Re: Fortran decimal anyone?
From: James Parsly (japarsly_at_tva.gov)
Date: 04/26/04
- Next message: Herman D. Knoble: "Re: How are derived data types stored in memory?"
- Previous message: Atmapuri: "Run time library?"
- In reply to: David Frank: "Re: Fortran decimal anyone?"
- Next in thread: Mike Cowlishaw: "Re: Fortran decimal anyone?"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Date: Mon, 26 Apr 2004 10:05:37 -0400
No, it supported the 8087. It actually came with four math libraries.
The default library would use an 8087 if it was available, but would
still work if you didn't have one.
If you knew that all machines would have an 8087, you could use
8087.LIB. If you knew that they all would not, you would use
ALTMATH.LIB.
If you wanted to use the decimal math option, you compiled with
the $DECMATH metacommand, and used DECMATH.LIB in
the link.
I still use this compiler (under XP pro), mostly when I want to fire off a
10 line program to dig data out of a binary file. I've also got CVF 6.6c
installed for real work. Things have come a long way since Fortran II
on the IBM 1620.
"David Frank" <dave_frank@hotmail.com> wrote in message
news:0Wbic.72866$I83.1501146@twister.tampabay.rr.com...
>
> "James Parsly" <japarsly@tva.gov> wrote in message
> news:newscache$7xvmwh$kjm$1@lyris.knx.tva.gov...
> > The old Microsoft Fortran 3.31 compiler (a F77 subset compiler)
> > has decimal data types. I've never used the option but, looking
> > at the manual, it refers to single and double precision
> > decimal floating point numbers:
> >
> > "Decimal floating point numbers consist of a byte containing a
> > sign bit and a 7 bit exponent in excess 64 notation followed by
> > a mantissa consisting of 6 (single) and 14 (double) binary coded
> > decimal digits packed two to a byte (if the exponent byte is 0,
> > the number is 0.) The notation for decimal floating point constants
> > follows the same format as standard FORTRAN real and double
> > precision constants."
> >
> > Apparently you turned this all on using the $DECMATH metacommand
> > and then linked with a special decimal match library.
> >
>
> Didnt know that, but that support was probably dropped in favor of
> supporting the 8087 co-processor chip at its next (4.0 ?) release.
>
>
- Next message: Herman D. Knoble: "Re: How are derived data types stored in memory?"
- Previous message: Atmapuri: "Run time library?"
- In reply to: David Frank: "Re: Fortran decimal anyone?"
- Next in thread: Mike Cowlishaw: "Re: Fortran decimal anyone?"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Relevant Pages
|
