[aklug] A developer's take on system resource consumption

Guys:

Caught this article on /.:

   http://blog.internetnews.com/skerner/2011/06/mozilla-launches-memshrink-eff.html

in which Mozilla is now going to dedicate a team to targetting memory
utilization. One of the statements in the article struck me as pertinent:

   There are plenty of people out there at have 1 GB or less of RAM. As
   I'm writing this post now, Firefox 4.0.1 is consuming 633 MB on my
   machine and that's not abnormal. Simply put, it's just waaay too much
   RAM for a web browser. Then again, if you've got 8 GB of DDR3 RAM,
   you likely aren't all that worried about using 633 MB for Firefox. So
   maybe it's just a question of focus. Is Mozilla still interested in
   the mainstream consumer?

Obviously, I expect a great deal of debate over what a "mainstream consumer"
really constitutes in terms of memory consumption. Even so, it highlights a
problem some of us have been bitching about since the '80s: with all of the
magnitude increases in system resources over the years why is it that all of
our actual gains are only incremental?

Ultimately, our incremental gains are entirely the fault of the developer.
While some of our ambitions can only legitimately be done on a bigger/faster
machine, the fact that we're targeting more "modern" hardware in general
allows us to take shortcuts and less aggresively optimize that we'd
otherwise do.

Ironically, while I have both harped and groused on this subject for as long
as I've been using computers, I've been just as guilty as anyone else.
Having recently made my primary personal computer a 700MHz toughbook with
256MB of RAM that point has been driven home. My missteps as a developer has
consequences. Having a development system with such meager resources
amplifies any frivolous use of CPU, RAM, or I/O, and it's something I now
more actively try to counter since I have to live with it. Even minor
missteps can cumulatively hurt.

All of which brings forth a good question: how much better a computing
situation would we all be in if *developers* were forced to be at the tail
end of the upgrade cycle, rather than the forefront? If we, as a matter of
routine, had lower-spec'd hardware than the majority of the users we have to
support?

I know my developmental process has become slower, but more deliberative. I
actively try to anticipate system resource utilization beforehand. I test
and benchmark multiple approaches to solving the same problem to see which
one taxes the system the least. Sometimes it pays to be poor, eh? ;-) (Side
note: I'm not really poor... cheap, yes, but not poor. And I've definitely
been blessed...)

So, let's start a movement to cripple developers with old machines! Who's
with me? ... Well, I suppose I should have seen that one coming... ;-)

         --Arthur Corliss
           Live Free or Die
---------
To unsubscribe, send email to <aklug-request@aklug.org>
with 'unsubscribe' in the message body.
Received on Sat Jun 11 11:06:10 2011