Chiuw 2021

Users
1

Hi all,

I noticed that IPDPS 2021 rumblings are happening on other lists. I saw
that it is just down the road from you guys this year. Not that this is
relevant because it will be virtual anyway.

What are the deadlines for CHIUW 2021 please? Is it still affiliated with
IPDPS 2021 or is that irrelevant in these days of virtual workshops?

Stay safe - Damian

Pacific Engineering Systems International, 277-279 Broadway, Glebe NSW 2037
Ph:+61-2-8571-0847 … Fx:+61-2-9692-9623 | unsolicited email not wanted here
Views & opinions here are mine and not those of any past or present employer

2

Hi Damian –

I don’t know when the deadlines will be - I think they are still TBD.

There was this exchange on gitter earlier:

Thomas:

on the topic of 2021, is there any news regarding CHIUW 2021?

Brad:

Good question. Nothing definitive yet. We expect to have it in a completely virtual format this year, hopefully sometime in the spring, and decided not to affiliate it with a specific conference because that seemed to add more overhead than benefit. We have a new group manager starting up today, and I expect this to be a topic of discussion in the next few weeks.

Hope that helps,

-michael

3

Hi Damian,

We are planning to hold CHIUW as an independent event this year. Planning for
the event is well underway and we expect to send the call for papers soon
(hopefully sometime this week). Please stay tuned!

Engin

4

In case you missed it, the announcement:

Call for Papers and Talks for CHIUW 2021

Full CFP:

Chapel: CHIUW 2021: Call for Papers and Talks

5

Hi Engin,

Chapel: CHIUW 2021: 8th Annual Chapel Implementers and Users Workshop

We have already written in the context of non-working partial reductions.
If they work, our paper will be a little simpler and have two paragraphs
less discussion. It is not done in Word.

I have tried to rewrite the paper in the 5-8 page 2 column format and it
does not work. The text only just fits 9 pages. The Chapel code examples
and tables in particular do not work in a narrow column format and those
two, and references, do not fall within that limit.

How critical is that 2 column ACM format and page limit? The ACM's own
flagship journal, Transactions on Mathematical Software does not even use
that format because it does not work to explain numerical algorithms!

Thanks - Damian

Pacific Engineering Systems International, 277-279 Broadway, Glebe NSW 2037
Ph:+61-2-8571-0847 .. Fx:+61-2-9692-9623 | unsolicited email not wanted here
Views & opinions here are mine and not those of any past or present employer

6

Hi Damian,

The format and page limit is really helpful for reviewers and organizers to
gauge the review workload. However, I have been in the position you are
describing many times, and I understand that it is unfortunate.

For tables:

  • It sounds like you are narrowing your tables to fit into 2 columns. Can you
    make them span two columns? In latex this is straightforward (change from
    table to table*). In Word, I don't know myself. I have seen some
    discussions online about how to make that happen, like:

    microsoft word - How to make a table span two columns? - Super User.

    Let me know if something like that works for you. I would like to believe that
    it is possible, but I don't know how straightforward that is.

For code blocks:

  • Is it possible to reduce some of the code by omitting some details that are
    unrelated to what you are trying to show? This maybe in the form of creating
    Chapel-like pseudocode for some cases.

  • Can you put some of the code on gist.github.com and link to them in footnotes
    or something?

  • If applicable, you can also put all your code in a github repo and provide
    only minimal snippets for it and link to your repo.

Let me know if any of this helps reduce the paper length a bit.

Engin

7

Hi Engin,

The format and page limit is really helpful for reviewers and organizers to
gauge the review workload.

I wrote this in Troff with one of the standard macro package (MM from the
Documenters Workbench. It is very close to the ACM submission template

https://www.acm.org/binaries/content/assets/publications/taps/acm_submission_template.docx

I have never before in my life seen the ACM format that the CHIUW 2021
guidelines mentioned although it looks like what is used for SIGGRAPH.

