Thread Links Date Links
Thread Prev Thread Next Thread Index Date Prev Date Next Date Index

Re: Motion 4

> Date: Thu, 9 Apr 2009 13:09:51 +0200
> From: Vincent Lefevre <vincent@xxxxxxxxxx>
> To: stds-1788@xxxxxxxxxxxxxxxxx
> Subject: Re: Motion 4
> 
> On 2009-04-09 09:57:05 +0200, Paul Zimmermann wrote:
> > I find the terms of this motion quite confusing. Did you mean "P1788
> > compliance implies IEEE 754-2008 compliance"?
> 
> I also find them very confusing.
> 
> IMHO, P1788 compliance should not be related to IEEE 754-2008
> compliance. For instance, NaN support (and even more sNaN/qNaN)
> is probably not needed for interval arithmetic (even though it
> can help the implementation). Ditto for signed zeros.
> 
> Note that even though a processor has NaN and signed zeros support
> for IEEE 754 compliance, a user may want to disable their support
> if he knows that they will not be useful and that this will make
> his program faster (compilers already have such options, so that
> one can't ignore them).
> 
> -- 
> Vincent Lefèvre <vincent@xxxxxxxxxx> - Web: <http://www.vinc17.org/>
> 100% accessible validated (X)HTML - Blog: <http://www.vinc17.org/blog/>
> Work: CR INRIA - computer arithmetic / Arenaire project (LIP, ENS-Lyon)

	Vincent,

	We have said nothing about NaN support yet.  And, indeed,
	there are good reasons for excluding NaNs from valid
	intervals.  I don't think we will attach any significance
	to signed zeros for practical reasons (we cannot control
	them) but I may be wrong.  So the compiler options you
	speak of my be useful.

	But if your real concern is for user defined precisions,
	I will point out that they exist now in 754-2008.  They
	may not be exactly what INRIA & others have been using
	up to now but they are the best & most general that we
	could agree on given the committee process.

	Again, since the purpose is KISS, please weigh the concerns
	you raise against the alternative of having to put up with
	language throughout 1788 that places the needed constraints
	on all sorts of arithmetics with characteristics that vary
	all over the place.

	I think the choice is clear.

	You may disagree.

	That's what voting is all about.

	Yours,

				Dan