C++ Logo

sg19

Advanced search

Re: [isocpp-sg19] [isocpp-sg14] SG14 and SG19 meeting @ CPPCON Wednesday Sept 17, 13:00 MDT

From: Michael Wong <fraggamuffin_at_[hidden]>
Date: Tue, 7 Oct 2025 16:24:56 -0400
Here are the notes for both sg14 and 19. The notes are also in the wiki as
this was an actual F2F meeting:
https://wiki.edg.com/bin/view/Wg21kona2025/SG14

The meeting discussed the process of joining and contributing to the C++
standardization committee. Michael Wong and others explained the role of
national bodies, the importance of persistence in proposing standards, and
the challenges of virtual participation. Guy Davidson presented his paper
on reproducible floating-point arithmetic, highlighting the need for
consistent floating-point behavior across different platforms, especially
in gaming and financial applications. The committee is currently reviewing
C++26, with national bodies providing feedback. The next meeting is in
Kona, followed by Croydon, where final revisions will be made before
publication. The discussion focused on the reproducibility of
floating-point calculations across different compilers and languages.
Someone highlighted differences in output when using NVIDIA C compiler
version 4.5 versus clang compiler version 81, noting variations in bit
representations. The conversation also covered the lack of standardization
in languages like Ada, COBOL, and Fortran regarding floating-point
arithmetic. Proposals for improving reproducibility included adding annexes
to C++, using pragmas, and introducing new core types.

Next we discussed Richard Dosselmann's basic stats proposal. The need for a
weighted variance function was debated, with some suggesting it should be
included despite the lack of a standard definition. The discussion
concluded with a vote to encourage further work on these proposals. Michael
Wong led a formal vote on a proposal, counting 12 in favor, 0 neutral, and
0 against in person, with 8 more voting online, totaling 20 in favor. Edwin
added his support, making it 21 in favor. The proposal was approved. A
discussion followed about scheduling an evening session for an event in
Kona, considering time zone differences. Michael Wong expressed
appreciation for the session's organization and participants' enthusiasm.
Additionally, there was a brief mention of computing standard deviation and
a book signing, indicating ongoing collaboration and technical discussions
among the participants.
Action Items

   - [ ] Get input from SG 6 on the question of whether to add a weighted
   variance function to the basic statistics proposal.
   - [ ] Revise the basic statistics proposal based on the feedback
   received and prepare to present it at the Kona meeting, pending resolution
   of any SG 6 issues.


Introduction to the Meeting and Participant Roles

   - Guy explains the structure of the meeting, mentioning different study
   groups and incubator groups for language and library evolution.
   - Michael Wong provides additional context on joining the standard and
   the role of national bodies in the ISO organization.
   - Guy and Michael Wong discuss the openness of the group and the
   importance of contributions from new members.
   - Guy introduces the GitHub repository for C++ and encourages
   participants to contribute and reach out to authors of papers.

Stories of National Bodies and Joining Procedures

   - Someone shares a story about a Czech national body starting a C++
   programming language standardization effort.
   - Inbal mentions the Israeli national body having a similar experience.
   - Inbal and Guy discuss the different regulations and fees associated
   with joining national bodies in various countries.
   - Michael Wong shares his personal experience as a Canadian national
   body member and the influence of national bodies on standardization.

Challenges and Strategies for Joining and Contributing

   - Michael Wong emphasizes the importance of persistence in pushing
   through proposals and the challenges of attending face-to-face meetings.
   - Guy and Inbal discuss the intricacies of the standardization process
   and the role of national bodies in reviewing and approving standards.
   - Michael Wong highlights the need for participants to show up for
   meetings, both in person and virtually, to contribute effectively.
   - Nevin provides insights into the mailing process and the importance of
   timely submissions for papers.

Technical Discussion on Reproducible Floating Point Results

   - Guy Davidson introduces his paper on reproducible floating point
   results, emphasizing the importance of determinism in multiplayer games.
   - Davidson explains the current challenges with floating point
   arithmetic in C++ and the need for a standardized approach.
   - The discussion includes the historical context of IEEE 754 and the
   current limitations of C++ in supporting reproducible floating point
   results.
   - Davidson outlines the proposed changes to the standard to ensure
   consistent behavior across different implementations.

Q&A and Further Discussion on Paper Proposals

   - Participants ask questions about the paper and the implications of the
   proposed changes for different use cases.
   - Davidson addresses concerns about the feasibility of implementing the
   changes and the potential impact on existing code.
   - The discussion includes a comparison with other programming languages
   and standards that have addressed similar issues.
   - Participants express interest in continuing the conversation and
   exploring potential collaborations on related topics.

Floating Point Reproducibility Issues

   - Someone discusses the variability in floating point calculations
   across different compilers, specifically comparing NVIDIA C compiler
   version 4.5 and clang compiler version 81.
   - The issue of different bit representations in running programs is
   highlighted, emphasizing the importance of bit representations in network
   terms.
   - Someone mentions the lack of standardization in languages like Ada,
   COBOL, and Fortran regarding floating point arithmetic, contrasting with
   C++'s well-written standards.
   - The discussion includes the challenges of implementing floating point
   arithmetic in C and C++, with C++'s optional extended floating point types
   being a point of contention.

