From: "Dmitry A. Kazakov" <mailbox@dmitry-kazakov.de>
Subject: Re: Generators/coroutines in future Ada?
Date: Tue, 18 Jul 2017 09:38:11 +0200
Date: 2017-07-18T09:38:11+02:00 [thread overview]
Message-ID: <okkdt3$1o2o$1@gioia.aioe.org> (raw)
In-Reply-To: okjioe$g91$1@franka.jacob-sparre.dk
On 18/07/2017 01:54, Randy Brukardt wrote:
> No, I think there is very little chance of it becoming possible. When I was
> a young buck at the University of Wisconsin, there was a lot of optimism and
> research into automatic parallelization of programs (particularly Fortran
> programs). I recall reading a number of papers in this area for classes.
>
> Fast forward (gulp!) about 38 years, and most of the optimism is gone, and
> there has not been a lot of progress in the area of automatic
> parallelization. Certainly, compilers can recognize certain patterns and
> parallelize them (generally in cases where all of the "tasklets" run the
> same code), but general-purpose parallelization has defied solution (at
> least for conventional languages -- there might be more luck with a
> purpose-built language like Parasail -- but this is an Ada forum and such
> languages are irrelevant for Ada).
>
> It should be clear that fine-grained parallelism (the kind that a compiler
> could conceivably do) is likely to be a failure, as scheduling takes a
> significant amount of overhead, on top of which a lot of conventional
> sequential code doesn't have much parallelism. (Remember that things like
> exceptions, assignment semantics, and the like add sequential requirements
> that probably aren't necessary to solve the actual problem.)
>
> I would never want to say never, but I think it is far more likely that
> programming will cease to be a human activity at all (perhaps because
> everything is replaced by neural nets) than automatic parallelization of
> "classic" languages become possible.
Huh, when I was young everybody believed AI round the corner. However
already then, long long before that actually, it was well known that
neuronal networks were a dead end.
The algorithm (Perceptron) was invented in 50's and already then shown
that it can separate only linear classes, that is by a flat surface in a
n-D space. Imagine how does it work with two concentric circles. Even a
hen can do that, and probably an insect too.
Fast forward, people forgot 50's, 80's and incidentally the very skill
of researching (reading?) scientific publications ... then Perceptron
was reborn!
> If you believe, like I do, that automatic parallelization is likely to be
> impossible in the reasonable term, the next best thing is to give the
> compiler hints. Moreover, nothing about "parallel" blocks and loops actually
> requires them to be run in parallel; it just suggests the possibility. It
> also includes checking that the "parallel" code is actually possible to
> execute in parallel, so you are actually giving an optimizer more help by
> ensuring that the code can really be run that way.
I fully agree.
> My goal for these constructs is that they are essentially semantically
> identical to the non-parallel versions, just with checking that parallel
> execution is possible. (I have no idea how close or far from that goal we
> will end up, as these constructs change a lot from meeting to meeting.)
IMO, it is pure time-wasting while so many problems stay unresolved.
Modern multi-core architectures of shared memory is clearly a dead end.
When new architectures emerge we will see what language support or even
paradigms will be needed to handle these, not before,
--
Regards,
Dmitry A. Kazakov
http://www.dmitry-kazakov.de
next prev parent reply other threads:[~2017-07-18 7:38 UTC|newest]
Thread overview: 40+ messages / expand[flat|nested] mbox.gz Atom feed top
2017-07-09 20:19 Generators/coroutines in future Ada? Victor Porton
2017-07-09 21:28 ` Dennis Lee Bieber
2017-07-09 23:56 ` Victor Porton
2017-07-10 7:33 ` Dmitry A. Kazakov
2017-07-10 13:38 ` Victor Porton
2017-07-10 17:01 ` Paul Rubin
2017-07-10 21:24 ` Victor Porton
2017-07-12 17:09 ` Paul Rubin
2017-07-12 17:39 ` Victor Porton
2017-07-13 6:35 ` Paul Rubin
2017-07-11 9:42 ` J-P. Rosen
2017-07-11 12:52 ` Victor Porton
2017-07-11 13:01 ` Victor Porton
2017-07-11 13:26 ` J-P. Rosen
2017-07-11 16:04 ` Dennis Lee Bieber
2017-07-11 16:59 ` Dmitry A. Kazakov
2017-07-11 19:52 ` Pascal Obry
2017-07-11 20:18 ` Dmitry A. Kazakov
2017-07-11 18:36 ` Victor Porton
2017-07-11 19:22 ` J-P. Rosen
2017-07-11 20:25 ` Dmitry A. Kazakov
2017-07-11 23:19 ` Victor Porton
2017-07-12 4:54 ` J-P. Rosen
2017-07-12 13:07 ` Victor Porton
2017-07-12 13:38 ` Dmitry A. Kazakov
2017-07-12 5:35 ` Randy Brukardt
2017-07-12 7:27 ` Dmitry A. Kazakov
2017-07-12 22:47 ` Shark8
2017-07-16 13:11 ` Robert Eachus
2017-07-17 23:54 ` Randy Brukardt
2017-07-18 7:38 ` Dmitry A. Kazakov [this message]
2017-07-12 7:19 ` Dmitry A. Kazakov
2017-07-12 6:35 ` G.B.
2017-07-12 7:34 ` Dmitry A. Kazakov
2017-07-12 20:49 ` G.B.
2017-07-13 8:18 ` Dmitry A. Kazakov
2017-07-12 17:34 ` Paul Rubin
2017-07-11 19:27 ` Simon Wright
2017-07-12 5:42 ` darkestkhan
2017-07-12 8:57 ` Maciej Sobczak
replies disabled
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox