Abstract: Read reference levels are calibrated by calibrating integration times. An integration time is the length of time for which the charge on a sense node is allowed to change while the memory cell is being sensed. Calibrating the integration time is much faster than calibrating the reference voltage itself. This is due, in part, to reducing the number of different reference voltages that need to be applied during calibration. Calibrating the integration time may use different test integration times for a given read reference voltage, thereby reducing the number of read reference voltages. Hence, calibrating the integration time(s) is very efficient timewise. Also, power consumption may be reduced.

Abstract: A storage system and method for adaptive scheduling of background operations are provided. In one embodiment, after a storage system completes a host operation in the memory, the storage system remains in a high power mode for a period of time, after which the storage system enters a low-power mode. The storage system estimates whether there will be enough time to perform a background operation in the memory during the period of time without the background operation being interrupted by another host operation. In response to estimating that there will be enough time to perform the background operation in the memory without the background operation being interrupted by another host operation, the storage system performs the background operation in the memory.

Abstract: Power regulation in an integrated memory assembly having control semiconductor dies and memory semiconductor is disclosed herein. A master control die regulates power usage by the integrated memory assembly. Each control die reports information about its expected power usage to the master control die. The master control die determines a plan that meets a power criterion for the integrated memory assembly. The plan may maximize the power usage in each time period, while staying within a power budget. The plan can include selecting which of the memory dies perform a memory operation (or phase of a memory operation) during a given time period. The master control die may send a die scheduling plan to each of the other control dies. Each die scheduling plan indicates when memory operations and/or phases of memory operations are to be performed.

Abstract: A method and apparatus for obtaining data from a memory, estimating a probability of data values of the obtained data based on at least one of a source log-likelihood ratio and a channel log-likelihood ratio, wherein each bit in the obtained data has an associated log-likelihood ratio, determining at least one data pattern parameter for the data and performing a decoding process using the at least one data pattern parameters to determine a decoded data set.

Abstract: A method and apparatus for obtaining data from a memory, estimating a probability of data values of the obtained data based on at least one of a source log-likelihood ratio and a channel log-likelihood ratio, wherein each bit in the obtained data has an associated log-likelihood ratio, determining at least one data pattern parameter for the data and performing a decoding process using the at least one data pattern parameters to determine a decoded data set.

Abstract: The disclosure relates in some aspects to on-chip processing circuitry formed within the die of a non-volatile (NVM) array to perform data searches. In some aspects, the die includes components configured to sense wordlines of stored data in the NVM array by applying voltages on the wordlines serially, and then search for an input data pattern within the serially-sensed wordlines. In some examples, the components of the die include latches and circuits configured to perform bitwise latch logic search operations. In other examples, the search components are configured with under-the-array or next-to-the-array dedicated search circuitry that uses registers and/or random access memory (RAM). Other aspects relate to a separate controller device for controlling the on-chip NVM search operations. For example, the controller may determine whether to search for data using search components of the NVM die or processors of the controller based, e.g., on a degree of fragmentation of data.

Abstract: Systems and methods for compacting and anonymizing telemetry data in a storage system. A controller of a storage device may generate telemetry data based on collected features indicative of the performance of the storage device. The controller may store the telemetry data in the telemetry memory of the storage device. The controller may then transform the telemetry data into transformed telemetry data based on a dimension reduction algorithm, and transmit the transformed telemetry data to the host device.

Abstract: An integrated memory assembly comprises a memory die and a control die bonded to the memory die. The memory die includes a memory structure of non-volatile memory cells. The control die is configured to program user data to and read user data from the memory die based on one or more operational parameters. The control die is configured to calibrate the one or more operational parameters for the memory die. The control die is also configured to perform testing of the memory die using the calibrated one or more operational parameters.

Abstract: Methods and apparatus are disclosed for managing the storage of dynamic neural network data within bit-addressable memory devices, such phase change memory (PCM) arrays or other storage class memory (SCM) arrays. In some examples, a storage controller determines an expected amount of change within data to be updated. If the amount is below a threshold, an In-place Write is performed using bit-addressable writes via individual SET and RESET pulses. Otherwise, a modify version of an In-place Write is performed where a SET pulse is applied to preset a portion of memory to a SET state so that individual bit-addressable writes then may be performed using only RESET pulses to encode the updated data. In other examples, a storage controller separately manages static and dynamic neural network data by storing the static data in a NAND-based memory array and instead storing the dynamic data in a SCM array.

Abstract: Power regulation in an integrated memory assembly having control semiconductor dies and memory semiconductor is disclosed herein. A master control die regulates power usage by the integrated memory assembly. Each control die reports information about its expected power usage to the master control die. The master control die determines a plan that meets a power criterion for the integrated memory assembly. The plan may maximize the power usage in each time period, while staying within a power budget. The plan can include selecting which of the memory dies perform a memory operation (or phase of a memory operation) during a given time period. The master control die may send a die scheduling plan to each of the other control dies. Each die scheduling plan indicates when memory operations and/or phases of memory operations are to be performed.

