Western Digital Technologies, Inc.

Sandisk Technologies Inc.

Inter-tel, Inc.

Mervyn Wongso

Jacob Schmier

Lakshmana Chintada

Rushang Karia

Kevin Corbin

James Fitzpatrick

James M Higgins

Li Li

Edoardo Daelli

Brian Willess

A method and apparatus is disclosed for using supervised learning with closed loop feedback to improvement of output consistency for memory arrangements, such as a solid state drive.

Supervised learning with closed loop feedback to improve input output consistency of solid state drives

A non-volatile memory (NVM) system receives host requests that each specify a memory operation to be performed by the NVM system, the specified memory operations including read operations and write operations, and performs a set of operations for each memory operation specified by a received host request. The set of operations performed for each such memory operation include: initiating performance the memory operation; determining a throttle interval for the memory operation in accordance with at least a first factor, corresponding to available space in a write cache of the non-volatile memory system, and a second factor, corresponding to a metric corresponding to prevalence of write operations in the memory operations specified by the received host requests; and returning to the host system a response associated with the memory operation at a time no earlier than a start time associated with the memory operation plus the determined throttle interval.

Method and system for input-output throttling to improve quality of service in a solid state drive

Supervised learning with closed loop feedback to improve ioconsistency of solid state drives

The various implementations described herein include systems, methods and/or devices used to enable adaptive erasure in a storage device. The method includes performing a plurality of memory operations including read operations and respective erase operations on portions of one or more non-volatile memory devices specified by the read operations and respective erase operations, where the respective erase operations are performed using a first set of erase parameters that has been established as a current set of erase parameters prior to performing the respective erase operations. The method includes, in accordance with each erase operation of at least a subset of the respective erase operations, updating one or more erase statistics that correspond to performance of multiple erase operations. The method includes, in accordance with a comparison of the erase statistics with an erasure performance threshold, establishing a second set of erase parameters as the current set of erase parameters.

Adaptive erase of a storage device

A firmware distribution system is disclosed with particular application to integrating a plurality of ancillary endpoints into an existing network of endpoints. The ancillary endpoints being coupled to an isolated network and a master device downloading the desired firmware to the ancillary endpoints in a peer-to-peer distribution. An administrator may be coupled to the isolated network for providing a management and control mechanism for an installer. Once the ancillary network has the firmware loaded, the network may be tested prior to integration with the existing network of endpoints. Upon verification of the download and operation of the ancillary endpoints, the ancillary endpoints can be readily integrated with the existing network without major disruption to the existing network.

Peer-to-peer distribution of firmware

Growing community of inventors

Gilbert, AZ, United States of America

Mervyn Wongso