Re: Ideas for Ada 200X

[Ole-Hjalmar Kristensen <oleh@vlinux.voxelvision.no>]

18k11tm001@sneakemail.com (Russ) writes:

> Bill Findlay <yaldnifw@blueyonder.co.uk> wrote in message news:<BB02E370.30D1%yaldnifw@blueyonder.co.uk>...
> > On 3/6/03 23:20, in article bebbba07.0306031420.69c20f71@posting.google.com,
> > "Russ" <18k11tm001@sneakemail.com> wrote:
> > 
> > > I know that Ada is preferable for safety-critical applications, but
> > > based on what you are telling me, it sure seems to be a poor choice
> > > for non-safety-critical simulation with heavy number crunching. Thanks
> > > for clarifying that. I guess C++ is good for something after all, eh?
> > 
> > Only if getting results *fast* is more important than getting them *right*.
> > 8-)
> 
> I prefer to get my results both fast *and* right. :^) Seriously, we
> are currently doing some "fast-time" sims that take something like 8
> hours of clock time. Two hours would be a lot more convenient, but 32
> would be a lot less.
> 
> > (And let's not forget that Ada *can* compute the result matrix in-place,
> > just not using the notation you prefer. It's a strange performance-oriented
> > programmer who gives up a factor of 4 for aesthetic reasons.)
> 
> But Mr. Duff says that "+=" is just "syntactic sugar" for "Add(A,B),"
> and Mr. Eachus says "+=" needs a temporary to preserve A in the case
> of a Constraint_Error. So, as far as efficiency is concerned, we're
> screwed either way, aren't we?

Not if you explicitly turn off the constraint checking.  If you really
want performance for matrix operations, consider rewriting the matrix
operations in assembler using SSE or equivalent instructions. A
speedup with a factor of 4 is realistic on Intel platforms.
Gnat has excellent assembler facilites, btw.

-- 
Ole-Hj. Kristensen

******************************************************************************
* You cannot consistently believe this sentence.
******************************************************************************