Abstract: A memory controller includes, in one embodiment, a memory interface, a plurality of decoders, and a controller circuit. The memory interface is configured to interface with a memory having a plurality of wordlines. Each decoder of the plurality of decoders is configured to determine a bit error rate (BER). The controller circuit configured to generate a plurality of bit-error-rate estimation scan (BES) hypotheses for one wordline of the plurality of wordlines, divide the plurality of BES hypotheses among the plurality of decoders, receive BER results from the plurality of decoders based on the plurality of BES hypotheses, and adjust one or more read locations of the one wordline based on the BER results from the plurality of decoders.

Abstract: Exemplary methods and apparatus are provided for implementing a deep learning accelerator (DLA) or other neural network components within the die of a non-volatile memory (NVM) apparatus using, for example, under-the-array circuit components within the die. Some aspects disclosed herein relate to configuring the under-the-array components to implement feedforward DLA operations. Other aspects relate to backpropagation operations. Still other aspects relate to using an NAND-based on-chip copy with update function to facilitate updating synaptic weights of a neural network stored on a die. Other aspects disclosed herein relate to configuring a solid state device (SSD) controller for use with the NVM. In some aspects, the SSD controller includes flash translation layer (FTL) tables configured specifically for use with neural network data stored in the NVM.

Abstract: A programmable and reprogrammable processor on a control semiconductor die is disclosed. The processor controls various operations on a memory semiconductor die to which it is bonded, such as read, write, and erase. The processor issues control signals to operate circuits such as voltage regulators, sense amplifiers, and data latches. Because the processor is reprogrammable, it is possible to modify the operation of the circuits. For example, the processor can execute updated instructions to control the voltage regulators to modify the timing and/or magnitude of voltages applied to control lines in the memory semiconductor die. In one aspect, a page mapping scheme is updated in order to more evenly distribute a bit error rate (BER) across the pages. In one aspect, a read equalization scheme is updated. In one aspect, a technique for reading soft bits is updated.

Abstract: A storage system and method for predictive block allocation for efficient garbage collection are provided. One method involves determining whether a memory in a storage system is being used in a first usage scenario or a second usage scenario; in response to determining that the memory is being used in the first usage scenario, using a first block allocation method; and in response to determining that the memory is being used in the second usage scenario, using a second block allocation method, wherein the first block allocation method allocates blocks that are closer to needing garbage collection than the second block allocation method.

Abstract: A programmable and reprogrammable processor on a control semiconductor die is disclosed. The processor controls various operations on a memory semiconductor die to which it is bonded, such as read, write, and erase. The processor issues control signals to operate circuits such as voltage regulators, sense amplifiers, and data latches. Because the processor is reprogrammable, it is possible to modify the operation of the circuits. For example, the processor can execute updated instructions to control the voltage regulators to modify the timing and/or magnitude of voltages applied to control lines in the memory semiconductor die. In one aspect, a page mapping scheme is updated in order to more evenly distribute a bit error rate (BER) across the pages. In one aspect, a read equalization scheme is updated. In one aspect, a technique for reading soft bits is updated.

Abstract: A storage system and method for stream management in a multi-host virtualized storage system are provided. In one embodiment, a method for stream management is provided that is performed in a storage system in communication with a host comprising a plurality of virtual hosts. The method comprises: receiving, from the host, identification of each virtual host of the plurality of virtual hosts; analyzing usage history of each virtual host of the plurality of virtual hosts; and assigning streams to a subset of the plurality of virtual hosts based on the usage history, wherein a maximum number of streams assignable by the storage system is less than a total number of virtual hosts in the plurality of virtual hosts. Other embodiments are provided.

Abstract: A decoding system and method of a non-volatile memory are provided in which information regarding a characteristic of a non-volatile memory is used to determine an initial log-likelihood-ratio (LLR) table from among a number of LLR tables. The decoding is then performed using the determined initial LLR table.

Abstract: The disclosure relates in some aspects to on-chip processing circuitry formed within the die of a non-volatile (NVM) array to perform data searches. In some aspects, the die includes components configured to sense wordlines of stored data in the NVM array by applying voltages on the wordlines serially, and then search for an input data pattern within the serially-sensed wordlines. In some examples, the components of the die include latches and circuits configured to perform bitwise latch logic search operations. In other examples, the search components are configured with under-the-array or next-to-the-array dedicated search circuitry that uses registers and/or random access memory (RAM). Other aspects relate to a separate controller device for controlling the on-chip NVM search operations. For example, the controller may determine whether to search for data using search components of the NVM die or processors of the controller based, e.g., on a degree of fragmentation of data.

Abstract: The present disclosure generally relates to scheduling zone-append commands for a zoned namespace (ZNS). Rather than taking zone-append commands in order or randomly, the zone-append commands can be scheduled in the most efficient manner consistent with the open zones of the ZNS. A zone priority is determined based upon the length of time that a zone has been open together with the zone status. Generally, the older the zone and/or the more full that a zone is increases the priority. Once the zone priority is established, the zone-append commands are scheduled to ensure the zone-append commands for the high priority zones are processed first so that the open zone can be filled prior to closing.

Abstract: A storage system has a controller with an encoder. The encoder is configured to perform first and second stages of an encoding process in parallel on pipelined data blocks. In this way, while the first stage of the encoding process is being performed on a first data block, the second stage of the encoding process is performed on a second data block.