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

Re: A proposal for the next motion



> Date: Sun, 17 May 2009 22:34:39 -0400
> From: Nate Hayes <nh@xxxxxxxxxxxxxxxxx>
> Subject: Re: A proposal for the next motion
> To: STDS-1788@xxxxxxxxxxxxxxxxx
> 
> Vincent Lefevre wrote:
> > Dear Nate,
> > 
> > On 2009-05-13 20:47:53 -0400, Nate Hayes wrote:
> >> Dear Vincent,
> >> 
> >>> Vincent Lefevre wrote:
> >>> On 2009-05-13 18:23:30 -0400, Nate Hayes wrote:
> >>>> Global flags are technology from the 70's and don't work well in
> >>>> modern multi-core, multi-threaded computing environments.
> >>> 
> >>> . . .
> 
> My understanding is IEEE 754-2008 attempts to stop speaking of global
> flags and instead speak of "attributes". For example, a conforming
> application could provide four op-codes for addition, one each with a
> different hard-wired rounding mode. This is not unlike the examples in
> Ulrich's letters to the IEEE.
> 
> It is a more modern concept, i.e., "attribute" (I think I also heard Michel
> refer to this idea as "local flag"). The main idea is there is nothing that
> requires it to be global, like the rounding-mode control flags on current
> x87 FPUs, for example (which can flush the pipeline and give processor
> stalls when you change them).
> 
> I don't actually have a copy of IEEE 754-2008 yet, so I'll let Dan or Michel
> correct me if I misrepresent this idea.
> 
> . . .
> 
> Nate

	Nate,

	Vincent has it correct with attributes as modes.  In a sense,
	Michel does as well but attributes as flags is a bit misleading.

	"Attributes" (nee modes) are the subject of Clause 4: Attributes
	and Rounding.  And we tried to make it clear that they can have
	a local scope.  This covers what used to be called rounding
	modes as well as exeption modes.

	But "flags" are the subject of Clause 7: Default Exception
	Handling & Clause 8: Alternate Exception Handling Attributes.

	In Clause 7 the use of 'flags' is much the same as in 754-1985
	but as have tried to make the distinction between 'exception'
	(the thing that caused the trouble) & 'flag' (the place it was
	recorded).

	In Clause 8 the word 'attributes' is used in an attempt to apply
	scope to exceptional results.

	But Clause 8 is, unfortunately, optional.  So we can't count on
	it to be there even if we were limiting ourselves to a 754-2008
	environment (which it seems we are not doing).

	Still, we can require more of an implementation than the minimum
	of 754-2008.  So if you think its important we could specify
	that they implement the option of Clause 8.  But if we couldn't
	get restriction to 754-2008 to pass I doubt we could get that to
	pass.  IMHO, of course.

	BTW, get your copy of 754-2008.  You may find that some of your
	problems with modals can be attacked by things in there.  My
	reply to Vincent's previous post should give a hint.

	Take care,

					Dan