Micron Technology Incorporated

Patrick Michael Sheridan

Aliasger Tayeb Zaidy

Lukasz Burzawa

David Andrew Roberts

Disclosed in some examples are improved address prediction and memory preloading that leverages next-delta prediction and/or far-delta prediction for scheduling using a DNN. Previous memory access sequence data that identify one or more memory addresses previously accessed by one or more processors of a system may be processed and then converted into a sequence of delta values. The sequence of delta values are then mapped to one or more classes that are then input to a DNN. The DNN then outputs a predicted future class identifier sequence that represents addresses that the DNN predicts will be accessed by the processor in the future. The predicted future class identifier sequence is then converted back to a predicted delta value sequence and back into a set of one or more predicted addresses.

Delta predictions for page scheduling

Disclosed in some examples are methods, systems, memory devices, memory controllers, and machine-readable mediums which provide for reserving physical memory device resources to specific execution units. Execution units may include processes, threads, virtual machines, functions, procedures, or the like. Physical memory device resources may include channels, modules, ranks, banks, bank groups, and the like. For example, a physical memory device resource that is reservable may be a smallest unit that allows for parallel access with another of the same size unit.

Zoned namespaces for computing device main memory

Disclosed in some examples are methods, systems, memory devices, and machine-readable mediums which store a single compressed value per line with a marker value in a front of the compressed version of the memory line. In some examples, the only value stored in the memory line is the value normally stored therein. This removes the complexity of the prediction tables and the inclusion of invalid values as well as preventing the penalty when those prediction tables are wrong. Furthermore, by inclusion of the marker in the beginning of the memory line, the system can quickly determine the compression status of the memory line without having to read the entire line. That is, it can quickly stop reading the rest of the memory line once the compressed data is read out which saves the memory device from having to read the entire line.

Early detection of compression status using inline metadata

A method for managing logical-to-physical (L2P) mappings in a memory subsystem is described. The method includes updating, by a set of processing units, an L2P table based on a set of journal pages from the non-volatile memory. The L2P table prior to the update includes a first set of entries from a table snapshot and the table includes a second set of entries following the update from a set of journal pages. Each L2P table entry (1) corresponds to a logical address in a set of logical addresses and (2) includes a physical address of a set of memory components. The set of logical addresses are categorized into zones and each processing unit is assigned to a separate zone such that each processing unit updates the first set of entries based on an assigned zone of a corresponding logical address and each zone includes at least two non-contiguous logical addresses.

Growing community of inventors

Boulder, CO, United States of America

Patrick Michael Sheridan