It can convert it to Latex but I have no time to do this until Easter.

For tables:

It sounds like you are narrowing your tables to fit into 2 columns.
Can you make them span two columns?

Not easily. Troff does not easily switch between 2-column and 1-column
modes. As I noted, if I discard code blocks, tables and references, drop
the point size from 12-point to 11-point, and quickly add the

.2C \" two column macro

just after the abstract, it produces a 9 page long PDF. This PDF converts
nicely to Word automatically.

The table of performance comparison data is deliberately minimalistic.

For code blocks:

Is it possible to reduce some of the code by omitting some details that
are  unrelated to what you are trying to show? This maybe in the form of
creating Chapel-like pseudocode for some cases.

I have long ago done this. Only the few lines relevent to the discussion
are in the document.

Can you put some of the code on gist.github.com and link to them in
footnotes or something?

I am happy to put the full code there. I provided the full code to a
2020 GSoC project last year but it was too late in the project so was
not able to be leveraged.

Let me know if any of this helps reduce the paper length a bit.

The report/paper is meant as a guide to people trying to do similar things
so throwing bits out destroys the usefulness of the document. I could have
chosen a simpler algorithm for this exercise and had less to say but then
it would have risked being trivialized as just "too simple".

Brad just gave me some code to better fake a partial reduction. That will
hopefully save 2 paragraphs from the discussion but I won't get a chance
to play with that for a few weeks.

I will submit it as is. Feel free to reject it.

Regards - Damian

Pacific Engineering Systems International, 277-279 Broadway, Glebe NSW 2037
Ph:+61-2-8571-0847 .. Fx:+61-2-9692-9623 | unsolicited email not wanted here
Views & opinions here are mine and not those of any past or present employer

8

I see submissions are not due until April 9th which is 6 weeks away. I
will incorporate Brad's fudged partial reduction code next month and see
how much it simplifies our explanation and hence reduces the text.

Will this actually get published in some ACM journalk or conference?

I notice that only slides(+video) were produced. last year. No papers

Regards - Damian

Pacific Engineering Systems International, 277-279 Broadway, Glebe NSW 2037
Ph:+61-2-8571-0847 .. Fx:+61-2-9692-9623 | unsolicited email not wanted here
Views & opinions here are mine and not those of any past or present employer

9

Let us know how it goes.

No, the work you have submitted will not be in any proceedings. We will post
them on the website. In the previous years, the papers appeared in the
affiliated conference's proceedings, and we only put slides and talk videos
online.

Also, if it helps, page limits do not include bibliography. I am realizing that
we haven't put that information in the call for papers. And I'll fix that.

Engin

10

Hi Damian / all —

Just for a big more context (because I think it's an interesting tension):

The main reasons we've affiliated CHIUW with larger conferences in past
years include:

a) it takes care of a lot of the logitics of organizing an event

b) it gives us the chance to pick up random people who happen to be at the
conference

c) it tends to make it easy to publish the papers in a proceedings

However, in practice, we also found:

d) that it was expensive given the value-add attendees got for it over a
homegrown event

e) that we didn't get as many random people walking in as we'd hoped

f) that we got a lot more talk submissions than paper submissions anyway,
perhaps due to the nature of our community and users

Then when Covid hit last year, we realized:

g) that we could easily get a larger audience virtually than in person

h) that being affiliated with a conference added more headaches than
benefits in a virtual format

So once it seemed certain that this year would need to be virtual again,
abandoning the conference for the time being seemed like the obvious
choice. We considered whether we could try for a special issue of a
journal or somesuch, but given the numbers of papers in recent years, it
seemed unlikely we'd meet their minimum numbers. So instead, I think of
the value proposition for submitting papers this year as being "Get expert
feedback on your paper before submitting it somewhere that it'll likely do
you and the Chapel community more good [in terms of reaching outside the
invested community]."

