[J3] [EXTERNAL] Re: Comments on 23-146

[Dick Hendrickson]

This is a problem for me. I accidently deleted something a minute ago, so I
can't directly reply to it; but, I think I remember a (long, acrimonious?)
thread from a few years ago about avoiding comments about people rather
than their ideas.  So "This argument is of no intellectual merit whatsoever
and I’m embarrassed for you that you made it" just seems wrong.  As Jim
Matheny said "it's the ideas I'm attacking, not the idiot who came up with
them."

Dick Hendrickson

On Mon, Feb 27, 2023 at 3:08 PM Jeff Hammond via J3 <
j3 at mailman.j3-fortran.org> wrote:

> Yes, I will show the use cases and the measurements.  >95% of the hundreds
> of GPU applications NVIDIA works with is using some form of asynchronous
> task parallelism.  It’s hard to write a CUDA program that doesn’t have it.
> It’s a critical feature of OpenACC, which is used primarily by Fortran
> applications.  It’s also present in HIP and DPC++, which are important to
> other members of WG5.  That Fortran alone among the HPC languages lacks a
> form of asynchronous task parallelism inhibits its use on modern
> supercomputers, or at least forces the adoption of directives, which many
> WG5 members have indicated they do not like.
>
>
>
> Jeff
>
>
>
> *From: *Clune, Thomas L. (GSFC-6101) <thomas.l.clune at nasa.gov>
> *Date: *Monday, 27February 2023 at 10:08 PM
> *To: *General J3 interest list <j3 at mailman.j3-fortran.org>, Van Snyder <
> van.snyder at sbcglobal.net>
> *Cc: *Jeff Hammond <jehammond at nvidia.com>, j3 <j3 at j3-fortran.org>
> *Subject: *Re: [EXTERNAL] Re: [J3] Comments on 23-146
>
> *External email: Use caution opening links or attachments*
>
>
>
> I will make a broader statement:  whether/how such features are
> implemented in other languages generally has limited relevance for
> Fortran.   We, of course, want to understand the strengths and weaknesses
> of the various approaches used in other languages, but those should stand
> on their own merits.
>
>
>
> Strengths and weaknesses are best examined in the light of *concrete* use
> cases.   If COBOL’s approach is the best for our prioritized use cases,
> that is just fine.   (And makes for an amusing story at the end of the
> day.)    I look forward to a constructive discussion about the
> strengths/weaknesses of varying approaches to accommodating independent
> tasks in an asynchronous context.  (Trying hard, but failing, to use
> completely neutral terminology.)
>
>
>
>
>
>
>
>
>
>
>
>
>
> *From: *J3 <j3-bounces at mailman.j3-fortran.org> on behalf of j3 <
> j3 at mailman.j3-fortran.org>
> *Reply-To: *j3 <j3 at mailman.j3-fortran.org>
> *Date: *Monday, February 27, 2023 at 2:47 PM
> *To: *Van Snyder <van.snyder at sbcglobal.net>
> *Cc: *Jeff Hammond <jehammond at nvidia.com>, j3 <j3 at mailman.j3-fortran.org>,
> j3 <j3 at j3-fortran.org>
> *Subject: *[EXTERNAL] Re: [J3] Comments on 23-146
>
>
>
>
>
>
> On 27. Feb 2023, at 21.39, Van Snyder <van.snyder at sbcglobal.net> wrote:
>
> *External email: Use caution opening links or attachments*
>
>
>
> On Mon, 2023-02-27 at 15:21 +0000, Jeff Hammond wrote:
>
> I would prefer that Fortran join this modern language set rather than fade
> into history like Ada.
>
>
>
> WG9 has more members than WG5.
>
>
>
> This argument is of no intellectual merit whatsoever and I’m embarrassed
> for you that you made it.
>
>
>
> Jeff
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mailman.j3-fortran.org/pipermail/j3/attachments/20230227/a3271223/attachment-0001.htm>