Data inspection for compression/decompression configuration and data type determination

A method comprising:

• selecting one or more compression techniques of data based, at least in part, on one or more identified data lengths of the data, wherein the data lengths of the data are identified based, at least in part, on one or more patterns of bits of data to be compressed by one or more datatype compression techniques;
• determining, based on distribution of data in a subset of the data, an underlying data type of the data;
• and
• replacing, using the underlying data type, one or more elements in the data having a negative value with zero values before compression of the data.

A data compressor including a processing circuit to:

• select one or more compression techniques of data based, at least in part, on one or more identified data lengths of the data, wherein the data lengths of the data are identified based, at least in part, on one or more patterns of bits of data to be compressed by one or more datatype compression techniques;
• determine, based on distribution of data in a subset of the data, an underlying data type of the data;
• and
• replace, using the underlying data type, one or more elements in the data having a negative value with zero values before compression of the data.

A non-transitory memory storing instructions that when executed by at least one processor control the at least one processor to perform steps comprising:

• selecting one or more compression techniques of data based, at least in part, on one or more identified data lengths of the data, wherein the data lengths of the data are identified based, at least in part, on one or more patterns of bits of data to be compressed by one or more datatype compression techniques;
• determining, based on distribution of data in a subset of the data, an underlying data type of the data;
• and
• replacing, using the underlying data type, one or more elements in the data having a negative value with zero values before compression of the data.

A computer system comprising:

• a processor;
• a main memory storing data to be compressed;
• and
• cache system including a cache memory, a cache encoder, and a cache controller, wherein the cache controller is in communication with the processor and the main memory, and is configured to:
• receive the data stored in the main memory;
• select one or more compression techniques based, at least in part, on one or more identified data lengths of the data, wherein the data lengths of the data are identified based, at least in part, on one or more patterns of bits of data to be compressed by one or more datatype compression techniques;
• determine, based on distribution of data in a subset of the data, an underlying data type of the data;
• replace, using the underlying data type, one or more elements in the data having a negative value with zero values before compression of the data;
• and
• configure the cache encoder to compress the data using the selected compression techniques.

A data decompressor comprising:

• an input circuit to;
• receive (a) compressed data, and (b) a detection parameter generated by a compressor that indicates one or more compression techniques selected based, at least in part, on one or more identified data lengths of the data, wherein the data lengths of the data are identified based, at least in part, on one or more patterns of bits of data to be compressed by one or more datatype compression techniques;
• determine, based on distribution of data in a subset of the data, an underlying data type of the data;
• replace, using the underlying data type, one or more elements in the data having a negative value with zero values before compression of the data;
• a mode selector that selects from a plurality of decoding modes based at least in part on the detection parameter;
• and
• a decoder that operates in the selected decoding mode to decode the compressed data.