APL BUG 21 Jan 2010: Robert Bernecky on "Functionality is Free"
Curtis A. Jones
ACM) will meet on Thursday the 21st of January to hear Robert Bernecky
speak on ways to speed up APL, J and related languages.
Thursday, 21 January 2010
6 p.m. - Bring takeout supper and network.
7 p.m. - Robert Bernecky speaks.
at San Jose State University,
College of Engineering Room ENG 339
(E San Fernando St. & S 7th St. San Jose, CA 95192)
http://www.sjsu.edu/about_sjsu/visiting/campus_maps/#maincampus
http://www.sjsu.edu/parking/
http://www.sjdowntownparking.com/evening_parking.html
Speaker: Robert Bernecky
Title: Functionality is Free
Abstract
Functional array languages, such as APL, J, SAC, and SISAL, are
computer languages with their roots in linear algebra. Being abstract,
functional, and array-based, they are excellent tools of thought. As
well as being amenable to formal analysis and algebraic
simplification, they exhibit parallelism at levels including primitive
verbs, expressions, defined verbs, compositions of verbs, and verbs
derived from adverbs and conjunctions.
Until recently, these benefits have been sullied by poor performance,
compared to C and Fortran, their scalar-oriented, imperative cousins,
with array-based code often executing several orders of magnitude
slower.
We present recent research results in functional array language
optimization that show a closing of that performance gap, with array
language code matching or beating scalar code on single core
benchmarks, as well as offering significantly better performance on
multi-core systems and GPU-based systems, all achieved without
application code alterations.
Biography
Robert Bernecky has designed and developed APL systems since 1971.
While at I.P. Sharp Associates Limited, he was one of the people
responsible for the design and development of SHARP APL, a system that
set the standard for performance of large-scale APL systems. He has
authored papers on language design, algorithm design, and interpreter
performance.
Bernecky is the CEO of Snake Island Research Inc, a consulting and
research firm headquartered in Toronto.
Bernecky developed APEX -- the APL Parallel Executor -- a high-
performance, retargetable APL compiler for serial and parallel
computers.
Bernecky holds a BA in philosophy from SUNY at Buffalo, an MSc in
Computer Science from the University of Toronto, and is trying to
finish his PhD before his PhD finishes him. He is a member of ACM and
IEEE. Bernecky lives on Ward's Island in Toronto with no cats anymore.
MichaelHughes
> The APL Bay Area Users' Group (The Northern California SIGAPL of the
> ACM) will meet on Thursday the 21st of January to hear Robert Bernecky
> speak on ways to speed up APL, J and related languages.
>
> Thursday, 21 January 2010
> 6 p.m. - Bring takeout supper and network.
> 7 p.m. - Robert Bernecky speaks.
>
> at San Jose State University,
> College of Engineering Room ENG 339
>
> Speaker: Robert Bernecky
>
> Title: Functionality is Free
>
Can I put in a request that a transcript be made available for those
of us who would love to attend but can't.
There is a regular meeting of APL people here in London, UK every last
Friday of the month and it would be good to be able to have the
material like this to discuss as well.
Regards
Mike
Curtis A. Jones
> ...
> Can I put in a request that a transcript be made available for those
> of us who would love to attend but can't.
Michael,
If all goes well Chuck Kennedy will get an MP3 audio recording from
the talk that can be made available. Curtis
Olivier Lefevre
This (I suspect largely laziness-driven) trend toward MPS3 and videos
is absolutely deplorable. In most case they bring nothing to the table
that a written transcript wouldn't and they exclude scores of people:
those with hearing deficiencies, those not proficient in the tongue
in question, those who can't make heads or tails of the speaker's
accent etc etc. Please do consider putting up a transcript.
Regards,
-- O.L.
Kerry Liles
proper transcript (I guess this is what you mean by laziness-driven) and
since I speak the same language as Bob (more or less: he *was* an American;
but I digress) I would much appreciate hearing it.
Producing a good transcript may be very difficult - especially considering
the difficulty of properly hearing comments made by others (questions from
the floor, asides that lead to clarifications, etc). The mp3 would be great;
a transcript (in addition) would be priceless. I don't think the use of mp3
or video is deplorable: they are merely a partial solution to the problem of
preserving useful events for historical record.
