[ale] HPC replies

[Russell L. Carter]

Hi Brian,

On 1/8/25 9:35 AM, Brian MacLeod wrote:

[... elided]

> And I'll also share one of the most maddening things I dealt with in
> my time in HPC was the focus on speed of results, not efficiency of 
> computing (though code efficiency was occasionally a focus).  There
> are such groups, but certainly in research the focus is on speed so
> results can be published before anyone else. That invariably means
> there's clock time available for other projects during lulls if they
> could be made ready, but we were always focused on faster hardware,
> not efficient use of it.

Yeah.  The efficiency part[1] is what always interested me.
But nowadays, for example, on Hacker News people like to
"Here's how we implemented this 30 yo numerical algorithm in our current
language l'amour de jour and no I have never heard of *LAPACK nor do we
care because it's fast enough" and I just sigh. And then I realize that
the article is either a marketing pitch for the company to customers
or it's just yet another "see how competent I am please hire me" plea.
The important part is not the efficiency.

I find myself writing a lot of tool programs for tools (I'm remodeling
my house) and after getting burnt YET AGAIN by FAANG C++ standard
"features" I said screw it and I'm writing everything now in "slow"
SBCL Common-Lisp.  Grandpa is happy with his 40 yo language.  (C++
lambdas are *good*, though).

All the best,
Russell

[1] Back in 1992 or so I implemented the NAS Parallel Benchmarks
tridiagonal and pentadiagonal solvers in templated C++, very first
to do so, I think.  Performance using the ancient g++ was quite
good.  When I left a couple of years later David Bailey was
quite eager to get a copy.  In those days, it was here it is,
have fun.


> bnm
> 
> On Mon, Dec 30, 2024 at 8:44 PM Russell L. Carter via Ale
> <ale at ale.org <mailto:ale at ale.org>> wrote:
> 
> I'm loving it!  Old school: see below for the reply
> 
> On 12/30/24 7:59 PM, Leam Hall via Ale wrote:
>> On 12/30/24 17:50, Russell L. Carter via Ale wrote:
>>> Yes, exactly so.  This is what I meant by economically
>>> inefficient. You can be a specialist and employed at say a
>>> National Lab. of some sort located *not* in an expensive housing
>>> city (ie, not NASA Ames) and enjoy the work immensely.  I know I
>>> did:  if you check out the original published NAS Parallel
>>> Benchmarks you will find my name
> right
>>> there, and that job was the grandest adventure of my life.
>>> 
>>> But!!  I don't know anything about SLURM because my own
> household/small
>>> biz clusters just can't justify the overhead for *all the other
> stuff*
>>> that is required to make such a system viable with a low
>>> headcount support crew of ahem uno ein un yep *1*.
>>> 
>>> I mean it's all cool stuff and if you believe it's for you (it
> was for
>>> me at the time) go for it but have no illusions that anybody on
>>> say Hacker News will understand such a thing.  (That could be a
>>> good
> reason
>>> to do it anyway, yep I get it)
>>> 
>>> Basically the whole cloud ecosystem which is heating up the
> planet so
>>> successfully is predicated on *waste*.  Maybe somebody has
> calculated
>>> how much of the cloud is simply tests.  You need tests, to be
> sure, but
>>> real work is always jobs of days, weeks, months.  What the cloud
> don't
>>> care about is efficiency, in the HPC context.
>>> 
>>> Good luck and all the best, Russell
>>> 
>>> Did I just top post... again?  I mean &*^@(*&^#$ Firefox for
>>> getting rid of the emacs editing ability.
>> 
>> I'll just remove all else and respond.  :D
>> 
>> You, Brian, Vernard, and Jim, became my target audience, although
> Jim
>> was the only person I knew doing HPC. Vernard, Russell, it is
> good to
>> meet you!
>> 
>> Much of my career has been on things most technologists don't care 
>> about: OS vulnerability scanning, supply chain security, and not
> dumping
>> 27,315 third-party packages on a node because the STDLIB method
> isn't
>> the newest or coolest thing. Having done cloud, and even
> supported an
>> enterprise-specific cloud backend, your opinion seems to match
> mine. The
>> cloud is flexible, but not overly efficient. Many cloud start-ups
> don't
>> seem to care what OS is underneath, they just want their app to
> run and
>> keep running. Which is a fine thing! If it pays the bills for
> them and
>> adds value to the world, then more power to them.
>> 
>> I have a chance at an HPC job. During the interview I was honest
> about
>> what I didn't know, even admitting that I had to do a web search
>> on "Slurm". I like to think I can learn, given a task list, but
>> with
> HPC "I
>> don't know what I don't know". Hence the original question. I
> chased the
>> "AppDev" career path for a while, and while I like playing with
> code,
>> front end development just isn't my joy. I like Linux, I've built
>> a career supporting it, and HPC seems like a cool path.
>> 
> 
> Ok, all good.
> 
>> One of the things I'm looking forward to is not being the smartest 
>> person in the "room". Which is funny, as I'm not particularly
>> smart. I've just been doing this for a long time and enjoy
>> learning more
> about
>> whatever I'm doing. If I can get into an HPC shop I'll have a lot
>> to learn and smart people to learn from.
> 
> No.
> 
> I worked at NASA Ames "NAS", and at Sandia Livermore, and traveled
> to about 30 research labs across the country, including later when I
> was doing SBIRS. You are as good as your technical chops are.  If
> you are in the room, or on the call, or responding constructively in
> a distributed email (or whatever the kidz do these days) you are
> already there.  You learn, yes, everybody learns.  Those smart
> people are learning from *you*, my friend.
> 
> The problem with software shops that I have had experience with is
> that tenure is so understandably tenuous that people don't speak
> frankly; they don't prepare you for the "life of the mind" that we
> had way back then.  I can tell you that it is very likely that as a
> software person with (proven?) experience in the cloud that the
> interviewing person is probably *very* interested in that
> perspective, as will be the people you support.  You can't possibly
> understand any of the codes though and no sane person at the
> institution would ever suspect it.
> 
> But!  As you may have surmised from the replies those technical 
> people mainly want their codes to run efficiently.  Days -> hours. 
> hours -> minutes.  That sort of thing.  I find that kind of puzzle 
> intoxicatingly seductive; if only it could have paid for our life.
> 
> 
> All the best, Russell
> 
>> So I'm hoping and praying for the job; we'll see what happens.
>> 
>> Thanks!
>> 
>> Leam
>> 
>> 
> 
> _______________________________________________ Ale mailing list 
> Ale at ale.org <mailto:Ale at ale.org> https://mail.ale.org/mailman/
> listinfo/ale <https://mail.ale.org/ mailman/listinfo/ale> See JOBS,
> ANNOUNCE and SCHOOLS lists at http://mail.ale.org/mailman/listinfo
> <http://mail.ale.org/mailman/ listinfo>
>