RE: Motion P1788/M0009.01_ExactDotProduct
George,
If by "people on the implementation side", you mean "hardware implementers",
my best guess is that major computer companies won't be willing to add exact
dot products in hardware unless there is a very clear commercial benefit to
doing so. Even though the hardware cost for exact dot products is not bad,
commercial companies also need to take into account the cost of verifying
the design, potential impacts on their schedule, and including it in any
subsequent architectures.
However, since std-1788 does not require that everything be implemented in
hardware, I think we should include exact dot products in the standard and
then people can decide if they want to implement it in hardware or software.
My impression is that a software implementation of exact dot product would
not be prohibitive (except possibly in some very cost-constrained embedded
devices).
Best regards,
Mike
-----Original Message-----
From: stds-1788@xxxxxxxx [mailto:stds-1788@xxxxxxxx] On Behalf Of Ralph
Baker Kearfott
Sent: Friday, October 30, 2009 10:07 AM
To: Corliss, George
Cc: stds-1788@xxxxxxxx
Subject: Re: Motion P1788/M0009.01_ExactDotProduct
George et al,
That's a very good point. All views and interests need to be represented
during our work, to avoid later problems.
Best regards,
Baker
Corliss, George wrote:
> I am VERY INTERESTED to hear the views of those who must build whatever we
> write the standard to require. As a USER, I strongly favor an exact dot
> product, and I hold the opinion that it should not be too hard to
implement.
> However, if the implementors were to say, "We WILL NOT implement an exact
> dot product. Period." I would vote against to avoid a standard that goes
> no where.
>
> Some of my personal life difficulties have convinced me that "I will do
..."
> are powerful words. "You should do ..." can be very damaging.
>
> I'd like to hear people from the implementation side say at least, "We are
> willing to do an exact dot product" before I am willing to say, "You must
do
> an exact dot product."
>
> Dr. George F. Corliss
> Electrical & Computer Engineering
> Haggerty Engineering #296
> Marquette University
> P.O. Box 1881
> 1515 W. Wisconsin Ave.
> Milwaukee WI 53201-1881
> George.Corliss@xxxxxxxxxxxxx
> 414-288-6599; -288-4400 (GasDay); -288-5579 (Fax)
> Www.eng.mu.edu/corlissg
>
>
>
>
> On 10/29/09 9:35 AM, "Ralph Baker Kearfott" <rbk@xxxxxxxxxxxx> wrote:
>
>> P1788 members:
>>
>> The formal discussion period for this motion has ended, and the
>> formal voting period has begun. Votes may be cast until the
>> end of Wednesday, November 18, 2009. The motion is listed in
>> attachments to this message, and may also be obtained from
>> the private area of our working group web site, at
>>
>> http://grouper.ieee.org/groups/1788/
>>
>> (Contact me if you need a password to the private area.)
>>
>> During the voting period, the
>> motion is not subject to amendment. A registered Voting Member may vote
>> "Yes"
>> "No"
>> "No," but propose an amendment that would cause the voter to
>> vote "Yes." The proposed amendment shall include detailed
>> wording and rationale. Such "No" votes on position papers
>> are NOT motions to amend. The purpose is to influence
>> other voters.
>> Of course, anyone may make any statements they wish, but those are not
>> votes.
>>
>> All votes on position papers should be public. The mechanism for voting
is a
>> message broadcast to <stds-1788@xxxxxxxxxxxxxxxxx>. The ideal vote is
>> Subject: Motion P1788/M0009.01_ExactDotProduct YES (or NO)
>> Body: YES (or NO and proposed changes)
>> Name
>>
>> The Voting Tabulator shall count as a vote any message in which the
intent
>> is clear.
>>
>> Instructions for registering for the working group may be found
>> at
>>
>> http://grouper.ieee.org/groups/1788/wg1788Reg.html
>>
>> Information about working group motions and supporting documents
>> can be found in the public and private areas of the P1788 web
>> site, accessible from
>>
>> http://grouper.ieee.org/groups/1788/
>>
>> Please contact me (rbk@xxxxxxxxxxxxx) if you need the user ID
>> and password for the private area.
>>
>>
>> A registered Voting Member may change her/his vote at any time during the
>> voting period simply by broadcasting a fresh voting message.
>>
>> A position paper requires a "Yes" vote by a majority of registered Voting
>> Members to pass. A quorum is 2/3 of the registered Voting Members. If
>> necessary to achieve a quorum, the Voting Tabulator may solicit further
>> votes, in which case, all not-yet-voted registered Voting Members shall
be
>> solicited equally.
>>
>>
>> William: Please record this transaction in the minutes.
>>
>> Juergen: Please update the status on the web page.
>>
>> Best regards to all,
>>
>> Ralph Baker Kearfott
>> (vice chair, P-1788)
>
--
---------------------------------------------------------------
R. Baker Kearfott, rbk@xxxxxxxxxxxxx (337) 482-5346 (fax)
(337) 482-5270 (work) (337) 993-1827 (home)
URL: http://interval.louisiana.edu/kearfott.html
Department of Mathematics, University of Louisiana at Lafayette
(Room 217 Maxim D. Doucet Hall, 1403 Johnston Street)
Box 4-1010, Lafayette, LA 70504-1010, USA
---------------------------------------------------------------