Quote:
> So BFP is not understood by the floating point feature on IBM S/370?
The hardware to handle BFP has been been available for a few years
now. The first POP I see mentioning it is in the bookshelf for OS/390
2.8, circa 1999.
Quote:
> I frankly do not understand the purpose of having data on a platform in
> a format not understood on that platform, but once you know what you
> want to do and the format of the data involved it should not be much of
> a task to write a conversion routine?
> What you need seems to be just a precise definition of your BFP format?
The POP has that. The problem of converting long BFP to 32 bit binary is
what to do about values that don't convert? Do you truncate past the
decimal point? What about values that are beyond +/- 2G (for signed
binary)? If there is such a routine for HFP, then that could be used by
first converting the BFP number to HFP (a single instruction).
A very nice essay on IEEE floating point and the decisions made on how
to implement on S/390 (and z/Arch of course) is here:
http://www.research.ibm.com/journal/rd/435/abbott.html
Quote:
> Sven
>> > Excuse me, but what is BFP?
>> Binary Floating Point. Same as IEEE floating point. Not the same as HFP
> (Hex
>> Floating Point) -- the scheme originally used on IBM mainframes.
>> --
>> -----------------------------------------------------------------
>> | Edward E. Jaffe | |
>> | Phoenix Software International | Tel: (310) 338-0400 x318 |
>> | 5200 W Century Blvd, Suite 800 | Fax: (310) 338-0801 |
>> | Los Angeles, CA 90045 | http://www.phoenixsoftware.com |
>> -----------------------------------------------------------------