Agenda for sync meeting 3/19/2021

[tlively]

We are returning to a monthly meeting schedule, so the next meeting will be Friday, March 19 at 9:00AM - 10:00AM PDT/ 5:00PM - 6:00PM CET (Note the different European time due to the start of daylight savings in the US). Please respond with agenda items you would like to discuss.

If this meeting doesn't already appear on your calendar, or you are a new attendee, please fill out this form to attend.

[tlively]

Bumping this in case anyone has items they want to discuss on Friday. If not, we should probably cancel the recurring meeting going forward, although of course we can pull together one-off meetings to discuss any issues that do come up in the future. Unless @ngzhian wants to take over this recurring meeting slot for discussion of relaxed SIMD?

[abrown]

I wouldn't mind using this meeting for relaxed/fast SIMD since it's already routine and many of the same people are involved?

[arunetm]

Its an good time to seek the interest level among the simd sub-group to collaborate on flexible vector proposal as well. Maybe we can add an item to discuss this. @penzn fyi.

If there a sizable interest among attendees, we can use the monthly sync to discuss Relaxed SIMD or Flexible vectors topics as they come up. As both proposals are in the early stages, sharing slots might be reasonable until a need arises to separate them.

[lars-t-hansen]

I agree with @abrown and @arunetm, we should just segue into the relaxed and flexible simd proposals. (We have the most inviting proposal names, hands down.) In particular, I wouldn't mind talking a bit around the scope expectations for relaxed simd, I just posted a bug about that.

[dtig]

Agenda items sgtm for this meeting, but perhaps we can add a quick item to discuss this meeting, and possibly formalizing a SIMD subgroup. We never officially did this - so maybe we should?

I agree keeping the same time slot is beneficial, but I'd like to propose some logistical changes and give the current champions of the new proposals some flexibility to make changes if needed. Once merged, this repo will stop being a good place for co-ordination. We could move the meeting notes etc. to the meetings repo and collect agenda items from both flexible, and relaxed SIMD proposals as an option, and talk about formalizing this meeting in a light weight way so it is easily discoverable for folks not already attending the meeting but are contributing to the other proposals that are in flight

[penzn]

I was just writing that at least trying a joint SIMD++ meeting would be a reasonable thing to do, when @dtig beat me to it 😄

I think we can try a joint meeting or even form a subgroup. Also, it is true that we would need to use meetings repo instead of either of the SIMD repositories. If we end up with too much agenda at any point or people working on one of the proposal feel like they are not really interested in the other one, we can split them apart.

[tlively]

Sounds good. The agenda items I've seen so far are:

1. Meta-discussion about making the subgroup official and how best to coordinate meetings and notes going forward
2. Discussion of scope and goals for relaxed SIMD.

[penzn]

I can give an informal update on flexible vectors as well.

[ngzhian]

Unless @ngzhian wants to take over this recurring meeting slot for discussion of relaxed SIMD?

Happy to do so! We can discuss more about the specifics this Friday (e.g. where to post agenda issues, meeting notes, etc).

[abrown]

If you all have time, please take a look at a PR to document problematic instructions in the Emscripten repository: https://github.com/emscripten-core/emscripten/pull/13706. (And I should add, can we add this to the agenda?).

[tlively]

Thanks, everyone! Here are the notes. We will meet again in 4 weeks, so watch for a new agenda issue from @ngzhian.