Abstract: Embodiments of the present disclosure are directed toward probabilistic in-memory computing configurations and arrangements, and configurations of probabilistic bit devices (p-bits) for probabilistic in-memory computing. concept with emerging. A probabilistic in-memory computing device includes an array of p-bits, where each p-bit is disposed at or near horizontal and vertical wires. Each p-bit is a time-varying resistor that has a time-varying resistance, which follows a desired probability distribution. The time-varying resistance of each p-bit represents a weight in a weight matrix of a stochastic neural network. During operation, an input voltage is applied to the horizontal wires to control the current through each p-bit. The currents are accumulated in the vertical wires thereby performing respective multiply-and-accumulative (MAC) operations. Other embodiments may be described and/or claimed.

Abstract: A data serializer, a latch data device using the same and a controlling method thereof are provided. The data serializer includes at least one data buffer and a de-skew buffer. The data buffer at least receives an inputting data and a controlling signal. An outputting signal and a complementary outputting signal, which is complementary to the outputting signal, are formed when the controlling signal is at a predetermined level. The de-skew buffer receives the complementary outputting signal to accelerate or slow down forming the outputting signal.

Abstract: Numerous embodiments are disclosed for programmable output blocks for use with a VMM array within an artificial neural network. In one embodiment, the gain of an output block can be configured by a configuration signal. In another embodiment, the resolution of an ADC in the output block can be configured by a configuration signal.

Abstract: Embodiments of the present disclosure provide an apparatus including: a sense amplifier coupled to a memory array and having a set of output terminals, a latch coupled to a first output terminal of the sense amplifier, and a comparator coupled to the latch and a second output terminal of the sense amplifier.

Abstract: A memory and a read and write method of memory can prevent the magnetic random-access memory (MRAM) from being easily damaged or degraded by excessive write current during use, and increase memory integration density. The memory includes: a storage unit, comprising a storage element; a source line, electrically connected to a first end of the storage element; the memory is configured to change a storage state of the storage element by a first current and a second current, the first current flowing through the storage element and the second current flowing through the source line without flowing through the storage element.

Abstract: An integrated circuit structure includes: a well region having a first conductivity type; a semiconductor structure extending away from the well region from a major surface of the well region, the semiconductor structure having the first conductivity type; a source/drain feature disposed on the semiconductor structure, the source/drain feature having a second conductivity type different from the first conductivity type; an isolation layer laterally surrounding at least a portion of the semiconductor structure; a dielectric layer disposed on the isolation layer, where at least a portion of the source/drain feature is disposed in the dielectric layer; and a conductive plug continuously extending through the dielectric layer and the isolation layer to physically contact the major surface of the well region, wherein the conductive plug is coupled to a power supply line to bias the well region.

Abstract: Techniques in advanced deep learning provide improvements in one or more of accuracy, performance, and energy efficiency. An array of processing elements performs flow-based computations on wavelets of data. Each processing element has a compute element and a routing element. Each router enables communication via wavelets with at least nearest neighbors in a 2D mesh. Routing is controlled by virtual channel specifiers in each wavelet and routing configuration information in each router. Execution of an activate instruction or completion of a fabric vector operation activates one of the virtual channels. A virtual channel is selected from a pool comprising previously activated virtual channels and virtual channels associated with previously received wavelets. A task corresponding to the selected virtual channel is activated by executing instructions corresponding to the selected virtual channel.

Abstract: Memory devices, memory systems, and methods of operating the same are disclosed in which a memory device, in response to receiving a mode register read (MRR) command directed to one or more write-only bits of a mode register, reads data indicative of a status of the memory device about the memory device from one or more cells of a memory array of the memory device that are different from the write-only mode register. The data can include device settings, environmental conditions, usage statistics, metadata, feature support, feature implementation, device status, temperature, etc. The status information mode can be optionally enabled or disabled. The memory devices can include DDR5 DRAM memory devices.

Abstract: One or more data units at a memory device and that are associated with one or more dice of a die group comprising a plurality of dice are programmed. A voltage offset bin associated with the plurality of dice in the die group is determined based on a subset of dice of the die group.

Abstract: Methods, systems, and devices for clock locking for frame-based communications of memory devices are described. A memory system may include a memory device and a host device. The memory device may receive one or more frames of data from the host device, the one or more frames of data communicated by the host device using a first frame clock. The memory device may generate a second frame clock aligned with the one or more frames on receiving the one or more frames and align one or more operations of the memory device with the second frame clock. In some examples, the host device may receive a second set of frames from the memory device based on transmitting the first set of frames. The host device may align one or more operations of the host device with the second set of frames received from the memory device.

Abstract: Magnetoelectric or magnetoresistive memory cells may include a plurality of reference layers and optionally a plurality of free layers to enhance the tunneling magnetoresistance (TMR) ratio.

