Abstract: Systems, apparatus and methods are provided for determining whether data accessed by the command in a storage system is hot or cold. An apparatus may include a first interface to be coupled to a host and a storage controller configured to: receive a command that contains an address in a data storage system; generate a set of hash addresses for the address; for each hash address of the set of hash addresses: obtain a stored hotness score associated with the hash address, update the stored hotness score to generate an updated hotness score associated with the hash address, and determine that the updated hotness score is above a hotness threshold; and determine that the address is hot.

Abstract: Systems, apparatus and methods are provided for performing computations of a neural network using hardware computational circuitry. An apparatus may include a controller, a configuration buffer and a data buffer. The controller may be configured to dispatch computing tasks of a neural network, load configurations into the configuration buffer and load input data and parameters including weights and biases into the data buffer. The apparatus may also include a multiply-accumulate (MAC) layer. The configurations may include at least one FNN configuration. The MAC layer may apply the at least one FNN configuration, which includes settings for a FNN operation topology for the MAC layer to perform computations for at least one FNN layer. Optionally, the neural network may be a CNN and the configurations may further include at least one CNN configuration for the MAC layer to perform computations for at least one CNN layer.

Abstract: Systems, apparatus and methods are provided for determining whether data associated with a command in a storage system is hot or cold. An apparatus may comprise a first interface to be coupled to a host and circuitry configured to receive a command via the first interface from the host. The command may contain an address for data storage in a data storage system and a length for a size of data associated with the command. The circuitry may be further configured to classify the command into a cluster of a plurality of clusters, obtain a plurality of historical commands in the cluster and determine whether the data associated with the command is hot or cold using a machine learning model. The address and the length of the command and addresses and lengths of the plurality of historical commands may be used as input features to the machine learning model.

Abstract: Systems, apparatus and methods are provided for compressing data. A method may include receiving an input data block to be compressed, determining numbers of occurrences for distinct symbols in the input data block, generating reduced numbers of occurrences for the distinct symbols based on the numbers of occurrences for the distinct symbols and encoding the input data block using the reduced numbers of occurrences as probability distribution of the distinct symbols in the input data block.

Abstract: Systems, apparatus and methods are provided for compressing data. A method may comprise receiving an input data block to be compressed, determining numbers of occurrences for distinct symbols in the input data block, generating reduced numbers of occurrences for the distinct symbols based on the numbers of occurrences for the distinct symbols and encoding the input data block using the reduced numbers of occurrences as probability distribution of the distinct symbols in the input data block.

Abstract: Systems, apparatus and methods are provided for compressing data and decompressing compressed data. A method may comprise receiving an input data block to be compressed, generating a number of occurrences table and a cumulative occurrences table for distinct symbols in the input data block, for each symbol in the input data block, based on the number of occurrences table and the cumulative occurrences table, dynamically obtaining a number of shifts for right-shifting a current state “x” to encode a current symbol, outputting right-shifted bits to encoded data and obtaining a next state “x” and obtaining a final state “X” from a last state “x” generated in a final loop.

Abstract: Systems, apparatus and methods are provided for decompressing compressed data. A method may include receiving encoded data to be decompressed, obtaining a “Stotal” of a total number of symbols, numbers of occurrences for distinct symbols in the encoded data, and a final state generated during an encoding process as a first state for decoding, building a decoding table containing a row of the distinct symbols, a row of substitutes for numbers of bits to be recovered and a row of substitutes for new states corresponding to encoding states, decoding using the decoding table including: obtaining a current symbol from the decoding table based on a current state X, dynamically determining a current number of bits to be recovered from the encoded data and a new state X based on corresponding substitutes and outputting symbols recovered. L may be a sum of the numbers of occurrences for the distinct symbols.

Abstract: Systems and methods are provided for decoding a codeword encoded by a linear block code. A method may comprise performing a hard decision decoding on a codeword, determining which check nodes are satisfied and which check nodes are unsatisfied after the hard decision decoding, scheduling a check node processing order by moving at least one unsatisfied check node to be processed ahead of at least one satisfied check node and performing a soft decision decoding on the codeword according to the check node processing order.

Abstract: Systems, apparatus and methods are provided for compressing data and decompressing compressed data. A method may include receiving encoded data to be decompressed, obtaining a size “Stotal” of a total number of symbols in the encoded data, numbers of occurrences for distinct symbols in the encoded data, a number L of sum of the numbers of occurrences for the distinct symbols, and a final state generated during an encoding process as a first state for decoding, building a decoding table containing the distinct symbols corresponding to L encoding states, decoding the encoded data using the decoding table including: obtaining a current symbol from the decoding table based on a current state X, dynamically determining a current number of bits to be recovered from the encoded data and a new state X, and outputting symbols recovered from the encoded data.

Abstract: Systems, apparatus and methods are provided for compressing data. An exemplary method may comprise interleaving one or more literal length fields with one or more literal fields to an output. The literal fields may contain a first data segment literally copied to the output, and each of the one or more literal length fields may contain a value representing a length of a succeeding literal field. The method may further comprise determining a second data segment being matched to a previously literally copied sequence of data and a match position and writing to the output one or more match length fields and a match position field containing the match position. The literal length fields may contain a total length of the first data segment and the match length fields may contain a total length of the second data segment.

