Abstract: Methods, apparatuses, and systems for in- or near-memory processing are described. Spiking events in a spiking neural network may be processed via a memory system. A memory system may store a group of destination neurons, and at each time interval in a series of time intervals of a spiking neural network (SNN), pass through a group of pre-synaptic spike events from respective source neurons, wherein the group of pre-synaptic spike events are subsequently stored in memory.

Abstract: Methods and apparatuses for thin film transistors and related fabrication techniques are described. The thin film transistors may access two or more decks of memory cells disposed in a cross-point architecture. The fabrication techniques may use one or more patterns of vias formed at a top layer of a composite stack, which may facilitate building the thin film transistors within the composite stack while using a reduced number of processing steps. Different configurations of the thin film transistors may be built using the fabrication techniques by utilizing different groups of the vias. Further, circuits and components of a memory device (e.g., decoder circuitry, interconnects between aspects of one or more memory arrays) may be constructed using the thin film transistors as described herein along with related via-based fabrication techniques.

Abstract: Methods and apparatuses for a cross-point memory array and related fabrication techniques are described. The fabrication techniques described herein may facilitate concurrently building two or more decks of memory cells disposed in a cross-point architecture. Each deck of memory cells may include a plurality of first access lines (e.g., word lines), a plurality of second access lines (e.g., bit lines), and a memory component at each topological intersection of a first access line and a second access line. The fabrication technique may use a pattern of vias formed at a top layer of a composite stack, which may facilitate building a 3D memory array within the composite stack while using a reduced number of processing steps. The fabrication techniques may also be suitable for forming a socket region where the 3D memory array may be coupled with other components of a memory device.

Abstract: Spiking events in a spiking neural network may be processed via a memory system. A memory system may store data corresponding to a group of destination neurons. The memory system may, at each time interval of a SNN, pass through data corresponding to a group of pre-synaptic spike events from respective source neurons. The data corresponding to the group of pre-synaptic spike events may be subsequently stored in the memory system.

Abstract: Methods and apparatuses for thin film transistors and related fabrication techniques are described. The thin film transistors may access two or more decks of memory cells disposed in a cross-point architecture. The fabrication techniques may use one or more patterns of vias formed at a top layer of a composite stack, which may facilitate building the thin film transistors within the composite stack while using a reduced number of processing steps. Different configurations of the thin film transistors may be built using the fabrication techniques by utilizing different groups of the vias. Further, circuits and components of a memory device (e.g., decoder circuitry, interconnects between aspects of one or more memory arrays) may be constructed using the thin film transistors as described herein along with related via-based fabrication techniques.

Abstract: Systems, methods, and apparatus related to memory devices that perform multiplication using logical states of memory cells. In one approach, a memory cell array has memory cells programmed to store weights for performing the multiplication. Voltages are applied to the memory cells. Each voltage represents one or more input bits to be multiplied by one of the weights. Output currents from the memory cells are accumulated in a common bitline. A sum of the output currents is digitized to provide a digital result. The digital results from several bitlines can be shifted based on bit significance and added to provide a final accumulation result from the multiplication.

Abstract: The present disclosure includes apparatuses, methods, and systems for multi-state programming of memory cells. An embodiment includes a memory having a plurality of memory cells, and circuitry configured to program a memory cell of the plurality of memory cells to one of a plurality of possible data states by applying a voltage pulse to the memory cell, determining the memory cell snaps back in response to the applied voltage pulse, turning off a current to the memory cell upon determining the memory cell snaps back, and applying a number of additional voltage pulses to the memory cell after turning off the current to the memory cell.

Abstract: Methods, systems, and devices for pulse based multi-level cell programming are described. A memory device may identify an intermediate logic state to store to a multi-level memory cell capable of storing three or more logic states. The memory device may apply a first pulse with a first polarity to the memory cell to store a SET or RESET state to the memory cell based on identifying the intermediate logic state. As such, the memory device may identify a threshold voltage of the memory cell that stores the SET or RESET state. The memory device may apply a quantity of pulses to the memory cell to store the identified intermediate logic state based on identifying the threshold voltage of the memory cell that stores the SET or RESET state. In some examples, the quantity of pulses may have a second polarity different than the first polarity.

Abstract: The present disclosure includes apparatuses, methods, and systems for three-state programming of memory cells. An embodiment includes a memory having a plurality of memory cells, and circuitry configured to program a memory cell of the plurality of memory cells to one of three possible data states by applying a voltage pulse to the memory cell, determining whether the memory cell snaps back in response to the applied voltage pulse, and applying an additional voltage pulse to the memory cell based on the determination of whether the memory cell snaps back.

Abstract: Methods, apparatuses, and systems for in- or near-memory processing are described. Spiking events in a spiking neural network may be processed via a memory system. A memory system may store a group of destination neurons, and at each time interval in a series of time intervals of a spiking neural network (SNN), pass through a group of pre-synaptic spike events from respective source neurons, wherein the group of pre-synaptic spike events are subsequently stored in memory.

