Abstract: Methods and apparatus are provided for collecting cell-level statistics for detection and decoding in flash memories. Data from a flash memory device is processed by obtaining one or more read values for a plurality of bits in a page of the flash memory device; and generating cell-level statistics for the flash memory device based on a probability that a data pattern was read from the plurality of bits given that a particular pattern was written to the plurality of bits. The cell-level statistics are optionally generated substantially simultaneously with a reading of the read values, for example, as part of a read scrub process. The cell-level statistics can be used to convert the read values for the plurality of bits to a reliability value for a bit among the plurality of bits.

Abstract: An apparatus having a device and a circuit is disclosed. The device is configured to convey a codeword. The circuit is configured to (i) receive the codeword from the device, (ii) generate a syndrome by performing a portion less than all of an iterative decoding procedure on the codeword, (iii) generate a value by counting a number of unsatisfied parity checks in the syndrome and (iv) generate a quality metric of the device according to the value.

Abstract: A method of characterizing a distribution of a maximum number of errors that first cause uncorrectable error correction code failure for hard low density parity check codes includes selecting a low density parity check code, generating encoded data with the low density parity check code and writing the encoded data to a number of memory blocks, reading the encoded data from the number of memory blocks and determining any pages having a first uncorrectable error correction code failure, determining a number of raw bit errors for each page having a first uncorrectable error correction code failure, incrementing an error count value corresponding to each of the numbers of raw bit errors determined, and repeating the generating, reading, determining, and incrementing steps for a predetermined range of values of a predetermined reliability statistic of the memory blocks.

Abstract: Systems and method relating generally to data processing, and more particularly to systems and methods for scaling messages in a data decoding circuit. In one embodiment, the systems and methods include applying a variable node algorithm, applying a check node algorithm, calculating a first number of errors, calculating a second number of errors, calculating a difference between the first and second number of errors, multiplying by scalar values to yield a scaled set of messages, and re-applying the variable node algorithm to the scaled set of messages.

Abstract: An apparatus having a circuit and a decoder is disclosed. The circuit is configured to adjust an initial one of a plurality of reference voltages in a read channel of a memory by shifting the initial reference voltage an amount toward a center of a window and read a codeword from the memory a number of times. The window bounds a sweep of the reference voltages. Each retry of the reads uses a respective reference voltage from a pattern of the reference voltages. The pattern is symmetrically spaced about the initial reference voltage. The pattern fits in the window. The decoder is configured to generate read data by performing an iterative decoding procedure on the codeword based on the reads.

Abstract: A method for system for dynamic channel Log Likelihood Ratio (LLR) quantization for a Solid State Drive (SSD) controller is a targeted approach to scaling which results in a scaled, quantized set of LLRs whose relative magnitude remains undisturbed from an original magnitude. The method reads a set of voltages from each channel of the SSD. The set of reads is configured in location and number for performance. Once a set is returned, the method determines an LLR for each of the voltages read resulting in a raw set of LLRs. Targeted scaling results in a scaled set of LLRs between an upper limit and a lower limit determined for reading by a decoder. Once scaled, the LLRs are rounded and quantized for use by the decoder to produce an Error Correction Code (ECC).

Abstract: An apparatus comprising a memory and a controller. The memory is configured to process a plurality of read/write operations. The memory comprises a plurality of memory unit granularities each having a size less than a total size of the memory. The controller is configured to process a plurality of I/O requests to the memory units of the memory that are not marked as bad on a memory unit list. The controller is configured to track a plurality of bad blocks of the memory. The controller is configured to determine which of the memory units to mark as bad based on a test of whether a unit of memory larger than a block of the memory has failed. The test is based on a threshold of the bad blocks in the unit of memory.

Abstract: An apparatus for reading a flash memory includes a read controller operable to read the flash memory to yield read patterns, a likelihood generator operable to map the read patterns to likelihood values, a decoder operable to decode the likelihood values, a data state storage operable to retrieve the likelihood values for which decoding failed, and a selective dampening controller operable to select at least one dampening candidate from among the likelihood values for which decoding failed, to dampen the likelihood values of the at least one dampening candidate to yield dampened likelihood values, and to provide the dampened likelihood values to the decoder for decoding.

Abstract: Systems and method relating generally to solid state memory, and more particularly to systems and methods for recycling data in a solid state memory. The systems and methods include receiving a data set maintained in a memory device, applying at least one iteration of a data decoding algorithm to the data set by a data decoder circuit to yield a decoded output, counting the number of iterations of the data decoding algorithm applied to the data set to yield an iteration count, and recycling the data set to the memory device. The recycling is triggered based at least in part on the iteration count.

Abstract: An apparatus includes an error correction code circuit and an error correction code selection circuit. The error correction code circuit may be configured to encode and decode data using any of a plurality of error correction codes. The error correction code selection circuit may be configured to control which of the plurality of error correction codes is used by the error correction code circuit to encode and decode data responsive to one or more reliability statistics and predetermined data characterizing distribution properties of each of the plurality of error correction codes.