Exploring Compiler Options and Annexes

   - Someone explores various compiler options for ensuring
   reproducibility, such as FP strict and fast math options, but notes that
   these still produce slightly different results.
   - The idea of introducing annexes or specifying language support for
   floating point arithmetic in C++ is discussed, but it is noted that this
   would be a significant undertaking.
   - Inbal asks about the intentions of the C standard committee (WG 14)
   regarding floating point arithmetic, and Guy explains the floating point
   group's meetings and the potential for adding a third annex.
   - The discussion touches on the historical compatibility and specialized
   systems supported by C's floating point annexes, with a focus on Annex F
   and its role in adapting IEEE 754.

Possible Approaches to Reproducibility

   - Guy outlines several possible approaches to achieving reproducibility
   in floating point calculations, including pragma options, function
   qualification, and new core types.
   - The challenges of specification overhaul and the potential for
   introducing new types are discussed, with a preference for less disruptive
   solutions.
   - The idea of using pragmas to control compiler behavior during
   translation is explored, but concerns about their effectiveness and
   compatibility with function templates are raised.
   - The concept of introducing a new core type for reproducible operations
   is considered, but it is noted that this would be a significant undertaking.

Discussion on Function-Based Approaches

   - Michael Wong suggests a function-based approach as a less disruptive
   solution, with a focus on correctly rounded functions and their
   implementation.
   - The discussion includes the potential for nested function calls and
   the need for strict evaluation modes to ensure correct results.
   - The idea of using pragmas to enforce specific orders of operations is
   considered, but concerns about their effectiveness and compatibility with
   existing standards are raised.
   - The potential for introducing new types or library types to support
   reproducibility is discussed, with a preference for less intrusive
   solutions.

P1780R10
<https://wiki.edg.com/bin/edit/Wg21kona2025/P1780R10?topicparent=Wg21kona2025.SG14>
Statistics
for SG19 ML Feedback and Next Steps

   - The discussion includes feedback from participants on the proposed
   approaches, with a focus on the practicality and efficiency of the
   solutions.
   - The importance of ensuring that the proposed solutions are compatible
   with existing standards and do not introduce unnecessary complexity is
   emphasized.
   - The need for further discussions and revisions to the proposals is
   acknowledged, with a preference for addressing the most critical issues
   first.
   - The decision to proceed with the proposals and seek further input from
   relevant committees, such as SG 19 and SG 6, is made, with a focus on
   ensuring that the solutions are well-received and effective.

Voting on the Proposal

   - Michael Wong initiates the voting process, asking if anyone has
   questions about the naming of the proposal.
   - Michael Wong counts the votes: 12 in favor, 0 neutral, 0 against, and
   8 in favor online.
   - Michael Wong confirms the final tally: 12 in favor, 0 neutral, 0
   against, and 8 in favor online, making it 20 in favor in total.
   - Edwin expresses his strong support, bringing the total to 21 in favor.

Approval and Next Steps

   - Michael Wong grants approval to take the proposal to the next level,
   - Inbal asks about the possibility of an evening session for the event,
   considering fewer in-person attendees in Kona.
   - Michael Wong and Inbal discuss the feasibility of an evening session,
   even for remote participants.
   - Michael Wong thanks everyone for their participation and enthusiasm,
   and closes the meeting.



On Wed, Sep 17, 2025 at 6:29 PM Andre Kostur via SG14 <sg14_at_[hidden]>
wrote:

