Re: Integers and standard

Ken.Fairfield@xxxxxxxxx wrote:

On Jul 29, 12:07 pm, nos...@xxxxxxxxxxxxx (Richard Maine) wrote:
(snip)
Ok. I was probably thinking of G because of my fiddling to convert
between VAX and IEEE formats. G was so easy (as long as you didn't worry
about things like denormals) that I think I did the other VAX formats by
first converting to G (I think there might have been some VAX library
procedure to do that) and then going from G to IEEE (for portable file
formats that wanted IEEE).

From www.bitsavers.org, it seems that VAX/VMS 1.5 only supported
D float, but by VMS 5.0 both D and G had Fortran support, with
D float the default. I believe that some hardware included
hardware support for G-float, while some did not.

And to be a bit more precise, I was talking about *VAX*,
not VMS in general. I make the distinction because G-float
*is* the default for Alpha/VMS, with S-float and T-float (IEEE
types) as compiler options that are natively supported by the
hardware. D-float is a compiler option (for compatibility)
but is really just a "storage format", with calculations done
in G-float and conversion from and to D-float on read and
store. And the IEEE types are default on IA64/VMS.

It seems that for Alpha, load/store for F, D, and G convert
to an IEEE form in registers, including changing the
byte/word ordering. Not so different from load/store on
the x87 which converts to/from extended real.

-- glen

.

Quantcast