Float Decimal (p) DFP, IEEE-754-2008 - Intel platforms

Attention: This topic applies to a feature that is in Early Adopter Program (EAP) release status. We intend to provide the finalized feature in a future release. Please contact Micro Focus Customer Care if you require further clarification.

IEEE-754-2008 Decimal Floating Point (DFP) is an IEEE standard that allows precise calculations that are identical regardless of platform. This standard allows data to be held in Densely Packed Decimal (DPD) format or in Binary Integer Decimal (BID) format.

Densely Packed Decimal (DPD) and Binary Integer Decimal (BID) formats

Enterprise Developer uses a DFP Binary Integer Decimal (BID) Library, provided by Intel as open source software, as the underlying software in our implementation for supporting Decimal Floating Point. Our implementation has the following key attributes:

Densely Packed Decimal (DPD) is always stored in big-endian format on Intel platforms. This format is mainframe compatible.
Binary Integer Decimal (BID) format is always stored in little-endian format on Intel platforms. BID format is different from DPD format and is NOT mainframe compatible, irrespective of big-endian or little-endian storage format.
Data manipulation, with a few exceptions, is always done using BID format native data.
You may store data in DPD format or BID format; however, because it requires less overhead, BID format is more efficient than DPD format. Therefore, we recommend that you use BID format unless compatibility is required with the mainframe due to file content, SQL/DB2, or other such mainframe technologies.

Storage requirements and maximum values for FLOAT DECIMAL (p), where p represents the precision, are as follows:

	Maximum Value
`p` <= 7 4 bytes	9.999999E+0096
7 < `p` <= 16 8 bytes	9.999999999999999E+0384
16 < `p` <= 34 16 bytes	9.999999999999999999999999999999999E+6144

Densely Packed Decimal (DPD) format: Float decimal data can be stored and processed as DPD float decimal data by using the -dfp compiler option or *PROCESS FLOAT(DFP).
Binary Integer Decimal (BID) format: Use the -bid compiler option in conjunction with the -dfp option to store DFP FLOAT DECIMAL data in Intel native BID format.