A method for data compression may include scanning input data, performing, based on the scanning, a compression operation to generate compressed data using the input data, finding, based on the scanning, a delimiter in the input data, and generating, based on a position of the delimiter in the input data, a portion of data using the compressed data. The input data may include a record, the delimiter indicates a boundary of the record, and the portion of data may include the record. The generating may include generating the portion of data based on a portion size. The portion size may be a default portion size. The portion size may be based on a default portion size and a length of a match in the input data.

A method for data compression may include scanning input data, performing, based on the scanning, a compression operation to generate compressed data using the input data, finding, based on the scanning, a delimiter in the input data, and generating, based on a position of the delimiter in the input data, a portion of data using the compressed data. The input data may include a record, the delimiter indicates a boundary of the record, and the portion of data may include the record. The generating may include generating the portion of data based on a portion size. The portion size may be a default portion size. The portion size may be based on a default portion size and a length of a match in the input data.

A hardware implementable lossless data compression decompression algorithm is disclosed, where the input data string is described in term of consecutive groups of alternating same type bits, where one of these groups of same type bits is defined as a preferred group with the other groups having either lower or higher number of same type bits, where the data string is partitioned into variable length processing strings where the variable length is determined by the occurrence of the preferred group or of a determined number of bits consisting of groups of lower number of same type bits, where these variable length processing strings are processed function of the configuration and content of each processing string only, where consecutive processing strings are additionally processed based on their content only, where processing is performed in a loop until a certain target performance is achieved, where processing is done without any data analysis, and where no negative compression gain is achieved for any content of an input string.

A hardware implementable lossless data compression decompression algorithm is disclosed, where the input data string is described in term of consecutive groups of alternating same type bits, where one of these groups of same type bits is defined as a preferred group with the other groups having either lower or higher number of same type bits, where the data string is partitioned into variable length processing strings where the variable length is determined by the occurrence of the preferred group or of a determined number of bits consisting of groups of lower number of same type bits, where these variable length processing strings are processed function of the configuration and content of each processing string only, where consecutive processing strings are additionally processed based on their content only, where processing is performed in a loop until a certain target performance is achieved, where processing is done without any data analysis, and where no negative compression gain is achieved for any content of an input string.

A system and a method provide compression and decompression of weights of a layer of a neural network. For compression, the values of the weights are pruned and the weights of a layer are configured as a tensor having a tensor size of H×W×C in which H represents a height of the tensor, W represents a width of the tensor, and C represents a number of channels of the tensor. The tensor is formatted into at least one block of values. Each block is encoded independently from other blocks of the tensor using at least one lossless compression mode. For decoding, each block is decoded independently from other blocks using at least one decompression mode corresponding to the at least one compression mode used to compress the block; and deformatted into a tensor having the size of H×W×C.

A system and a method provide compression and decompression of weights of a layer of a neural network. For compression, the values of the weights are pruned and the weights of a layer are configured as a tensor having a tensor size of H×W×C in which H represents a height of the tensor, W represents a width of the tensor, and C represents a number of channels of the tensor. The tensor is formatted into at least one block of values. Each block is encoded independently from other blocks of the tensor using at least one lossless compression mode. For decoding, each block is decoded independently from other blocks using at least one decompression mode corresponding to the at least one compression mode used to compress the block; and deformatted into a tensor having the size of H×W×C.

Embodiments of the subject matter described herein relate to a wireless programmable media processing system. In the media processing system, a processing unit in a computing device generates a frame to be displayed based on a graphics content for an application running on the computing device. The frame to be displayed is then divided into a plurality of block groups which are compressed. The plurality of compressed block groups are sent to a graphics display device over a wireless link. In this manner, both the generation and the compression of the frame to be displayed may be completed at the same processing unit in the computing device, which avoids data copying and simplifies processing operations. Thereby, the data processing speed and efficiency is improved significantly.

A multichannel data packer includes a plurality of two-input multiplexers and a controller. The plurality of two-input multiplexers is arranged in 2.sup.N rows and N columns in which N is an integer greater than 1. Each input of a multiplexer in a first column receives a respective bit stream of 2.sup.N channels of bit streams. Each respective bit stream includes a bit-stream length based on data in the bit stream. The multiplexers in a last column output 2.sup.N channels of packed bit streams each having a same bit-stream length. The controller controls the plurality of multiplexers so that the multiplexers in the last column output the 2.sup.N channels of bit streams that each has the same bit-stream length.

A multichannel data packer includes a plurality of two-input multiplexers and a controller. The plurality of two-input multiplexers is arranged in 2.sup.N rows and N columns in which N is an integer greater than 1. Each input of a multiplexer in a first column receives a respective bit stream of 2.sup.N channels of bit streams. Each respective bit stream includes a bit-stream length based on data in the bit stream. The multiplexers in a last column output 2.sup.N channels of packed bit streams each having a same bit-stream length. The controller controls the plurality of multiplexers so that the multiplexers in the last column output the 2.sup.N channels of bit streams that each has the same bit-stream length.

Embodiments of the present disclosure provide systems and methods for reverse decompression. According to one embodiment of the present disclosure, the method for reverse decompression includes receiving encoded and compressed input data in a form of one or more data blocks and locating an end of block marker for a last block of the one or more data blocks of the input data. The method also includes traversing the input data, bit by bit, in a reverse direction starting from a last bit of the end of block marker of the last block of the one or more data blocks of the input data towards a beginning of the input data, determining if one block of the one or more blocks of the input data can be designated as a valid block, designating the one block as a valid block and decompressing the valid block in a forward direction.