Seagate Technology Incorporated

Jeffrey John Pream

Ara Patapoutian

Jeremy Blair Goolsby

Ryan James Goss

Jonathan M Henze

Indrajit Zagade

Charles McJilton

Thomas V Spencer

Deepak Sridhara

Young Pil Kim

Ivana Djurdjevic

David O Sluiter

Bengt Anders Ulriksson

Nicholas Odin Lien

Christopher A Smith

Eric Michael Beck

Bijoy Purushothaman

Venugopal Rao Garuda

A data storage system may connect a non-volatile memory to a quarantine module that generates a quarantine strategy in response to a pending data access request to the non-volatile memory. The quarantine strategy can proactively prescribing a plurality of status levels for physical data addresses of the non-volatile memory. A comparison of a volume of errors for the non-volatile memory to a first threshold of the quarantine strategy with the quarantine module may prompt the alteration of a first status level of the plurality of status levels for a first physical data address of the non-volatile memory, as directed by the quarantine strategy.

Data storage system with intelligent error management

Method and apparatus for managing data in a storage device, such as a solid-state drive (SSD). A non-volatile memory (NVM) is arranged into multiple garbage collection units (GCUs) each separately erasable and allocatable as a unit. Read circuitry applies read voltages to memory cells in the GCUs to sense a programmed state of the memory cells. Calibration circuitry groups different memory cells from different GCUs into calibration groups that share a selected set of read voltages. A read command queue accumulates pending read commands to transfer data from the NVM to a local read buffer. Read command coalescing circuitry coalesces selected read commands from the queue into a combined command for execution as a single batch command. The combined batch command may include read voltages for use in retrieval of the requested data.

Coalescing read commands by location from a host queue

Method and apparatus for managing data in a storage device, such as a solid-state drive (SSD). In some embodiments, a main memory has memory cells arranged on dies arranged as die sets accessible using parallel channels. A controller is configured to arbitrate resources required by access commands to transfer data to or from the main memory using the parallel channels, to monitor an occurrence rate of collisions between commands requiring an overlapping set of the resources, and to adjust a ratio among different types of commands executed by the controller responsive to the occurrence rate of the collisions. In further embodiments, the controller may divide a full command into multiple partial commands, each of which are executed as the associated system resources become available. In some cases, the ratio is established between read commands and write commands issued to the main memory.

Collision reduction through just-in-time resource allocation

Method and apparatus for managing data in a storage device, such as a solid-state drive (SSD). A data storage device includes a main non-volatile memory (NVM) and a command queue that lists pending data transfer commands to transfer data between the NVM and a local memory. A collision manager predicts future collisions among the pending data transfer commands, such as but not limited to commands involving pending host commands from a host. A storage manager enacts a change in a storage policy to reduce a future rate of the predicted future collisions. The change in storage policy may involve duplication of write data so that the write data are written to multiple locations within the NVM. The change in storage policy may further involve a pre-emptive garbage collection operation upon an existing location to distribute current version data blocks to multiple locations within the NVM.

Pre-emptive storage strategies to reduce host command collisions

Systems and methods are disclosed for command status polling at a flash queue of a non-volatile memory device. The flash queue may be configured to perform polling on the status of flash operations without direct oversight from the data storage controller or firmware. In certain embodiments, a flash queue circuit may be configured to receive, from a data storage controller of a nonvolatile solid state memory (NVSSM) data storage device, one or more commands to access a flash memory of the NVSSM data storage device, each command of the one or more commands including one or more instructions. The flash queue circuit may execute the one or more commands to access the flash memory, evaluate a status response from the flash memory at the flash queue circuit, and re-execute a sequence of instructions of the one or more commands based on the status response.

Flash queue status polling

A low-density parity check (LDPC) decoder includes a variable node unit (VNU) comprising a plurality of variable nodes configured to perform sums. A first message mapper of the LDPC decoder receives first n1-bit indices from likelihood ratio (LLR) input and maps the first n1-bit indices to first numerical values that are input to the variable nodes of the VNU. A second message mapper of the LDPC decoder receives second n2-bit indices from a check node unit (CNU) and maps the second n2-bit indices to second numerical values that are input to the variable nodes of the VNU. The CNU includes a plurality of check nodes that perform parity check operations. The first and second numerical values having ranges that are larger than what can be represented in n1-bit and n2-bit binary, respectively.

Method and apparatus for LDPC decoding using indexed messages

Method and apparatus for managing data decoder circuits, such as LDPC (low density parity check) decoders in a solid state drive (SSD). In some embodiments, a non-volatile memory (NVM) is configured to store data in the form of code words. Each code word has a user data payload and associated code bits. A plurality of data decoder circuits are configured to use the code bits to detect and correct bit errors in the code words during a read operation. A power transition circuit is configured to successively transition each of the data decoder circuits in turn from a first power mode to a second power mode, such as from an active mode to an idle mode, at a different time and at a conclusion of a predetermined time interval. In this way, voltage spikes or other anomalous conditions on a voltage source pathway may be reduced.

Sequential power transitioning of multiple data decoders

The disclosed technology provides for multi-dimensional data randomization in a memory cell array using circular shifts of an initial scrambling sequence. Data addressed to a first row of a data array is randomized using the initial scrambling sequence and data addressed to each row of the memory cell array is randomized using a scrambling sequence that is equal to a circular shift of the initial sequence.

Multi-dimentional data randomization

A load adaptive pipeline system includes a data recovery pipeline configured to transfer data between a memory and a host. The pipeline includes a plurality of resources, one or more of the plurality of resources in the pipeline have multiple resource components available for allocation. The system includes a pipeline controller configured to assess at least one parameter affecting data transfer through the pipeline. The pipeline controller is configure to allocate resource components to the one or more resources in the pipeline in response to assessment of the at least one data transfer parameter.

Load adaptive data recovery pipeline

A device comprising a data transfer channel is configured to transfer data between multiple memory devices and a host device. The channel includes multiple decoders and a buffer coupled between the multiple memory devices and the multiple decoders. The buffer is configured to store code words received from the memory devices. Channel control logic is configured to determine availability of one or more of the multiple decoders and to distribute the code words to the one or more decoders based on decoder availability.

Growing community of inventors

Berthoud, CO, United States of America

Jeffrey John Pream