Abstract: An adaptive channel tracking algorithm performed by a flash memory system obtains disparity metrics and derivative metrics and uses a combination of the disparity and derivative metrics to estimate an optimal read reference voltage. The estimation of the optimal read reference voltage does not rely on assumptions about the underlying cell voltage distributions and results in a good estimate of the read reference voltage even if the standard deviations of the cell voltage distributions are different. In addition, the algorithm is relatively simple and less computationally intensive to perform than the known tracking algorithms.

Abstract: An apparatus having a memory and a controller is disclosed. The memory is configured to (i) program a protected lower unit in a lower page of a location, (ii) generate a corrected lower unit by correcting the protected lower unit using a first error correction code and (iii) program a protected upper unit in an upper page of the location based on the corrected lower unit. The controller is configured to generate the protected upper unit by encoding an upper write data item using a second error correction code. The controller is on a separate die as the memory.

Abstract: An apparatus having a circuit and an interface to a nonvolatile memory is disclosed. The circuit is configured to (i) perform one or more attempts of a soft-decision decode of data stored in the nonvolatile memory, where soft-decision decode uses a plurality of log likelihood ratio values stored in a table, (ii) generate one or more adjusted log likelihood ratio values by adding a constant value to one or more of the log likelihood ratio values in response to a failure to decode the data using the log likelihood ratio values and (iii) re-decode the data using the adjusted log likelihood ratio values.

Abstract: In a data storage system having multi-level memory cells, a counter tracks the number of program/erase cycles, anticipated read cycles, or anticipated length of data retention of a cell. When a threshold number of cycles is reached or length of retention or read frequency are anticipated, the cell is programmed using more program voltage pulses, narrower program voltage pulses or some other modification to the incremental step programming pulse to reduce the range where the intermediate least significant (lower) bit read voltage may be erroneous, thereby reducing the probability of write errors when the most significant page (upper) is programmed.

Abstract: Systems and method relating generally to solid state memory, and more particularly to systems and methods for recovering data from a solid state memory. In one embodiment, the systems and methods include providing a flash memory circuit including a superset of memory cells, accessing a data set from a group of memory cells using a standard reference value to distinguish bit values in the group of memory cells, and based at least in part on determining that the group of memory cells was a last written group of memory cells, re-accessing a data set from the group of memory cells using a last written reference value to distinguish bit values in the group of memory cells.

Abstract: Methods and apparatus are provided for soft data generation for memory devices using reference cells. At least one soft data value is generated in a memory device by writing a known data to one or more reference cells; reading one or more of the reference cells; obtaining a read statistic based on the read one or more reference cells; and obtaining the at least one soft data value based on the obtained read statistic. The read statistics can optionally be obtained for one or more desired locations of a memory array; or for a given pattern, PATT, in one or more aggressor cells. The read statistic can optionally comprise asymmetric statistics obtained for a plurality of possible values.

Abstract: Systems and method relating generally to data processing, and more particularly to systems and methods for scaling messages in a data decoding circuit. In one embodiment, the systems and methods include applying a variable node algorithm, applying a check node algorithm, calculating a first number of errors, calculating a second number of errors, calculating a difference between the first and second number of errors, multiplying by scalar values to yield a scaled set of messages, and re-applying the variable node algorithm to the scaled set of messages.

Abstract: Systems and methods relating generally to data processing, and more particularly to systems and methods for characterizing a solid state memory. In one embodiment, the systems and methods may include programming a first cell of a solid state memory device to a negative voltage, programming a second cell of the solid state memory device to a positive voltage, wherein the second cell is adjacent to the first cell, calculating a voltage shift on the negative voltage programmed to the first cell, characterizing a shifted voltage level on the first cell as an interim voltage, and subtracting the voltage shift from the interim voltage to yield an actual voltage on the first cell.

Abstract: An apparatus comprising a memory and a controller. The memory may be configured to process a plurality of read/write operations. The memory may comprise a plurality of memory modules each having a size less than a total size of the memory. The controller may be configured to recover data stored in the memory determined to exceed a maximum number of errors after performing a first read operation using a first read reference voltage. The controller may perform a second read operation using a second read reference voltage. The controller may identify a victim cell having a threshold voltage in a region between the first read reference voltage and the second read reference voltage. The controller may perform a third read operation on aggressor cells of the victim cell. The controller may perform a fourth read operation using the first read reference voltage with bit-fixed values on the victim cell based on a type of interference from the aggressor cells.

Abstract: Systems and method relating generally to solid state memory, and more particularly to systems and methods for generated data from a solid state memory. A data processing system includes a solid state memory device, a soft data generation circuit operable to receive multiple instances of an element of a read data set accessed from the solid state memory device, and access a scramble compensating extended look up table using the multiple instances of the element to receive corresponding soft data, and a data decoder circuit operable to yield a decoded output from the soft data.