the first memory node is rewritten by a signal supplied through the selection transistor to the wiring WBL1. When the selection transistor is off, data in the first memory node is rewritten by a signal supplied through the second memory node to the wiring WBL1.

An object is to shorten the time for rewriting data in memory cells. A memory module includes a first memory cell, a second memory cell, a selection transistor, and a wiring WBL1. The first memory cell includes a first memory node. The second memory cell includes a second memory node. One end of the first memory cell is electrically connected to the wiring WBL1 through the selection transistor. The other end of the first memory cell is electrically connected to one end of the second memory cell. The other end of the second memory cell is electrically connected to the wiring WBL1. When the selection transistor is on, data in the first memory node is rewritten by a signal supplied through the selection transistor to the wiring WBL1. When the selection transistor is off, data in the first memory node is rewritten by a signal supplied through the second memory node to the wiring WBL1.

A memory device with large storage capacity is provided. A NAND memory device includes a plurality of connected memory elements each provided with a writing transistor and a reading transistor. An oxide semiconductor is used in a semiconductor layer of the writing transistor, whereby a storage capacitor is not necessary or the size of the storage capacitor can be reduced. The reading transistor includes a back gate. When a reading voltage is applied to the back gate, data stored in the memory element is read out.

In a semiconductor device 100, at least one of a first transistor and a second transistor that supply a second voltage in a step-down circuit stepping down a first voltage to the second voltage and outputting the second voltage from an output portion is configured such that the number of second contacts of a source electrode which is connected to a ground voltage or is supplied with the first voltage is larger than the number of first contacts connecting a diffusion layer and a first metal layer of a drain electrode connected to the output portion, and the number of second vias of the source electrode connected to the ground voltage or supplied with the first voltage is larger than the number of first vias connecting the first metal layer and a second metal layer of the drain electrode connected to the output portion.

In a semiconductor device 100, at least one of a first transistor and a second transistor that supply a second voltage in a step-down circuit stepping down a first voltage to the second voltage and outputting the second voltage from an output portion is configured such that the number of second contacts of a source electrode which is connected to a ground voltage or is supplied with the first voltage is larger than the number of first contacts connecting a diffusion layer and a first metal layer of a drain electrode connected to the output portion, and the number of second vias of the source electrode connected to the ground voltage or supplied with the first voltage is larger than the number of first vias connecting the first metal layer and a second metal layer of the drain electrode connected to the output portion.