Actually, for any "numerical intensive applications" or High-Performance 
computing need to be "Parallel" to be efficient. Any other way of coding 
is totally, a waste of time. Ada 95 should of had threads, and by 2005 a 
parallel design should of been in place.

In paralleling an algorithm there are two main designs, The first is to 
add a paralleling optimizer component to the Ada compiler. This component 
optimizes an algorithm by rewriting parts in parallel form. But most of 
these optimizers requires standardization of the arithmetic processes 
such as using 754.

The second method is to allow the Ada programmer to construct a parallel 
routines. In the second case Ada is not parallel or even threaded. Using 
"Distributed Parallel" design might gives some help depending upon the 
lag time between systems. Too long a lag time kill the efficient also.

Using IEEE-754 standards on an Intel processor for example which does 
not fully support the 754 standards can easily done. By providing a 
conversion or interfacing packages to deal with those non-implement designs 
and as for efficiency its a trade off of being fast or being accurate. With 
efficiency using less accuracy finding someone with GPS, being off a couple 
of meters might not hurt, but using a laser in the operating room accuracy 
counts more than efficiency.

And using standards is more "portable" than depending on the computer 
systems or the OS.



In <isffbc$2jj$1@dont-email.me>, "J-P. Rosen" <rosen@adalog.fr> writes:
>Le 04/06/2011 23:35, anon@att.net a �crit :
>> Now, having a generic Ada math package or design would insure at least 
>> for math that Ada is System Independent. And using a standard like IEEE
>> 754 would insure application accuracy while maintaining that System 
>> Independent.
>> 
>The trouble is that numerical intensive applications often need
>efficient implementations.
>
>If you require IEEE arithmetic, and you are on a machine whose hardware
>does not support IEEE-754 (note: supports and not implements, since
>IEEE-754 is a software standard), you have a choice of 1) implement the
>whole standard in software and say bye-bye to efficiency, or 2) ignore
>the standard and use the native floating-point, which is a failure of
>the standardization.
>
>The Ada model allows portable guaranteed bounds to inaccurracy, without
>requiring any particular arithmetic model, and is therefore efficiently
>implementable on any hardware. That was part of the requirements.
>-- 
>---------------------------------------------------------
>          J-P. Rosen (rosen@adalog.fr)
>Adalog a d�m�nag� / Adalog has moved:
>2 rue du Docteur Lombard, 92441 Issy-les-Moulineaux CEDEX
>Tel: +33 1 45 29 21 52, Fax: +33 1 45 29 25 00