Abstract: Systems and methods for predicting read commands and pre-fetching data when a storage device is receiving random read commands to non-sequentially addressed data locations from a plurality of host sources are disclosed. A storage device having a memory with a plurality of separate prior read command data structures includes a controller having a next read command prediction module that separately predicts a next read command based on a received read command from the one of the plurality of prior read command data structures associated with the host from which the received command originated. The storage device then pre-fetches the data identified in the predicted next read command.

Abstract: A memory with asymmetric command latency characteristics for WRITE operations utilizing WOM coding methodologies to reduce programming latency across a number of WRITE operations.

Abstract: A memory with asymmetric command latency characteristics for WRITE operations utilizing WOM coding methodologies to reduce programming latency across a number of WRITE operations.

Abstract: Various method and apparatus embodiments for data dependent error correction code (ECC) encoding are disclosed. In one embodiment, a data object may include multiple portions, with each portion having different characteristics. An ECC encoder may allocate error correction resources (e.g., parity bits) to the different portions at respectively different data rates (e.g., more error correction resources to some portions relative to other portions). Upon completion of the allocation, the data object and the associated error correction resources are forwarded to a storage medium for storage therein.

Abstract: A controller for a phase change memory forms a dedicated burst write partition in the phase change memory and initializes memory cells of the dedicated burst write partition to a SET state. Programming of selected memory cells in the dedicated burst write partition is carried out using only RESET pulses.

Abstract: Disclosed herein is a memory device and a method of descrambling and decoding encoded data. In one aspect, encoded data is received. A scrambling seed is obtained from the encoded data prior to decoding the encoded data. The encoded data is descrambled according to the scrambling seed, and the descrambled data is decoded. The descrambled data may be decoded according to statistics of the descrambled data.

Abstract: Systems and methods for predicting the next host source of a command directed to a storage device, and configuring the storage device with the host memory configuration parameters of that predicted host, are disclosed. A predictive model may be used to identify which of a plurality of different actual or virtual hosts will be sending a next command based on a pattern of prior host origins of commands. The storage device may include a plurality of different memory configuration parameters associated with each different host source and may update the current memory configuration based on the predicted next host origin prior to actual receipt of a next command.

Abstract: Systems and methods for predicting read commands and pre-fetching data when a storage device is receiving random read commands to non-sequentially addressed data locations from a plurality of host sources are disclosed. A storage device having a memory with a plurality of separate prior read command data structures includes a controller having a next read command prediction module that separately predicts a next read command based on a received read command from the one of the plurality of prior read command data structures associated with the host from which the received command originated. The storage device then pre-fetches the data identified in the predicted next read command.

Abstract: Systems and methods for permitting flexible use of volatile memory for storing read command prediction data in a memory device, or in a host memory buffer accessible by the memory device, while preserving accuracy in predicting read commands and pre-fetching data are disclosed. The read command prediction data may be in the form of history pattern match table having entries indexed to a search sequence of one or more commands historically preceding the read command in the indexed table entry. A host trigger requesting the limited volatile memory space, a lower power state that is detected, or a memory device-initiated need may trigger generation of and subsequent use of a smaller table for the prediction process while the larger table is released. The memory device may later regenerate the larger table when more space in the volatile memory becomes available.

Abstract: A system and method for reducing performance penalties of a host that is supplying a host memory buffer (HMB) for use by a storage device. The method may include modeling desired HMB access timing by the storage device in an initial offline analysis for multiple classes of workloads, periodically updating the access timing data stored in the storage device based on actual use and using the current HMB access timing information to modify storage device access to the HMB on the host. The system may include a storage device controller that quantifies different HMB access timing for different host workloads based on individual HMB regions.

Abstract: A non-volatile storage apparatus comprises a non-volatile storage and a control circuit connected to the non-volatile storage. The non-volatile storage structure is organized into multiple partitions. Each partition is preassigned to a different data shaping level. Data to be stored in the non-volatile storage is shaped based on its entropy. The control circuit is configured to write shaped data to a partition of the multiple partitions that is preassigned to a same shaping level as the shaped data.

Abstract: A storage device may utilize de-centralized latch management to remove functions from the device controller to the memory die. NAND die located on a common bus may share a pool of latches with one die acting as a proxy or manager for the other die. A bridge or bridges may be used between NAND connections to allow additional die to be controlled by a leader die for the partially de-centralized management of latches. The latch management operations may include a sequence of commands/operations performed by the leader die.

Abstract: A method for calibrating read threshold voltages includes receiving, from at least one memory die, a number of page bits corresponding to a number of read operations performed on a page associated with the at least one memory die. The method further includes determining voltage bins for each bit of the number of page bits. The method further includes determining, for each voltage bin, a bit error rate. The method further includes adjusting read threshold voltages associated with the at least one memory die using the bite error rate for each voltage bin.

Abstract: A method and apparatus for cache management and eviction polices using unsupervised reinforcement learning schemes is disclosed.

Abstract: Systems and methods for managing performance and quality of service (QoS) with multiple namespace resource allocation. NVM Express (NVMe) implements a paired submission queue and completion queue mechanism, with host software on the host device placing commands into the submission queue. The memory device processes the commands through various phases including fetching, processing, posting a completion message, and sending an interrupt to the host. NVMe may support the use of namespaces. Namespace configuration may be modified to include performance criteria specific to each namespace. The memory device may then receive commands directed to specific namespaces an apply memory device resources to commands in each namespace queue such that QoS may be applied to control execution of commands such that commands in each namespace receive resources based on host selected performance parameters for each namespace.

Abstract: Various method and apparatus embodiments for data dependent error correction code (ECC) encoding are disclosed. In one embodiment, a data object may include multiple portions, with each portion having different characteristics. An ECC encoder may allocate error correction resources (e.g., parity bits) to the different portions at respectively different data rates (e.g., more error correction resources to some portions relative to other portions). Upon completion of the allocation, the data object and the associated error correction resources are forwarded to a storage medium for storage therein.

Abstract: Disclosed herein is a memory device and a method of descrambling and decoding encoded data. In one aspect, encoded data is received. A scrambling seed is obtained from the encoded data prior to decoding the encoded data. The encoded data is descrambled according to the scrambling seed, and the descrambled data is decoded. The descrambled data may be decoded according to statistics of the descrambled data.

Abstract: Embodiments of the present disclosure generally relate to storage devices, such as SSDs. A data storage device comprises an encrypted interface, one or more flash memory devices, and a controller configured to receive one or more workloads of data through the encrypted interface. Upon a threshold being met, the controller performs a diagnosis of one or more operating parameters of the one or more workloads of data. Based on the diagnosis, the data storage device is optimized by recalibrating one or more of: a partitioning of bits per cell of the one or more flash memory devices, one or more flash management parameters of the data storage device, and a programming rate of the storage device.

Abstract: Apparatuses, systems, and methods are disclosed for dynamic read operations. An on-die controller monitors one or more read statistics during a read operation for data of a non-volatile memory die. An on-die controller determines whether one or more read statistics satisfy a threshold for a read operation. An on-die controller dynamically modifies a read operation based on determining that one or more read statistics fail to satisfy a threshold.

Abstract: Technology is disclosed for reading non-volatile memory when a host does not need perfect data. By allowing the memory to return data with some errors, the data will be provided to the host much quicker. Therefore, in response to one or more host read commands, the memory system returns multiple copies of the data over time, progressively getting better so that later in time copies of the data have lower number of errors. The host decides when the error rate is good enough and stops the process (or ignores the rest of the results).