Abstract: One or more requests to store data in a flash memory portion of a storage device are received by a storage device controller of a storage device from a storage system controller. An indirection unit size to use in a flash translation layer (FTL) for mapping the data is determined from information associated with the one or more requests. The data is stored in the flash memory portion of the storage device. The data is mapped in the FTL using the indirection unit size.

Abstract: A failure of a flash memory device of multiple flash memory devices positioned on corresponding printed circuit boards is detected by a storage device controller detachably coupled to the flash memory devices and positioned on an interposer board. An indication that the flash memory device has failed is generated. A replacement of the flash memory device with a subsequent flash memory device is detected. In response to detecting the replacement, data stored on the flash memory device is rebuilt and stored on the subsequent flash memory device.

Abstract: Techniques for improving operational decisions of an autonomous vehicle are discussed herein. In some cases, a system may generate reference graphs associated with a route of the autonomous vehicle. Such reference graphs can comprise precomputed feature vectors based on grid regions and/or lane segments. The feature vectors are usable to determine scene context data associated with static objects to reduce computational expenses and compute time.

Abstract: A system, method, and product for flexible RAID layouts in a storage system, including: determining a reliability of an individual storage device of a plurality of storage devices, the individual storage device containing a plurality of portions of a Redundant Array of Independent Disks (RAID) stripe in a storage system, wherein the RAID stripe includes user data and inter-device parity data; detecting a change in the reliability of the individual storage device that contains the portion corresponding to the RAID stripe; and changing an amount of intra-device protection corresponding to the RAID stripe by decreasing, in the RAID stripe, an amount of space used to store the inter-device protection data.

Abstract: Probabilistic error correction in flash-based storage systems, including: determine that a number of data shards in a data stripe having an unrecoverable error correction code (UECC) error exceeds a number of failed shards allowable under a redundancy level of the data stripe; identifying, in a portion of data in a data shard having the UECC error, a plurality of low-confidence bits, including: performing multiple reads of the portion of data; identifying, as the plurality of low-confidence bits, a plurality of bits having values differing across the multiple reads; and correcting the portion of data by assigning corrected values to the one or more low-confidence bits.

Abstract: Data protection in a storage system that includes a plurality of Non-Volatile Memory Express (‘NVMe’) Solid State Drives (‘SSDs’), including: retrieving, from a plurality of NVMe SSDs (‘Non-Volatile Memory Express Solid State Drives’) of a storage system, one or more unencrypted shares of a master secret; reconstructing the master secret using the shares of the master secret; decrypting one or more encrypted device keys using the master secret; and using the decrypted device keys to perform a plurality of accesses to one or more of the NVMe SSDs.

Abstract: Methods, computer systems, and computer readable medium are described for low latency reads using cached deduplicated data, including: receiving a request to read data from a storage system; query, using a generated hash value associated with the request to read data, one or more deduplication tables that corresponds to the hash value; and responsive to determining that the one or more deduplication tables includes an entry that corresponds to the hash value, using a mapping contained in the entry to perform the requested to read data, wherein the mapping includes a pointer to a physical location where at least a portion of the data is stored.

Abstract: Techniques for generating a prediction about a vehicle environment and for controlling the vehicle based on the prediction are described herein. In some cases, a prediction about a vehicle environment can be determined based on at least one of contextual data associated with the vehicle environment, object state data associated with an object (e.g., another vehicle) in the vehicle environment, or image data associated with the object.

Abstract: Techniques for training a codebook usable by a machine learned model to predict an object trajectory or scene data are described herein. For example, the techniques may include generating tokens representing discrete object behavior into a machine learned model that outputs a sequence of tokens that is usable by another machine learned model to generate the object trajectory (e.g., position data, velocity data, acceleration data, etc.) or the scene data associated with the environment. The object trajectory can be sent to a vehicle computing device for consideration during vehicle planning, which may include simulation.

Abstract: Providing for increased flexibility for large scale parity, the including: writing data to a storage system, including utilizing a first data protection scheme; identifying, for storage media in the storage system, characteristics of the storage media; identifying, in dependence up the characteristics for the storage media, a second data protection scheme to use for the data; and writing the data to the to the storage system utilizing the second data protection scheme.