"Olivier Lefevre" <lefe...@yahoo.com> wrote in message
news:hjeset$etl$02$1...@news.t-online.com...
Stan Clitherdawes
> Capturing an informal talk as a recording is a lot easier than making a
> proper transcript (I guess this is what you mean by laziness-driven) [... deleted ...]
A reservation I have about audio/video capture is that using it is a
real-time experience - with paper/slides/text/etc. I can scan and decide
how much time I want to put into reviewing the material - and it's easy for
me to annotate.
I really think the right way to approach this is for the presenter to
create whatever materials they think the subject matter deserves. I don't
like the idea of third parties summarising.
If someone says "this is a one-off talk and isn't worth preserving" that's
their prerogative.
MichaelHughes
Many many thanks - I would have replied earlier but I'm just back in
the UK from a week in South Africa. I'm sorry I provoked the
controversy, mine was just a simple request for ANYTHING covering the
talk, as I couldn't be there in person, so anything was better than
nothing. Knowing the cost of these things and the time people have to
put into producing them, I was really asking this as a favour rather
than as a commercial product to receive complaints.
Again if you can put something up for download/email I would be
grateful.
Michael
Although I speak English too I can see Oliviers
Olivier Lefevre
one cannot *demand* anything. I just don't want people
to delude themselves that audio and video recordings are
just fine and that the need for transcripts is gone:
there's still nothing like the written word.
"Producing a good transcript may be very difficult -
especially considering the difficulty of properly
hearing comments made by others (questions from
the floor, asides that lead to clarifications, etc)."
If even someone who was present can't make sense of
the MP3 afterward, how easy do you suppose it will be
for listeners who didn't attend the original event?
If that is the case, then the MP3 will be useless.
But I think you're just wheeling out the old excuse
whereby if one can't do a perfect job, then one should
not do it at all, which is of course nonsense: just
like according to you an MP3 is better than nothing,
a partial transcript is better than no transcript.
Regards,
-- O.L.
Kerry Liles
I would like to have both the .mp3 and a transcript.
If a transcript is not available, I would still like to have the .mp3
phil chastney
>
> But I think you're just wheeling out the old excuse
> whereby if one can't do a perfect job, then one should
> not do it at all, which is of course nonsense: just
> like according to you an MP3 is better than nothing,
> a partial transcript is better than no transcript.
an attitude which was once summed up (if somewhat ambiguously) as
"if a job's worth doing, it's worth doing badly"
/phil
Björn Helgason, j-programming
A perfectionist never delivers anything
Curtis A. Jones
> of us who would love to attend but can't.
>
> There is a regular meeting of APL people here in London, UK every last
> Friday of the month and it would be good to be able to have the
> material like this to discuss as well.
>
> Regards
> Mike
Joey Tuttle agreed to post Bob Bernecky's notes for "Functionality is
Free".
They're at
http://keiapl.org/docs/RBEhandout.pdf
Joey added "That directory (docs) is an open index with several things
of potential interest to APL folk."
Curtis
Curtis A. Jones
<kerry.removethisandoneperiod.li...@gmail.com> wrote:
> ..
> I would like to have both the .mp3 and a transcript.
> If a transcript is not available, I would still like to have the .mp3
Kerry,
I hope Bob Bernecky's notes that Joey Tuttle has posted at
http://keiapl.org/docs/RBEhandout.pdf
are some help.
The MP3 recorder quit early in the talk, so we don't have an MP3. I'm
sorry.
Since this is part of Bob's Ph.D. research there will certainly be
real papers. I didn't ask Bob if he's submitted a paper for APL2010,
but I'd keep an eye on
http://www.apl2010.de/
Curtis
Curtis A. Jones
...
> If someone says "this is a one-off talk and isn't worth preserving" that's
> their prerogative.
Stan,
Thank you. Jones's own bias: Recording talks can make them less worth
preserving. A camera and microphone can have an inhibiting effect on
a speaker. It may not be obvious. An informal, face-to-face
conversation has a different quality than something that might be
investigated 30 years later.
In the case of this talk, Bob Bernecky happened to be visiting the
area and agreed to talk to the APL BUG. He was talking about a work
in progress, maybe practicing for his defense, maybe getting an idea
of interest in elements of his work, and maybe even getting advice
from the people who came with real experience in writing compilers and
interpreters. There was also a significant element of IPSA reunion.
