Quote:
Originally Posted by Peter Ward
... was programmed in Assembler  . A dark art indeed  |
Hi Peter,
Early in my career I swallowed the Red Pill and have been much deeper than that.
For example, one of the projects I was the manager on was the design and
implementation of a 32-bit microprocessor.
Starting with a blank sheet of paper, a first-cut of the architecture was a simulation
down at the register level, which also included the first iteration of the instruction
set, pipeline and caches.
Concurrently the code generation back-end of a C-compiler and assembler
was written which produced assembler and machine code for our first iteration
of the instruction set.
A series of target applications were compiled and the generated code run on
the simulator. What is referred to as profiling.
The designer has one foot in how efficient the instruction set is on the software side
and another foot in how it is most efficiently implemented on the hardware
implementation side.
Trade-offs abound.
What was of particular interest were aspects such as pipeline stalls. Whenever
a branch instruction is executed it can dramatically affect pipeline performance.
There are several methods with how to best deal with the problem, some of
which are compile-time schemes.
Profiling the execution of the applications thus provided clues as to
where improvements could be made. The instruction set was changed, the
pipeline altered and the cache sizes and architecture tuned, at the same
time being mindful that this then has to be implemented to fit on to a single
semiconductor die for the right price.
This particular device was then designed at the gate level.
Earlier than that in my career I had designed digital devices down at the
semiconductor transistor level.
That's a long way down the rabbit hole.
However, in turn, guys I worked with had been much deeper than that.
For example, before I had worked for him, my previous boss, an Aussie,
had earlier in his own career conceived, researched and implemented
semiconductor devices down at the quantum level.
Based on a prediction of his boss, he discovered the "quantum well".
They filed the patent for the first quantum well laser.
Useful if you want to build a CD or DVD player.
My boss went on to play an important part in the invention of what was then the
world's fastest transistor. The High Electron Mobility Transistor or HEMT
is fabricated in gallium arsenide (GaAs). They happily run at over 600GHz.
If you have a satellite dish on your roof there is one in there. They are deployed
in some mobile phone base stations. They are at the heart of radio telescopes and
there are some on-board spacecraft that have left the solar system.
Later and again before I worked for him, my boss was the manager of a semiconductor
research lab. One of the people he hired came to him as a postdoc. My boss
outlined to the postdoc what the team had been up to and sketched on
his whiteboard (he loved to use his whiteboard) some semiconductor structures.
As a result of that conversation, the postdoc made what he described in his
Nobel Prize acceptance speech as "a casual, almost trivial observation, which,
however, turned out to have big impact." The three lab workers in the team
discovered the fractional quantum Hall effect which won them the 1998 Nobel
Prize in Physics.
One afternoon I walked into my bosses office and he was staring at his whiteboard.
He said to me, "You know, I spent a great deal of my career thinking about things
in the first dimension".
He was alluding of course to his earlier involvement in the conception of
semiconductor devices using quantum confinement techniques. When he said
it I thought to myself that I can get my head around the third dimension and
the second dimension but I smiled and asked him "How on earth could you spend
all day just thinking about the first dimension?". He just smiled back.
He had clearly been down the rabbit hole much deeper than me.
What he had been involved in seemed like a dark art to us and it was
so mind-boggling, mathematical and abstract that all we could do was be in awe of
the people that burrow down that far.
But sometimes when we were giving our boss progress reports on what we were
doing much further up the hole at the hardware/software level, he would say,
"You know, you guys are really smart". Coming from someone who had
been so deep down the hole, that made us feel pretty good!
Peter, I guess if I were to get to sit where you do, I might look around the
cockpit at the instrumentation and at some level of abstraction have an
appreciation of, say, how the voting mechanism might be implemented
in hardware/software on the multiple redundant flight computers (research into
fault tolerant computer systems was my first paid job). My previous boss
might look around and seeing the light emitting from the LEDs have some
deep appreciation of how they work at the quantum level.
If an LED were to stop operating in-flight, you of course have an appreciation that
it is an electronic device that uses a quantum effect, but you don't need to know
exactly how it does that at the quantum level. You intimately understand what
the LED was meant to indicate and can instantly make a decision as to whether
you can obtain the same information but my some other annunciator.
As the rest of us look around and imagine how various sub-systems might
work at the levels of abstraction of which we have knowledge, we rely totally
on you to get us rolling and into the air and most critically, we rely on your
skill, training, intelligence and calmness to get us back down if something might
go wrong.
In other words, we'd all be saying you are a pretty smart guy!
It is interesting when you look around at the modern world and see
how complex it is. It requires many levels of specialization and
we rely on each other in complex ways.
Like most of us, I am unashamedly passionate about Australia and really
believe if we all keep pulling on the rope in the same direction we can
achieve even greater things.
I am also unashamedly passionate about the role science and technology has
played in getting us to where we are today. When the culture of our forebears
embraced the scientific and industrial revolutions all those years ago,
they helped set us up pretty nicely.
It is a trajectory that we would be silly to abandon.
You can divide the world into two camps of guys. If I had been lucky enough
to be present at Kill Devil Hills in 1903 watching the Wright Bros. take that
first flight, I am the type of guy that would be among those bellowing out "Go! Go! Go!".
But unfortunately there would always be some guy up the back laughing at them
and calling out "Crash. Crash.. Crash." When they only flew 37m, the same
guy would have no doubt made it clear to the rest of us that it was "useless".
Later when Kingsford Smith crossed the Pacific, the same guy would have declared with authority,
"That's good enough. We need do no more. Any more is a waste of time and money".
Like the man said, I guess we choose to do these things not because they are easy,
but because they are hard.
Best regards
Gary Kopff
Mt Kuring-Gai NSW 2080
Phone: 02 9457 9049
Into the Rabbit Hole - a personal perspective
Would you like to know what Tony Abbott said about mandates in 2007?
I think you are confusing my use of the word "platform" with mandate? 
