Ada, Alpha & IEEE Computational Models?

Ada, Alpha & IEEE Computational Models?

[John Connett]

The Alpha architecture provides a choice of three IEEE computational
models:

    *   IEEE compliant arithmetic;
    *   IEEE compilant arithmetic without inexact exception;
    *   High performance IEEE-format arithmetic.

The model used is controlled by instruction qualifiers.  The choice of
model can have an impact on the speed of the code.

For the language systems with which I am familiar, gcc/egcs on Alpha
Linux, the choice of computational model is made on a per compilation
unit basis through the use of compile time switches.

The use of pragmas would seem to offer a mechanism for finer control of
the computational model within compilation units.  Do any existing Ada
language systems which target the Alpha architecture offer this level of
control?

Any pointer to further information on this topic would be gratefully
received.

Regards
--
John Connett (jrc@art.co.uk)

Re: Ada, Alpha & IEEE Computational Models?

[dewar]

In article <36A88A73.BF261F63@art.co.uk>,
  John Connett <jrc@art.co.uk> wrote:

> The use of pragmas would seem to offer a mechanism for
> finer control of the computational model within
> compilation units.  Do any existing Ada language systems
> which target the Alpha architecture offer this level of
> control?

GNAT compilers for the Alpha fully implement the DEC
pragmas (Float_Representation, Long_Float) which provide
fine control of the kind you are looking for. The default
is IEEE, which is of course usually preferable, but
Vax_Float is supported particularly for compatibility with
existing legacy code and data files.

Robert Dewar
Ada Core Technologies

-----------== Posted via Deja News, The Discussion Network ==----------
http://www.dejanews.com/       Search, Read, Discuss, or Start Your Own