(I deliberately used "IPSA", a form that might not be familiar to some
readers here. A talk may be for a specific audience.)
Speakers do "create whatever materials they think the subject matter
deserves." They're often papers for publication. I'm sure Bob is
working on a thesis.
Curtis
Kerry Liles
news:3aab1403-ed79-47b5...@p13g2000pre.googlegroups.com...
Thank you Curtis! I will indeed have a look...
PS: I understood IPSA (being ex-IPSA made that rather easy of course). I
look forward to anything Bob does (as always).
Curtis A. Jones
Paul Jackson tell about a set of classes he's developed to
provide APL functionality for the .NET programmer.
Thursday, 1 April 2010
6 p.m. - Bring takeout supper and network.
7 p.m. - Paul Jackson speaks.
at San Jose State University,
College of Engineering Room ENG 339
(E San Fernando St. & S 7th St.
San Jose, CA 95192)
http://www.sjsu.edu/about_sjsu/visiting/campus_maps/#maincampus
http://www.sjsu.edu/parking/
http://www.sjdowntownparking.com/evening_parking.html
Paul Jackson writes:
The free APL I've been using is over twenty years old. In
addition to being a DOS program, it has severe limitations
on the size of variables and reading modern files would mean
mapping UTF-8 and Unicode. Years of using other development
environments has demonstrated that they have moved well
beyond what was available in the early years of APL's
success.
I've worked with VB.Net since it arrived, and I felt it had
the tools necessary to develop a compiled APL. This is not
an effort to compete with those who've learned the internals
of the .Net CLR and built a traditional interpreter.
Instead, I've produced a set of .Net classes which provide
APL functionality for the .Net programmer. It will be
provided in a way which makes writing your own functions and
operators relatively easy.
Briefly, one must declare variables as APL, but nothing
more.
Dim myA As APL
myA = _Index(10)(_a.Plus, _Of(100))
myA = _Of(2, 5)(_a.Reshape, myA)
myA = _Of("ABCD")
myA = myA(_c.IndexOf, _Of("AX"))
If you step through this one line at a time, you will find
the APL value contains exactly what you would expect as an
APL programmer.
Biography:
I learned APL in 1969 while working at Trinity University in
San Antonio, Texas. I taught it there and at Dallas County
Community College District for the first twelve years of my
career. I then joined I.P. Sharp Associates, where I
developed several of their shared variable processors and
the character left argument for thorn. I eventually lead
the APL development group, before leaving in 1993.
For the last decade, I've been leading the development group
at Dialog. Dialog is a company which generalized and
formalized search engine expressions, much like APL did for
logical expressions. I retired last September, and am
enjoying the ability to work on what interests me.
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
Other Events
22 Mar 2010 Andrea Montanari, "DMSIG - Large Matrices
beyond Singular Value Decomposition
(Mountain View, CA)
http://www.sfbayacm.org/?cat=41
15 - 16 Apr 2010 Finnish Forest Seminar, "somewhere in
Southern Finland"
http://vectoreditor.blogspot.com/2009/10/finnish-forest-seminar-15-16-apr-2010.html
26 - 27 Apr 2010 APL2000 User Conference, Bethesda, MD
http://apl2000.com/2010UserConference.php
5 - 10 Jun 2010 ACM SIGPLAN 2010 Conference, Toronto
http://cs.stanford.edu/pldi10/
13 - 16 Sep 2010 APL2010 in Berlin
http://www.apl2010.de/
New APL Books
Eugene McDonnell, "At Play With J [Edn 2]"
http://www.lulu.com/content/paperback-book/at-play-with-j-%5Bedn-2%5D/7405758
Bernard Legrand, "Mastering Dyalog APL..."
http://www.amazon.co.uk/s/ref=nb_sb_ss_i_0_16?url=search-alias%3Daps&field-keywords=mastering+dyalog&sprefix=mastering+Dyalog
Curtis A. Jones
Northern California SIGAPL of the ACM) on Thursday, 1 April. Details
are in the previous post.
Curtis Jones
cjl
http://aprogramminglanguage.wordpress.com/2010/03/31/sharp-people/