> Mine was camping with no cell signal.
>
> On Wed, Sep 17, 2025 at 3:16 PM Patrice Roy via SG14 <
> sg14_at_[hidden]> wrote:
>
>> Yeah, work meetings ate up my entire day too. Sorry for my absence :(
>>
>> Le mer. 17 sept. 2025 à 15:38, Bryan St. Amour via SG14 <
>> sg14_at_[hidden]> a écrit :
>>
>>> I'm sorry I missed this, the day completely got away from me.
>>>
>>>
>>> Bryan
>>>
>>>
>>> On 2025-09-17 13:15, Robert A.H. Leahy via SG14 wrote:
>>>
>>>
>>> https://cppcon2025.sched.com/event/27l1Z/official-iso-c++-wg21-sg-meeting-of-low-latencyembeddedgamesfinance-and-machine-learning
>>>
>>> --Robert
>>> ------------------------------
>>> *From:* SG14 <sg14-bounces_at_[hidden]>
>>> <sg14-bounces_at_[hidden]> on behalf of andrew drakeford via SG14
>>> <sg14_at_[hidden]> <sg14_at_[hidden]>
>>> *Sent:* Wednesday, September 17, 2025 12:49
>>> *To:* sg14_at_[hidden] <sg14_at_[hidden]>
>>> <sg14_at_[hidden]>
>>> *Cc:* andrew drakeford <andreedrakeford_at_[hidden]>
>>> <andreedrakeford_at_[hidden]>
>>> *Subject:* Re: [isocpp-sg14] SG14 and SG19 meeting @ CPPCON Wednesday
>>> Sept 17, 13:00 MDT
>>>
>>> Where about in cppcon
>>>
>>> Sent from Outlook for iOS <https://aka.ms/o0ukef>
>>> ------------------------------
>>> *From:* SG14 <sg14-bounces_at_[hidden]>
>>> <sg14-bounces_at_[hidden]> on behalf of Michael Wong via SG14
>>> <sg14_at_[hidden]> <sg14_at_[hidden]>
>>> *Sent:* Wednesday, September 17, 2025 10:21:36 AM
>>> *To:* Low Latency:Game Dev/Financial/Trading/Simulation/Embedded
>>> Devices <sg14_at_[hidden]> <sg14_at_[hidden]>;
>>> sg19_at_[hidden] <sg19_at_[hidden]> <sg19_at_[hidden]>
>>> *Cc:* Michael Wong <fraggamuffin_at_[hidden]> <fraggamuffin_at_[hidden]>
>>> *Subject:* [isocpp-sg14] SG14 and SG19 meeting @ CPPCON Wednesday Sept
>>> 17, 13:00 MDT
>>>
>>> Hi all, this month's SG14/SG19 meeting will be held at CPPCON Denver at
>>> 13:30 MDT.
>>>
>>>
>>>
>>>
>>>
>>> Michael Wong is inviting you to a scheduled Zoom meeting.
>>>
>>> Topic: SG14/SG19 monthly
>>>
>>> Join from PC, Mac, Linux, iOS or Android:
>>> https://iso.zoom.us/j/93151864365?pwd=aDhOcDNWd2NWdTJuT1loeXpKbTcydz09
>>> Password: 789626
>>>
>>> Or iPhone one-tap :
>>> US: +12532158782,,93151864365# or +13017158592,,93151864365#
>>> Or Telephone:
>>> Dial(for higher quality, dial a number based on your current
>>> location):
>>> US: +1 253 215 8782 or +1 301 715 8592 or +1 312 626 6799 or +1
>>> 346 248 7799 or +1 408 638 0968 or +1 646 876 9923 or +1 669 900 6833
>>> or 877 853 5247 (Toll Free)
>>> Meeting ID: 931 5186 4365
>>> Password: 789626
>>> International numbers available: https://iso.zoom.us/u/abRrVivZoD
>>>
>>> Or Skype for Business (Lync):
>>> https://iso.zoom.us/skype/93151864365
>>>
>>> Agenda:
>>>
>>> 1. Opening and introduction
>>>
>>> ISO Code of Conduct
>>> <
>>>
>>> https://isotc.iso.org/livelink/livelink?func=ll&objId=20882226&objAction=Open&nexturl=%2Flivelink%2Flivelink%3Ffunc%3Dll%26objId%3D20158641%26objAction%3Dbrowse%26viewType%3D1
>>> *>*
>>>
>>> ISO patent policy.
>>>
>>> https://isotc.iso.org/livelink/livelink/fetch/2000/2122/3770791/Common_Policy.htm?nodeid=6344764&vernum=-2
>>>
>>> IEC Code of Conduct:
>>>
>>> https://www.iec.ch/basecamp/iec-code-conduct-technical-work
>>>
>>> WG21 Code of Conduct:
>>>
>>>
>>> https://isocpp.org/std/standing-documents/sd-4-wg21-practices-and-procedures
>>>
>>> 1.1 Roll call of participants
>>>
>>> 1.2 Adopt agenda
>>>
>>> 1.3 Approve minutes from the previous meeting, and approve publishing
>>> previously approved minutes to ISOCPP.org
>>>
>>> 1.4 Action items from previous meetings
>>>
>>> 2. Main issues (125 min)
>>>
>>> 2.1 General logistics
>>>
>>>
>>>
>>> 2.2 Paper reviews
>>>
>>> 1. P1780R10 Statistics for SG19 ML
>>> Are there any other papers of interest? Thanks.
>>> 2. Other topics
>>>
>>> _______________________________________________
>>> SG14 mailing listSG14_at_[hidden]://lists.isocpp.org/mailman/listinfo.cgi/sg14
>>>
>>> _______________________________________________
>>> SG14 mailing list
>>> SG14_at_[hidden]
>>> https://lists.isocpp.org/mailman/listinfo.cgi/sg14
>>>
>> _______________________________________________
>> SG14 mailing list
>> SG14_at_[hidden]
>> https://lists.isocpp.org/mailman/listinfo.cgi/sg14
>>
> _______________________________________________
> SG14 mailing list
> SG14_at_[hidden]
> https://lists.isocpp.org/mailman/listinfo.cgi/sg14
>

Received on 2025-10-07 20:25:15