Abstract: A substrate includes a plurality of memory units each including an anti-fuse element and a switching element configured to switch application of a predetermined voltage to the anti-fuse element, a wiring to which the plurality of memory units are connected, a first electrode pad to which a voltage for supplying the predetermined voltage to the wiring is applied, and a second electrode pad to which a voltage for supplying the predetermined voltage to the wiring is applied.

Abstract: A memory device includes a first plurality of volatile memories, a non-volatile memory, and a controller coupled to the non-volatile memory and including a first controller output. The memory device further includes a registering clock driver (RCD) including a first RCD output, and a first multiplexer including a first mux input coupled to the first RCD output, a second mux input coupled to the first controller output, and a first mux output coupled to the first plurality of volatile memories. The first multiplexer can be configured to provide command/address signals from one of the RCD and the controller to the first plurality of volatile memories.

Abstract: A dual-port, dual function memory device can be configured to perform operations on data written to the memory device using artificial intelligence (AI) circuitry, such as a neuromorphic array and/or a deep learning accelerator (DLA), of the memory device. The memory device can include a port dedicated for communication between the AI circuitry and a host device and another port dedicated for communication between a memory array of the memory device and a host device. Performing operations, such as image processing operations, using AI circuitry of a memory device can reduce data transfers, reduce resource consumption, and offload workloads from a host device.

Abstract: A processing-in-memory (PIM) device includes a multiplication-and-accumulation (MAC) circuit, a memory circuit, and an address pipeline circuit. The MAC circuit is configured to perform a MAC arithmetic operation or an element-wise multiplication (EWM) calculation for first input data and second input data to generate result data. The memory circuit is configured to output the first input data and the second input data to the MAC circuit in response to a read control signal and is configured to store the result data in response to a write control signal. The address pipeline circuit is configured to receive the read control signal to store an address signal used as a target address signal for designating a region of the memory circuit into which the result data are stored. In addition, the address pipeline circuit is configured to receive the write control signal to output the target address signal to the memory circuit.

Abstract: The present disclosure generally relates to circuit architectures for programming and accessing resistive change elements. The circuit architectures can program and access resistive change elements using neutral voltage conditions. The present disclosure also relates to methods for programming and accessing resistive change elements using neutral voltage conditions. The present disclosure additionally relates to sense amplifiers configurable into initializing configurations for initializing the sense amplifiers and comparing configurations for comparing voltages received by the sense amplifiers. The sense amplifiers can be included in the circuit architectures of the present disclosure.

Abstract: A memory bit cell includes a first memory cell including a first antifuse transistor and a first selection transistor, the first antifuse transistor being selectable between a first state or a second state in response to a word line program signal, the first selection transistor being configured to provide access to the first antifuse transistor in response to a word line read signal; a second memory cell including a second antifuse transistor and a second selection transistor, the second antifuse transistor being selectable between the first state or the second state in response to the word line program signal, the second selection transistor being configured to provide access to the second antifuse transistor in response to the word line read signal; a first word line to selectively provide the word line program signal; a second word line to selectively provide the word line read signal; and a bit line.

Abstract: A memory device and an operation method thereof are provided. The memory device includes: a memory array including a plurality of memory cells for storing a plurality of weights; a multiplication circuit coupled to the memory array, for performing bitwise multiplication on a plurality of input data and the weights to generate a plurality of multiplication results; a counting unit coupled to the multiplication circuit, for performing bitwise counting on the multiplication results to generate a MAC (multiplication and accumulation) operation result.

Abstract: Embodiments provide a data processing circuit and a device. The circuit includes: a first bank group 301 and a second bank group 302, a write circuit 303 including one write input buffer circuit 3031, and a write circuit 304 including one write input buffer circuit 3041. The two write circuits 303 and 304 are configured to: receive stored data from a same write bus 306 by means of the write input buffer circuits 3031 and 3041 respectively, write the stored data into the first bank group 301 by means of a first read-write bus 307, and write the stored data into the second bank group 302 by means of a second read-write bus 308. Frequencies of control signals employed by the two write input buffer circuits 3031 and 3041 both are half of a clock frequency configured for writing the stored data by the write bus 306.

Abstract: A memory system is provided. The memory system includes a controller configured to refresh a memory array at a first temperature before a first refresh time that is acquired from a lookup table and corresponds to a time period for stored data in the memory array being lost at the first temperature. After the controller acquires a second refresh time from the lookup table, the controller resets a refresh time period to refresh the memory array before the second refresh time. The second refresh time corresponds to a time period for stored data in the memory array being lost at a second temperature different from the first temperature. The refresh time period corresponds to a time period after refreshing the memory array.