The interesting question will be what we do in post-Covid years, where my
personal hope is that the technology will be at the point that doing a
hybrid virtual + in-person conference will feel natural, and a way to
re-attach to a larger conference without losing the ability to involve
remote attendees, or possibly even speakers.

-Brad

11

What page size is required for these papers?

When I try and look at the template, it seems to say A4. But I am
wondering if it is confused with my printer settings.

Regards - Damian

Pacific Engineering Systems International, 277-279 Broadway, Glebe NSW 2037
Ph:+61-2-8571-0847 .. Fx:+61-2-9692-9623 | unsolicited email not wanted here
Views & opinions here are mine and not those of any past or present employer

12

Hi Damian,

US Letter or A4 should be fine.

Engin

13

Hi Engin,

US Letter or A4 should be fine.

Great. I have formatted mine for A4.

When you submit to EasyChair, if you suddenly discover mistakes, can you
upload an update easily? Sorry, I have only used Easychair once, and that
was 10 years ago.

Regards - Damian

Pacific Engineering Systems International, 277-279 Broadway, Glebe NSW 2037
Ph:+61-2-8571-0847 .. Fx:+61-2-9692-9623 | unsolicited email not wanted here
Views & opinions here are mine and not those of any past or present employer

14

Hi Damian,

Yes, you can. And I just confirmed that our configuration allows that. You can
update your submission until the deadline.

Please let me know if you found that not to be the case.

Engin

15

I want to use some words from the Chapel web page, specifically

"a programming language designed for productive parallel computing at
scale"

I deliberately do not want to use quotes because with quotes, it seems
like I am quoting the Chapel sales pitch (which I am) but not in a
salesman sense! I want to use the precise description.

Does anybody know the ethics of this please, or would Cray/HPE object?

I also want to say

Chapel provides data-parallel constructs X, Y, and Z and idioms A,
B and C

Chapel provides task-parallel constructs X, Y, and Z

When I took the words out of my brain and corrected the English, I found
that in the data-parallel case I had reconstructed a sentence from the
documentation page word for word. Maybe I had read it too often and the
words got implanted. Mind you, the sentence was largely a list of nouns
and adjectives

forall statement
forall expression
several idioms
whole array assignment
function and operator assignment
reductions and scans

Again, the same question.

And in one sense, when describing a list, you want a standard definition
so you do not get it wrong.

In describing the human body we have a head, neck, torso, abdomen, groin,
hips, legs and feet. It is pretty hard to say it any other way!

Thank - Damian

Pacific Engineering Systems International, 277-279 Broadway, Glebe NSW 2037
Ph:+61-2-8571-0847 .. Fx:+61-2-9692-9623 | unsolicited email not wanted here
Views & opinions here are mine and not those of any past or present employer

16

Hi Damian —

I want to use some words from the Chapel web page, specifically

"a programming language designed for productive parallel computing at
scale"

I deliberately do not want to use quotes because with quotes, it seems
like I am quoting the Chapel sales pitch (which I am) but not in a
salesman sense! I want to use the precise description.

Does anybody know the ethics of this please, or would Cray/HPE object?

I'm quite certain Cray/HPE wouldn't, and I/we wouldn't either. I think
what you're saying here is sufficiently short and innocuous (and accurate)
that it isn't the kind of thing that would trigger a plagiarism concern.

I also want to say

Chapel provides data-parallel constructs X, Y, and Z and idioms A,
B and C

Chapel provides task-parallel constructs X, Y, and Z

When I took the words out of my brain and corrected the English, I found
that in the data-parallel case I had reconstructed a sentence from the
documentation page word for word. Maybe I had read it too often and the
words got implanted. Mind you, the sentence was largely a list of nouns
and adjectives

forall statement
forall expression
several idioms
whole array assignment
function and operator assignment
reductions and scans

Again, the same question.

I don't see a problem here either.

Thanks for your sensitivity to these issues,
-Brad