Toshiba Memory Corporation

Kioxia Corporation

Kabushiki Kaisha Toshiba

David Malcolm Symons

Paul Edward Hanham

Francesco Giorgio

Neil Buxton

Ezequiel Alves

Jonghyeon Kim

Senthilkumar Diraviam

Josh Bowman

A method performed by a controller comprising assigning a first status indicator to entries in a first address line in a volatile memory belonging to a first region of an LUT stored in a non-volatile memory, and a second status indicator to entries in the first address line in the volatile memory belonging to a second region of the LUT, setting either the first or second status indicator to a dirty status based on whether a cache updated entry at an address m in the volatile memory belongs to the first or second region of the LUT, and writing, based on the dirty status of the first and second status indicator at the address m, all entries in the volatile memory associated with the first region or the second region containing the updated entry to the non-volatile memory.

Enhancing cache dirty information

Various implementations described herein relate to correcting errors in Dynamic Random Access Memory (DRAM). A memory controller uses an Error Correcting Code (ECC) to store an encoded data word within a DRAM die. The DRAM die is communicatively coupled the memory controller by a memory data bus. The DRAM die includes on-die error correction for data bits stored in the DRAM. Upon reading the encoded data word, the memory controller corrects and detects one or more errors. The one or more errors are introduced by at least one of the on-die error correction of the DRAM die or the memory data bus.

Systems and methods for detecting or preventing false detection of three error bits by SEC

Various implementations are directed to systems and methods for maintaining integrity and reliability of data in an SSD device using error correction coding. According to certain aspects, for frames of data having an ECC code with two or more sub-codes, while one sub-decoder is not in use it could be used to start a decode of another frame. By 'interleaving' and alternating the frames between sub-decoders, two or more frames can be decoded simultaneously in an efficient manner. This can clearly be extended to more sub-codes (i.e. dimensions greater than two).

Efficient decoding of n-dimensional error correction codes

There is provided a method of providing adjusted LLR values of a plurality of bits in a codeword to an LDPC decoder, the plurality of bits representing a plurality of charge states of a plurality of memory cells of a non-volatile memory. The method comprises storing in a non-volatile memory controller associated with the non-volatile memory LLR values of the plurality of bits. The controller then determines a plurality of levels of the charge states represented by the plurality of bits. The controller then generates, by a distribution processor, distributions of a population of the plurality of bits in the codeword at each of the plurality of levels at a first and a second time after the first time. The controller then generates the adjusted LLR values based on a comparison between the first and second distributions, and then decodes the codeword according to the adjusted LLR values.

Non volatile memory controller device and method for adjustment

Various implementations described herein relate to systems and methods for correcting data from memory systems such as a plurality of non-volatile memory devices of a Solid State Drive (SSD), including but not limited to, receiving frames of the data from the plurality of non-volatile memory devices, allocating the frames among pooled frontline Error Correction Code (ECC) decoders, decoding, by the pooled frontline ECC decoders, the frames to output first decoded frames, and returning the first decoded frames to the read channels.

Pooled frontline ECC decoders in memory systems

A method for fast calculation of a frame error rate (FER) of an error correcting code (ECC) soft decoder using a soft read process includes determining an MI-FER conversion data structure based on a relationship between mutual information (MI) of input channels and output channels of a memory, and FER of the ECC soft decoder, and decoding an encoded data codeword stored in a memory page of the memory and read using a soft read process. The method further includes generating a set of joint probability values using the information from the soft read process and data indicating true bit values for the data codeword, determining an MI value using the set of joint probability values, and determining an FER estimate using the MI-FER conversion data structure.

System and method for improved memory error rate estimation

A system for an Error Correction Code ('ECC') decoder includes a first decoder and a second decoder. The first decoder is configured to determine a first estimated number of errors in encoded data received at the first decoder and to compare the first estimated number of errors to a first threshold and a second threshold. The second decoder is configured to receive the encoded data when the first estimated number of errors is below the first threshold and is above a second threshold. When the first estimated number of errors is above the first threshold, the first decoder passes the encoded data out of the ECC. The first decoder has a lower power consumption than the second decoder.

Low power error correcting code (ECC) system

A solid state storage device comprises a non-volatile memory controller configured to store data in a non-volatile memory, wherein the stored data is encoded using a first error-correcting code and a second Low Density Parity Check (LDPC) code. The non-volatile memory controller includes a hard-decision LDPC decoder to decode encoded data received from the non-volatile memory and provide a decoded data output. The hard-decision LDPC decoder selects a voting scheme at each iteration in a sequence of iterations of decoding to determine when to implement bit flipping at a variable node amongst a plurality of check nodes, each of the plurality of check nodes connected to a plurality of variable nodes.

Optimal LDPC bit flip decision

A non-volatile memory controller includes a hard-decision Low Density Parity Check (LDPC) decoder with a capability to dynamically select a voting method to improve the decoding in low bit error rate (BER) situations. The hard-decision LDPC decoder dynamically selects a voting method associated with a strength requirement for bit flipping decisions. In one implementation, the voting method is selected based on the degree of a variable node and previous syndrome values.

Low BER hard-decision LDPC decoder

A method of providing, by a controller, a log likelihood ratio (LLR) to a low-density parity check (LDPC) decoder, the method comprising storing, in a non-volatile memory controller, a look-up table for storing LLR values of at least one bit representing a charge state of a cell of the plurality of cells in the memory. The controller determines a cell charge state of the target cell, calculates a value representative of the difference in charge states of the target cell and at least one of a plurality of neighboring cells. The controller compares the calculated value with at least one predetermined threshold value, and sets at least one address bit of an address to the look-up table if the calculated value exceeds the at least one threshold value. The controller extracts a new LLR value from the look-up table, and provides the new LLR value to the LDPC decoder.

Soft decision LDPC decoder with improved LLR from neighboring bits

A non-volatile memory controller for a solid state drive includes a soft-decision LDPC decoder. The soft-decision LDPC decoder includes a probability generation module. A processor reads collected statistics collated from decoded frames and tunes the performance of the soft-decision LDPC decoder performance. Additional parameters may also be taken into account, such as the scramble seed and the type of non-volatile memory. An asymmetry in errors may also be detected and provided to a hard-decision LDPC decoder to adjust its performance.

Growing community of inventors

Kidlington, United Kingdom

David Malcolm Symons