Abstract: Methods and apparatuses for thin film transistors and related fabrication techniques are described. The thin film transistors may access two or more decks of memory cells disposed in a cross-point architecture. The fabrication techniques may use one or more patterns of vias formed at a top layer of a composite stack, which may facilitate building the thin film transistors within the composite stack while using a reduced number of processing steps. Different configurations of the thin film transistors may be built using the fabrication techniques by utilizing different groups of the vias. Further, circuits and components of a memory device (e.g., decoder circuitry, interconnects between aspects of one or more memory arrays) may be constructed using the thin film transistors as described herein along with related via-based fabrication techniques.

Abstract: The present disclosure includes apparatuses, methods, and systems for three-state programming of memory cells. An embodiment includes a memory having a plurality of memory cells, and circuitry configured to program a memory cell of the plurality of memory cells to one of three possible data states by applying a voltage pulse to the memory cell, determining whether the memory cell snaps back in response to the applied voltage pulse, and applying an additional voltage pulse to the memory cell based on the determination of whether the memory cell snaps back.

Abstract: Methods and apparatuses for thin film transistors and related fabrication techniques are described. The thin film transistors may access two or more decks of memory cells disposed in a cross-point architecture. The fabrication techniques may use one or more patterns of vias formed at a top layer of a composite stack, which may facilitate building the thin film transistors within the composite stack while using a reduced number of processing steps. Different configurations of the thin film transistors may be built using the fabrication techniques by utilizing different groups of the vias. Further, circuits and components of a memory device (e.g., decoder circuitry, interconnects between aspects of one or more memory arrays) may be constructed using the thin film transistors as described herein along with related via-based fabrication techniques.

Abstract: Methods, a memory device, and a system are disclosed to reduce power consumption in a cross-point memory device, including providing a first portion of a first pulse of a memory operation to a memory cell at a first time using a first capacitive discharge from a first discharge path, and providing a second portion of the first pulse of the memory operation to the memory cell at a second time, later than the first time, using a second discharge path.

Abstract: Methods, systems, and devices for edgeless memory clusters are described. Systems, devices, and techniques are described for eliminating gaps between clusters by creating groups (e.g., domains) of clusters that are active at a given time, and using drivers within inactive clusters to perform array termination functions for abutting active clusters. Tiles on the edges of a cluster may have drivers that operate both for the cluster, and for a neighboring cluster, with circuits (e.g., a multiplexers) on the drivers to enable operations for both clusters.

Abstract: Interfaces between higher voltage and lower voltage wafers and related apparatuses and methods are disclosed. An apparatus includes a memory wafer and a logic wafer. Data storage elements of an array are configured to perform an operation responsive to an operational voltage potential. The memory wafer also includes bitlines electrically connected to the data storage elements and isolation devices electrically connected to the bitlines. The logic wafer is bonded to the memory wafer. The logic wafer includes logic circuitry electrically connected to the bitlines through the isolation devices. A maximum voltage potential difference tolerance of the logic circuitry is less than an operational voltage potential difference between the operational voltage potential and a reference voltage potential of the logic circuitry.

Abstract: Techniques are provided for programming a self-selecting memory cell that stores a first logic state. To program the memory cell, a pulse having a first polarity may be applied to the cell, which may result in the memory cell having a reduced threshold voltage. During a duration in which the threshold voltage of the memory cell may be reduced (e.g., during a selection time), a second pulse having a second polarity (e.g., a different polarity) may be applied to the memory cell. Applying the second pulse to the memory cell may result in the memory cell storing a second logic state different than the first logic state.

Abstract: Methods, systems, and devices for pulse based multi-level cell programming are described. A memory device may identify an intermediate logic state to store to a multi-level memory cell capable of storing three or more logic states. The memory device may apply a first pulse with a first polarity to the memory cell to store a SET or RESET state to the memory cell based on identifying the intermediate logic state. As such, the memory device may identify a threshold voltage of the memory cell that stores the SET or RESET state. The memory device may apply a quantity of pulses to the memory cell to store the identified intermediate logic state based on identifying the threshold voltage of the memory cell that stores the SET or RESET state. In some examples, the quantity of pulses may have a second polarity different than the first polarity.

Abstract: Semiconductor memory dies bonded to logic dies and associated systems and methods are disclosed. In an embodiment, a semiconductor die assembly includes a logic die and one or more memory dies directly bonded to the logic die. The logic die includes integrated circuits generated using relatively high temperature process steps whereas the memory dies include memory cells with materials generated using relatively low temperature process steps. The logic die and the memory dies have been separately fabricated in two different wafers such that process steps generating them can be optimized independently of each other. The resulting semiconductor device including the memory dies bonded to the logic die functions as a single device as if they were formed in a monolithic substrate. The resulting semiconductor device may be configured to perform artificial intelligence tasks.

Abstract: Semiconductor die stacks and associated systems and methods are disclosed. In an embodiment, a semiconductor die stack corresponds to a pair of a logic die and a memory die directly bonded together. The logic die includes integrated circuits generated by relatively high temperature process steps whereas the memory die includes memory cells with materials generated using relatively low temperature process steps. A logic wafer including the logic dies and a memory wafer including the memory dies are separately fabricated. Subsequently, the logic wafer and the memory wafer are directly bonded to generate the semiconductor die stacks. Either the logic dies or the memory dies include through-substrate vias (TSVs) to provide power and signals for the semiconductor die stacks. The resulting semiconductor devices operate as a single device as if they were formed in a monolithic substrate.