P754 Working Group Members:The Ballot Review Committee has completed the task of addressing the 659 comments received on P754 Draft 1.5 and now asks that you review the Draft 1.6 of the IEEE Standard for Floating Point Arithmetic. For all non-ballotiing body member of the working group; please submit rogue comments to Dan or myself.
I draft attached is the difference draft with the changes from 1.5 to 1.6 clearly shown in underline and, for those using a pdf viewer, the details of the change.
The results of the ballot were:64 eligible people in this ballot group64 eligible people in this ballot group
45 Affirmative votes 16 Negative votes with comments 1 Abstention vote ----- 62 votes received = 97% return rate 2% abstain rate Approval Rate 45 Affirmative vote 16 Negative votes ---- 61 votes = 74% affirmative THIS BALLOT FAILED to meet the 75% requiredOf the comments received, ~350 were accepted as Agree, or Agree-in-Principle, with resulting changes to the text of the standard. Note that many of the comments were duplicates. The BRC disagreed with less than 150 of the comments, one was ruled as out of scope, 5 were truly Unresolvable.
We, the BRC, wish to thank the very diligent effort of the members of the balloting body and the other commenters for finding typos and other errors in the document that have been present for many versions of the draft. We hope they are all fixed now.
Major changes to the draft resulting from multiple comment on each were:1. Substantial work went into revision of Clause 3 “Floating-point formats” to simplify the description of the formats used in this standard. N.B. No changes have been made to the basic formats.
2. The addition of the Index of operations.3. Clause 11 has changes with additional guidance to programmers to support reproducible floating point results in keeping with the purpose of the standard.
4. Clause 9 “Recommended operations” has received a lot of attention to get the details correct.
Please pay particular attention to these sections.Interval arithmetic is the largest unresolved issue with this draft. The submissions from the IFIP’s WG2.5 were considered. Adding interval arithmetic to this standard at this time was a task was beyond the scope of the BRC’s work and will need a working group created for an interval arithmetic standard. The Microprocessor Standards Committee is willing to sponsor such a standard activity.
This is a 15 day recirculation ballot. Sincerely, The Ballot Review Committee: Bob Davis, Chair Dan Zuras, P754 Chair Roger Golliver Mike Cowlishaw Peter Markstein Jeff Kidder William Kahan
Attachment:
P754-160-diff-150.pdf
Description: Adobe PDF document