George A. Madrid
						6.033 - Comp. Sys. Eng.
						8 February 1994

	The  main thrust  of Lyon's paper  is  that an incommensurate  scaling
	argument leads to the conclusion  that parallelism can be exploited to
	reduce  battery drain.  But  chapter 1  (reading 1) suggested  several
	different kinds  of system  thinking, complexity-producing situations,
	and complexity-attacking approaches.   Identify  and discuss at  least
	one other  such  generality from Chapter   1 that emerges  as you read
	Lyon's paper.

One interesting  point  which Lyon  raises  is a   problem of  propagation  of
effects.   He says  that  we have  some  of  these absolutely  amazingly  fast
processors, like the DEC Alpha, which run at  clock speeds of 200MHz, but that
in a system that uses this chip, most of  the cost comes  from the cooling and
power supply systems.  So, we see the speed of the chip  affecting the rest of
the system.  Also, in the interest of speed, the Alpha  designers have made no
efforts to gate the  system clock from unused  portions of the processor; this
causes even greater power  usage.  This fact, along with  the usual  quirks of
fast chip design, means that the  Alpha will most likely not  be usable with a
slower clock speed.  This effect is  a result of propagation  of effects.  One
single design motivation, speed, has increased the power usage requires of the
Alpha, and probably reduced  significantly the chances that  the Alpha will be
usable in a low-power, portable environment.

In  Saltzer, Chapter 1,  Saltzer presents abstraction  as a method for dealing
with  complexity.  Lyons  suggests  doing away  with  some  abstraction in the
interests of power-savings.  He  proposes tossing away the digital abstraction
in favor of an analog approach.  This wins  in terms of power-usage, but loses
in terms of  precision.    So, we see the   trade-off  between  generality  and
efficiency which Saltzer discussed.  

Lyon's paper is  full of trade-offs.  Since  Lyon's goal is very specific,  he
often trades generality  for  specificity or  even simplicity for  complexity.
Since he has a  very specific goal, that  of low power   usage, he is  free to
pursue that goal, even at the cost of added system complexity.  This is a good
argument  that  complexity has  its place, and  that  Saltzer's ideas are only
guidelines.  In the end, one  must keep in mind  the goals  of the system  and
work towards those goals.