Hi,
that question was asked a long time ago. I hope this answer helps you.
It would be better to ask it in the 'COBOL general discussion' forum.
In CA-REALIA COBOL you have the following comp source definitions:
COMP.
COMP-3.
COMP-4.
COMP-5.
I have never seen the implementation of floating point with COMP-1 and COMP-2. But my knowledge of CA-REALIA COBOL is until version 5. (I asked CA for an update-offer, but never got one)
In DB2 you will use COMP-3: the packed-decimal format. In mainframe environment COMP, COMP-4 and COMP-5 are the same. In PC environment, COMP and COMP-4 are the same - binary - but COMP-5 is the native format. On the PC as implemented in CA-REALIA COBOL it is the INTEL format. It is a little bit faster than the binary format. I use it all the time.
Regards,
Crox