Abstract:
A storage device is provided as follows. A nonvolatile memory device includes blocks, each block having sub-blocks erased independently. A memory controller performs a garbage collection operation on the nonvolatile memory device by selecting a garbage collection victim sub-block among the sub-blocks and erasing the selected garbage collection victim sub-block to generate a free sub-block. The memory controller selects the garbage collection victim sub-block using valid page information of each sub-block and valid page information of memory cells adjacent to each sub-block.
Abstract:
A method of decoding may include performing fewer than ⌊ 2 k - 1 k ⌋ number of sensing operations of multilevel cells within a nonvolatile memory device, decoding pages corresponding to each of the sensing operations while correcting a channel error using an RIO code, and extracting user data from the decoded pages.
Abstract:
A memory controller includes a state shaping encoder that receives k-bit write data, selects a logical page with reference to state shape mapping information, and changes data of the logical page to decrease an occurrence probability of a high-order program state among program states used to program the k-bit data in multi-level memory cells.
Abstract:
A method for controlling a nonvolatile memory device includes requesting a plurality of first sampling values from the nonvolatile memory device, each of the first sampling values representing the number of memory cells having a threshold voltage between a first sampling read voltage and a second sampling read voltage. The first sampling values are processed through a non-linear filtering operation to estimate the number of memory cells having the threshold voltage between the first sampling read voltage and the second sampling read voltage.
Abstract:
A redundant array of inexpensive disks (RAID) controller of a RAID storage system that includes one or more storage devices includes an error correction code (ECC) result manager configured to manage information of ECC result indicators when a data chunk that includes one or more ECC data units having an uncorrectable ECC error is read from among a plurality of data chunks dispersively stored in the one or more storage devices, each of the plurality of data chunks including a plurality of ECC data units, the ECC result indicators respectively indicating whether the plurality of ECC data units included in the plurality of data chunks has an uncorrectable ECC error; and an uncorrectable error counter configured to calculate a number of ECC result indicators indicating an uncorrectable ECC error among ECC result indicators corresponding to ECC data units having a same order in each of the plurality of data chunks.
Abstract:
The inventive concepts relate to an operation method of an error correction decoder correcting an error of data read from a nonvolatile memory. The operation method may include receiving the data from the nonvolatile memory, performing a first error correction with respect to the received data in a simplified mode, and performing, when the first error correction fails in the simplified mode, a second error correction with respect to the received data in a full mode. When the first error correction of the simplified mode is performed, a part of operations of the second error correction of the full mode may be omitted.
Abstract:
Disclosed is a bit-state mapping method of a flash memory system which maps m-bit data (m being a natural number more than 2) onto one of 2m states (voltage threshold distributions). The bit-state mapping method includes performing a subset partitioning operation during first to (m−1)th levels under a condition that two adjacent states are processed as one state; and distinguishing between the adjacent states while processing an (m)th level.
Abstract:
A data processing method is provided for processing data read from a nonvolatile memory. The data processing method includes receiving first bit data from the nonvolatile memory at a memory controller, and performing erasure decoding based on the first bit data and second bit data stored in the memory controller. The first bit data indicates a memory cell that is erasure, and the second bit data is read using a read voltage during previous error correction decoding.
Abstract:
A solid-state drive (SSD) may include a volatile buffer such as DRAM, a non-volatile memory (NVM) such as NAND Flash connected to the volatile buffer, and a capacitor connected to both, where the capacitor may have an energy capacity insufficient to supply the buffer and NVM using a normal supply voltage in a normal mode, but sufficient to supply the buffer and NVM using at least one reduced supply voltage in a temporary mode; and a related method may include programming data to the NVM by temporarily reducing the supply voltage to the NVM, and writing data to the NVM using the reduced supply voltage.
Abstract:
A low-density parity check (LDPC) decoder may include a variable node processing unit and a check node processing unit. The check node processing unit includes memory elements storing a check node value. The memory elements are interconnected through two or more paths, and each of the paths may include a total or partial cyclic permutation of the memory elements to transmit the check node value.