[
Date Prev][
Date Next][
Thread Prev][
Thread Next][
Date Index][
Thread Index]
Re: Meeting the Scope and Purpose of P754
Nick,
It's probably going to happen and what you could contribute, from you
reputed knowledgeable position, is a solution, or set of conditions that
would meet the requirements and not be that bothersome to exciting field
of parallel computation which may not need the same results on a regular
basis.
Telling it to go away is far less useful.
Bob.
Nick Maclaren wrote On 07/13/07 13:49,:
Bob Davis <Bob.Davis@xxxxxxx> wrote:
There has been a large volume of "You can not do this!" email on the
reflector. Much/most of this has been recently directed at the
reproducible results/identical results problem.
Yes, and I plead, er, guilty as charged.
How about making a positive contributions to solve the problems at had
with the P754 Draft. The BRC is pounding through the 200+ comments and
rogue comments in ongoing meetings spending 4-6 hours a week at it.
Hmm. What would you regard as a worthwhile contribution from someone
who has both theoretical and practical reasons to believe that the
objective is neither feasible nor desirable?
Please consider that the pupose of a standard is to provide a common
method of doing something. What other parts of the scope and purose have
been overlooked?
Exception handling, definitely. LIA-1 "Termination with message"
should not merely be permitted, it should be required as an option.
It is related to reproducibility, in that some systems may be able
to provide reproducibility except when an exception occurs.
The goal, is to create a clause that can be quoted in a contract ie "All
computations required by this contract/agreement/treaty shall be
performed in accordance with IEEE754-2007 Clause 1.x.y in binary
mode."(or decimal) with no other elaborating specification. The parties
to the contract/agreement/treaty may have a lot at stake and may be
prepared to back there position/solution with unlimited weapons. There
is good reason to believe that trust level in anything beyond the
written document, will be exceeding low.
What is required to get the same answer by all interested parties? It
does not need to be fast but it must yield the same answer. It does not
need to be applied to all parts of 754 in all applications. What
constraint must be applied?
There are three fundamental issues here, two of which have been
overlooked and the other not taken proper notice of:
1) This is not just a floating-point issue, and requires a great
many other constraints on the language, including character sets,
integer sizes and representations and much else.
2) There are a lot of basic facilities that a program may not use
if you are to attempt reproducibility. For example, it may not read
the time or read any data outside the control of the programmer.
3) The algorithmic and language constraints will be considerable,
I do not know of any research that indicates a suitable subset, and
I have seen no evidence that anyone has put serious thought into
what that subset would look like.
Executive summary: I think that this objective should be abandoned.
That IS a positive contribution, in that I am warning that you are
trying to solve an intractable or impossible problem, and therefore
you should change the problem.
Regards,
Nick Maclaren,
University of Cambridge Computing Service,
New Museums Site, Pembroke Street, Cambridge CB2 3QH, England.
Email: nmm1@xxxxxxxxx
Tel.: +44 1223 334761 Fax: +44 1223 334679
--
Respectfully,
Bob Davis
bob.davis@xxxxxxx
Cell 408.857.1273