Presentation at lenexa: We are SG14!

[Michael Wong]

Sorry I am now at Las Vegas to give a talk at an IBM conference after Lenexa, so I have not had much time to compose a report. I am in conference now, and trying to type this while listening to some presentation that I am terribly interested in.

In short, it went GREAT! We (meaning I had Billy Baker's help) presented on Saturday morning to EWG, and the result was more then we could hope for. I kind of hoped that presenting to EWG( and if all the right people were there including Bjarne, Herb, Sony guy) would have the effect of leading to a new SG whereas presenting to LEWG would not have similar result. However, that having been said, the more appropriate group for your papers should be LEWG (except reflection). There was great support for us to become SG14 and it was declared so in the formal vote that afternoon. There was a guy from Sony's compiler team who was there, and he echoed everything I said.

In fact, in post presentation lunch discussion with Herb and a bunch of people interested in games, we like to offer this to this group:
1. Form an official SG, SG14 named as Games dev (& Low Latency). I am interim chair until someone else wants the job, as I am not really part of your industry. My only qualification is I like playing games when I used to have time to play games (10 years ago), but no longer.
2. Have CPPCON2015 be partly focused with a Games track on one day, and also an official SG14 meeting on the second day so that we can have all the games people present to discuss the issues and concerns that trouble them.
3. Sony has offered to also host a similar meeting at GDC2016 (March 14-18? which unfortunately is during the Canadian March break when I usually have my family vacation but let's see how it will work out), where we will have our second SG14 meeting which will allow us to look at concrete paper. Quite a few WG21 committee members have committed to attend. I guess I finally get to go to GDC which I have always wanted to go. May be I can give a talk at GDC2016 so I can justify my trip there.
4. The idea is to have the WG21 committee members go to GDC and CPPCON SG14 meetings to help evaluate the papers. These papers would be presented at the Feb and June 2016 C++ Std meetings. In this way, the games dev people would not be required to be present.

I hope this short summary convey some of the thoughts. As soon as I finished my talk here, I will give more detail. Or Billy who was there can also add in more details. Thanks.

[Sean Middleditch]

Wow, this is really awesome.

Thank you Michael for championing this and all the work and time
you've put into it!

> --
> You received this message because you are subscribed to the Google Groups
> "unofficial-real-time-cxx" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to unofficial-real-ti...@googlegroups.com.
> To post to this group, send email to
> unofficial-r...@googlegroups.com.
> To view this discussion on the web visit
> https://groups.google.com/d/msgid/unofficial-real-time-cxx/be592401-d2dc-4b7f-bd4d-2fa804da3017%40googlegroups.com.
> For more options, visit https://groups.google.com/d/optout.



--
Sean Middleditch
http://seanmiddleditch.com

[Nicolas Guillemot]

Heroic!

To view this discussion on the web visit https://groups.google.com/d/msgid/unofficial-real-time-cxx/CALQmNFgxtjntp6WQVb7GrwoN9nFMC4R%2Bww2YiUL4SwUH8pdDOw%40mail.gmail.com.

[Scott Wardle]

Nice work guys. 

Scott Wardle

To view this discussion on the web visit https://groups.google.com/d/msgid/unofficial-real-time-cxx/CAFcDmHNQsXZF1duO%3Dyz6hx4ee%2BaJh9aDUh-%3D0aUdpdH48x5ZMw%40mail.gmail.com.

[Klaim - Joël Lamotte]

That's a very good news!

As both employed in a non-gaming-but-embedded-soft C++ shop (Aldebaran/Softbank) and independent game (and digital narration tools) developer I hope this will help a lot.

By the way, there was mention in the paper you published that there might be a merge with the potential Embedded SG,

so is SG14 considered to be that SG too from the commitee point of view?

[Michael Wong]

On Wednesday, May 13, 2015 at 1:10:17 PM UTC-4, Joel Lamotte wrote:

On Tue, May 12, 2015 at 12:54 AM, Michael Wong <fragga...@gmail.com> wrote:

Sorry I am now at Las Vegas to give a talk at an IBM conference after Lenexa, so I have not had much time to compose a report. I am in conference now, and trying to type this while listening to some presentation that I am terribly interested in.

In short, it went GREAT! We (meaning I had Billy Baker's help) presented on Saturday morning to EWG, and the result was more then we could hope for. I kind of hoped that presenting to EWG( and if all the right people were there including Bjarne, Herb, Sony guy) would have the effect of leading to a new SG whereas presenting to LEWG would not have similar result. However, that having been said, the more appropriate group for your papers should be LEWG (except reflection). There was great support for us to become SG14 and it was declared so in the formal vote that afternoon. There was a guy from Sony's compiler team who was there, and he echoed everything I said.

In fact, in post presentation lunch discussion with Herb and a bunch of people interested in games, we like to offer this to this group:
1. Form an official SG, SG14 named as Games dev (& Low Latency). I am interim chair until someone else wants the job, as I am not really part of your industry. My only qualification is I like playing games when I used to have time to play games (10 years ago), but no longer.
2. Have CPPCON2015 be partly focused with a Games track on one day, and also an official SG14 meeting on the second day so that we can have all the games people present to discuss the issues and concerns that trouble them.
3. Sony has offered to also host a similar meeting at GDC2016 (March 14-18? which unfortunately is during the Canadian March break when I usually have my family vacation but let's see how it will work out), where we will have our second SG14 meeting which will allow us to look at concrete paper. Quite a few WG21 committee members have committed to attend. I guess I finally get to go to GDC which I have always wanted to go. May be I can give a talk at GDC2016 so I can justify my trip there.
4. The idea is to have the WG21 committee members go to GDC and CPPCON SG14 meetings to help evaluate the papers. These papers would be presented at the Feb and June 2016 C++ Std meetings. In this way, the games dev people would not be required to be present.

