A method and system for reducing power consumed in processing units when processing units are used to calculate computationally expensive linear functions on a sequence of correlated data. Processing of a new data sample may be performed to consume less power by using results obtained from the processing a previous reference data sample.

A processor-implemented method of processing neural network data includes: setting first limit data by performing a first operation based on first input data and weight data generated from weights included in a filter; comparing the first limit data with an intermediate result of a second operation performed based on second input data and the weight data; and determining whether to perform a subsequent second operation based on a result of the comparing.

A processor-implemented method of processing neural network data includes: setting first limit data by performing a first operation based on first input data and weight data generated from weights included in a filter; comparing the first limit data with an intermediate result of a second operation performed based on second input data and the weight data; and determining whether to perform a subsequent second operation based on a result of the comparing.

Compressed domain processors configured to perform operations on data compressed in a format that preserves order. The Compressed domain processors may include operations such as addition, subtraction, multiplication, division, sorting, and searching. In some cases, compression engines for compressing the data into the desired formats are provided.

Compressed domain processors configured to perform operations on data compressed in a format that preserves order. The Compressed domain processors may include operations such as addition, subtraction, multiplication, division, sorting, and searching. In some cases, compression engines for compressing the data into the desired formats are provided.

A processor-implemented method of processing neural network data includes: setting first limit data by performing a first operation based on first input data and weight data generated from weights included in a filter; comparing the first limit data with an intermediate result of a second operation performed based on second input data and the weight data; and determining whether to perform a subsequent second operation based on a result of the comparing.

A processor-implemented method of processing neural network data includes: setting first limit data by performing a first operation based on first input data and weight data generated from weights included in a filter; comparing the first limit data with an intermediate result of a second operation performed based on second input data and the weight data; and determining whether to perform a subsequent second operation based on a result of the comparing.

A data processing platform, method, and program product perform compression and decompression of a set of data items. Suffix data and a prefix are selected for each respective data item in the set of data items based on data content of the respective data item. The set of data items is sorted based on the prefixes. The prefixes are encoded by querying multiple encoding tables to create a code word containing compressed information representing values of all prefixes for the set of data items. The code word and suffix data for each of the data items are stored in memory. The code word is decompressed to recover the prefixes. The recovered prefixes are paired with their respective suffix data.

A structure of floating-point number data stored in a storage medium according to an embodiment is provided with a first partial code obtained by encoding all or part of an exponent of a floating-point number using variable-length coding, and a second partial code including a significand of the floating-point number. The length of the combined code of the first partial code and the second partial code is fixed, and the end bit of the first partial code and the least significant hit of the second partial code are adjacent to each other.

A structure of floating-point number data stored in a storage medium according to an embodiment is provided with a first partial code obtained by encoding all or part of an exponent of a floating-point number using variable-length coding, and a second partial code including a significand of the floating-point number. The length of the combined code of the first partial code and the second partial code is fixed, and the end bit of the first partial code and the least significant hit of the second partial code are adjacent to each other.