Encoding densely packed decimals

A system for encoding densely packed decimals, comprising:

• a decimal floating-point unit comprising:
• an expansion unit operable to expand decimal data in a densely packed decimal (DPD) format to decimal data in a binary coded decimal (BCD) format;
• a pipeline operable to execute a floating point instruction configured to perform a floating point operation on the BCD-formatted decimal data;
• and
• a multiplexer configured to selectively output a decimal data result of the floating point operation in DPD format using one of a standard DPD compression process or a fast pack DPD compression process based on a detection of a rounded mantissa overflow.

A computer-implemented method for encoding densely packed decimals, comprising:

• receiving, from a pipeline of a decimal floating-point unit, decimal data in a binary coded decimal (BCD) format, the BCD-formatted decimal data comprising a result of a floating point operation in the pipeline;
• rounding the BCD-formatted decimal result data to a designated precision;
• determining whether the rounded result data includes a mantissa overflow from the designated precision;
• responsive to determining that the rounded result data does not include a mantissa overflow, applying by the decimal floating-point unit Boolean operations to the rounded result data to compress the rounded result data to decimal data in a densely packed decimal (DPD) format;
• and
• responsive to determining that the rounded result includes a mantissa overflow, compressing the rounded result data from the BCD format to decimal data in the DPD format by mapping select bit values of the BCD formatted rounded result decimal data to select bit positions in the DPD format.