Abstract: Systems, apparatus and methods are provided for determining whether data associated with a command in a storage system is hot or cold. An apparatus may comprise a first interface to be coupled to a host and circuitry configured to receive a command via the first interface from the host. The command may contain an address for data storage in a data storage system and a length for a size of data associated with the command. The circuitry may be further configured to classify the command into a cluster of a plurality of clusters, obtain a plurality of historical commands in the cluster and determine whether the data associated with the command is hot or cold using a machine learning model. The address and the length of the command and addresses and lengths of the plurality of historical commands may be used as input features to the machine learning model.

Abstract: The present disclosure relates to methods and systems for decoding a Bose-Chaudhuri-Hocquenghem (BCH) encoded codeword. The methods-may include receiving a codeword over a data channel; determining a plurality of syndrome values for the codeword during a first time interval; determining a set of initial elements during the first time interval; generating an error locator polynomial based on the plurality of syndrome values, the error locator polynomial representing one or more errors in the codeword; evaluating, based on the set of initial elements, the error locator polynomial to identify one or more error locations corresponding to the one or more errors in the codeword; and correcting the codeword based on the one or more error locations.

Abstract: Systems, apparatus and methods are provided for compressing data. An exemplary method may comprise interleaving one or more literal length fields with one or more literal fields to an output. The literal fields may contain a first data segment literally copied to the output, and each of the one or more literal length fields may contain a value representing a length of a succeeding literal field. The method may further comprise determining a second data segment being matched to a previously literally copied sequence of data and a match position and writing to the output one or more match length fields and a match position field containing the match position. The literal length fields may contain a total length of the first data segment and the match length fields may contain a total length of the second data segment.

Abstract: The present disclosure relates to methods and systems for decoding a low density parity check (LDPC) encoded codeword. The methods may include receiving a codeword over a data channel. The codeword may be encoded with a preset number of data bits having one or more shortened data bits. The methods may also include obtaining a parity check matrix that defines relationships between a plurality of variable nodes and a plurality of check nodes. The methods may further include decoding the codeword by iteratively estimating values with respect to the codeword at the plurality of variable nodes and the plurality of check nodes. During each iteration, a same part of the plurality of variable nodes related to one or more shortened data bits are skipped from estimation.

Abstract: The present disclosure relates to methods and systems for decoding a Bose-Chaudhuri-Hocquenghem (BCH) encoded codeword. The methods may include receiving a codeword over a data channel; determining a plurality of syndrome values for the codeword during a first time interval; determining a set of initial elements during the first time interval; generating an error locator polynomial based on the plurality of syndrome values, the error locator polynomial representing one or more errors in the codeword; evaluating, based on the set of initial elements, the error locator polynomial to identify one or more error locations corresponding to the one or more errors in the codeword; and correcting the codeword based on the one or more error locations.

Abstract: Systems, apparatus and methods are provided for compressing data. An exemplary method may comprise interleaving one or more literal length fields with one or more literal fields to an output. The literal fields may contain a first data segment literally copied to the output, and each of the one or more literal length fields may contain a value representing a length of a succeeding literal field. The method may further comprise determining a second data segment being matched to a previously literally copied sequence of data and a match position and writing to the output one or more match length fields and a match position field containing the match position. The literal length fields may contain a total length of the first data segment and the match length fields may contain a total length of the second data segment.

Abstract: Systems and methods are provided for decoding a codeword encoded by a linear block code. A method may comprise performing a first decoding on a codeword using a first decoder, determining a number of satisfied check nodes and a number of unsatisfied check nodes for a symbol value of the codeword for a decoding result of the first decoding, generating a soft log-likelihood ratio (LLR) for the symbol value based on the number of satisfied check nodes and the number of unsatisfied check nodes, and performing a second decoding using a second decoder with the soft LLR as an input to the second decoder.

Abstract: Systems and methods are provided for decoding a codeword encoded by a linear block code. A method may comprise performing a hard decision decoding on a codeword, determining which check nodes are satisfied and which check nodes are unsatisfied after the hard decision decoding, scheduling a check node processing order by moving at least one unsatisfied check node to be processed ahead of at least one satisfied check node and performing a soft decision decoding on the codeword according to the check node processing order.

Abstract: The present disclosure relates to methods and systems for decoding a low density parity check (LDPC) encoded codeword. The methods may include receiving a codeword over a data channel. The codeword may be encoded with a preset number of data bits having one or more shortened data bits. The methods may also include obtaining a parity check matrix that defines relationships between a plurality of variable nodes and a plurality of check nodes. The methods may further include decoding the codeword by iteratively estimating values with respect to the codeword at the plurality of variable nodes and the plurality of check nodes. During each iteration, a same part of the plurality of variable nodes related to one or more shortened data bits are skipped from estimation.

Abstract: Systems and methods are provided for decoding a codeword encoded by a linear block code. A method may comprise performing a soft decision decoding on a codeword, generating a hard decision for each bit of the codeword at an end of the soft decision decoding, generating a hard decision for each bit of the codeword at an end of the soft decision decoding, generating a reliability determination for each hard decision and performing a hard decision decoding using the hard decision for each bit and reliability determination for each hard decision.