I hope this short summary convey some of the thoughts. As soon as I finished my talk here, I will give more detail. Or Billy who was there can also add in more details. Thanks.

That's a very good news!

As both employed in a non-gaming-but-embedded-soft C++ shop (Aldebaran/Softbank) and independent game (and digital narration tools) developer I hope this will help a lot.

Please help the effort if you can. Thanks.

By the way, there was mention in the paper you published that there might be a merge with the potential Embedded SG,

so is SG14 considered to be that SG too from the commitee point of view?

 

No merging. Herb was of the opinion that these 2 are different groups. Also we got somewhat "fast-tracked" as the Embedded group is not officially an SG yet. 

--
You received this message because you are subscribed to the Google Groups "unofficial-real-time-cxx" group.

To unsubscribe from this group and stop receiving emails from it, send an email to unofficial-real-time-cxx+unsub...@googlegroups.com.

[Michael Wong]

Ok, I finished my talk, and before I get to my next destination in barcelona. Here are a few thanks.
Thanks to Sean on the perfect  slides.  I incorporated Paul Pedriana's quote on Gaming status into the talk as well while adding my own embellishments. They were hot items. Bjarne, Gaby, and Paul McKenney (my mentor who is a Linux kernel maintainer) has asked me for them. Paul said:
"I now have a renewed appreciation for Linux kernel being coded in C rather
than C++".  Then again, games are probably much more object-oriented
than are kernels.  ;-)



Thank everyone who contributed to the original and the revised paper.

I plan to update the paper with the results from Lenexa and submit a new paper for the post-meeting mailing so people can be updated on the status.
 The new paper is already changed in google docs and is called N4526. Please have a look:
https://docs.google.com/document/d/1sdllzgrl4rLMn6jvV-AJB1Vp4p8l142iaj0RAtgEyAk/edit#

I will have to submit by next Friday.

Thanks.

[Guy Davidson]

This is excellent news.  I will assist in any way I can.

Many thanks and congratulations,

Guy

[Billy Baker]

One area that received enthusiastic support from high performance computing was the idea of adding intrusive containers.

[Marcelo Zimbres]

"One area that received enthusiastic support from high performance computing
was the idea of adding intrusive containers. "

Is it a consensus that we cannot achieve the same performance with the current
allocator model, so that we do not have to add the intrusive
containers? For a couple
of months I have discussed real-time allocators in the C++-embedded mailing
list, which I believe would a possible solution or a way towards it. But since
it is a (small) breaking change in the allocator interface (for node-based
containers) it would be hard to get this accepted.

The thread is:

http://www.open-std.org/pipermail/embedded/2014-December/000335.html

continuation

http://www.open-std.org/pipermail/embedded/2015-January/000342.html

Regards,
Marcelo


2015-05-15 9:55 GMT-03:00 Billy Baker
<billy.baker%flightsa...@gtempaccount.com>:

> --
> You received this message because you are subscribed to the Google Groups
> "unofficial-real-time-cxx" group.
> To unsubscribe from this group and stop receiving emails from it, send an

> email to unofficial-real-ti...@googlegroups.com.

> To post to this group, send email to
> unofficial-r...@googlegroups.com.
> To view this discussion on the web visit

> https://groups.google.com/d/msgid/unofficial-real-time-cxx/e3ea5bbb-4076-45fc-8f6c-e81d94ba8e96%40googlegroups.com.

[Sean Middleditch]

Custom allocators do not make it possible to insert pre-existing
elements into a container with zero additional allocations or guesses
at an appropriate pre-allocated pool. If you have already allocated
your objects, you're going to have to follow an indirection to access
them anyway when iterating the container. Having to also then
allocate, resize, and manage another container that is nothing more
than a holder of pointers in addition to the objects themselves is
unnecessary work.

The custom allocator approach you outline has many strengths, and I'm
all for an improved allocator interface that more directly supports
node-based vs contiguous allocations. It's a restriction I've seen
placed in custom STL-replacements in game engines for some time; e.g.
an allocator must have a single-element allocate and optionally an
n-element allocate, and contiguous containers require their allocators
to support the n-element interface.

The rebinding thing is also something we usually see killed-off. It's
silly to have to specify an object type to an allocator that is just
going to be replaced with some internal implementation detail anyway.
Better to have allocators passed to containers as template types and
let the container decide exactly how to instantiate it.

Between intrusive lists or your proposal, though, the best choice is
going to be dependent on the specific problem the programmer is trying
to solve.

> To view this discussion on the web visit https://groups.google.com/d/msgid/unofficial-real-time-cxx/CAPGXEnGVVCZ10FkHYDWEs8r%3DWODgSAAsm4B8ZLjgAM7V5gCz8g%40mail.gmail.com.

> For more options, visit https://groups.google.com/d/optout.

[Ion Gaztañaga]

El 15/05/2015 a las 14:55, Billy Baker escribió:
> One area that received enthusiastic support from high performance
> computing was the idea of adding intrusive containers.

[Sorry, the default reply action is to reply to the sender instead of
the mailingl list, so this went only for Billy. Reposting here]

Great. I'm pretty interested in this idea given that I wrote most of
Boost.Intrusive. That library might not be the ideal starting point for
standard intrusive containers but at least it could contain some useful
ideas.

Best,

Ion