Abstract: A system and method comprising: receiving a request to write data stored at a first range of a first volume to a second range of a second volume, where first metadata for the first range of the first volume is associated with a range of physical addresses where the data is stored in the storage system; and responsive to receiving the request: creating second metadata for the second range of the second volume, wherein the second metadata is associated with the range of physical addresses where the data is stored in the storage system; and associating the second volume with the second metadata.

Abstract: Techniques for generating a graph using a generative adversarial network are described herein. For example, the graph can represent an input for a graph neural network whereby the graph can represent map data and object(s) in an environment. The graph can be used to train the generative adversarial network. Such techniques may be used to generate scenes that resemble potential physical environments in which the autonomous vehicle may navigate.

Abstract: Prioritizing garbage collection based on the extent to which data is deduplicated, including: determining, for one or more data elements, a number of deduplicated references to each data element; storing, for each of the data elements, the data element in an area of the storage device that contains other data elements with a similar number of deduplicated references; and adjusting a garbage collection schedule for the storage device, wherein garbage collection operations are performed more frequently on areas of the storage device that contain data elements with a relatively low number of deduplicated references.

Abstract: A method of cache management, the method comprising: identifying, among a plurality of storage items, storage items having an access count above a first threshold to generate a set of storage items; identifying, among the set of storage items, storage items having an updated access count above a second threshold to generate a subset of storage items, wherein, for each storage item, the updated access count is dependent upon a number of accesses subsequent to generating the set of storage items; and adding the storage items of the subset of storage items to a cache.

Abstract: Techniques for determining unified futures of objects in an environment are discussed herein. Techniques may include determining a first feature associated with an object in an environment and a second feature associated with the environment and based on a position of the object in the environment, updating a graph neural network (GNN) to encode the first feature and second feature into a graph node representing the object and encode relative positions of additional objects in the environment into one or more edges attached to the node. The GNN may be decoded to determine a first predicted position of the object. The first predicted position may be determined to be outside of a bounded area of the environment. Based on this determination, a second predicted position of the object may be determined using map data associated with the object.

Abstract: Methods, computer systems, and computer readable medium are described for low latency reads using cached deduplicated data, including: receiving a request to read data from a storage system; query, using a generated hash value associated with the request to read data, one or more deduplication tables that corresponds to the hash value; and responsive to determining that the one or more deduplication tables includes an entry that corresponds to the hash value, using a mapping contained in the entry to perform the requested to read data, wherein the mapping includes a pointer to a physical location where at least a portion of the data is stored.

Abstract: Aspects of the present disclosure are directed to a collaborative workspace for an artificial reality (XR) environment. Some implementations can take advantage of the capabilities of XR systems when working on collaborative large documents, such as design sheets and spreadsheets. For example, when a user launches a link to a large design sheet in XR, the workspace can show both a large view of the entire collaborative design space, as well as a personal viewport into the area of the design space that the user is working on. The workspace can include a number of collaboration controls, such as being able to follow someone else's viewport, filtering of the large view to include only the area of the sheet where a user's team is working, replaying a user's edits or view while moving around the design sheet, etc.

Abstract: Periodically re-encrypting user data stored on a storage device, including: determining that data stored in a first location of a storage device is encrypted with a data encryption key that has been decommissioned; re-encrypting the data utilizing a current data encryption key; and writing the data that is encrypted utilizing the current data encryption key to a second location of the storage device.

Abstract: A system, method, and product for flexible RAID layouts in a storage system, including: determining a reliability of an individual storage device of a plurality of storage devices, the individual storage device containing a plurality of portions of a Redundant Array of Independent Disks (RAID) stripe in a storage system, wherein the RAID stripe includes user data and inter-device parity data; detecting a change in the reliability of the individual storage device that contains the portion corresponding to the RAID stripe; and changing an amount of intra-device protection corresponding to the RAID stripe by decreasing, in the RAID stripe, an amount of space used to store the inter-device protection data.

Abstract: Preparing data for deduplication including: generating, by a storage system for a compressed data block, a padded compressed data block by padding the compressed data block to conform to a fixed block size, wherein the fixed block size is greater than a size of the compressed data block; storing, in the storage system, the padded compressed data block beginning at a block boundary of a storage device in the storage system; and performing block-based deduplication on the storage system, wherein the block-based deduplication determines whether the padded compressed data block matches one or more other padded compressed data blocks stored in the storage system.