Boy,
that escalated
Techno
Talk quickly
Techno Talk
From
humble
8-bit microprocessors and Arduino boards to chiplets,
Max
the Magnificent
XPUs, stacked silicon and edge-AI accelerators, a ramble through the
astonishing (and slightly bewildering) evolution of modern computing.
I’m starting to feel like an old
fool. Things are moving so fast these
days that I don’t know whether I’m
coming or… I’m sorry, what were we
talking about?
Hello Arduino, my old friend
As I’ve mentioned on many an occasion (and as I’ll doubtless mention
again), I hail from the world of early 8-bit
processors—those far-off days when a
1MHz system clock seemed scary fast.
Just to set the scene, typical highend microprocessors in the mid-1970s,
such as the Intel 8080 and Motorola
6800 (both announced in 1974), plus
the MOS Technology 6502 (introduced
in 1975), all had 8-bit data paths and
clocked at about 1-2MHz. All required
external RAM, external ROM, external
clock circuitry, and external input/
output (I/O) chips. If you needed an
analog-to-digital converter (ADC), that
would be external as well.
It’s worth remembering that the introductory prices for the 8080 and 6800
were $300+ (~£150) and $200+ (~£100),
respectively (that’s $1800/£1450+ and
$1200/£950+ in today’s money). After
that, you still needed to splash the cash
for external memory devices and support logic! So it isn’t surprising that
when the 6502 appeared on the scene at
just $25/£10 (about $150/£120 today), it
sent shockwaves through the industry.
Compare this to today’s Arduino
That’s not an Arduino...
this is an Arduino!
Practical Electronics | May | 2026
microcontroller development boards.
I typically use Arduino Uno or Nano
boards for my hobby projects. I recently
purchased three Arduino Nanos for only
$5 each, which still blows me away.
The ATmega328P microcontroller chip
on one of these boards boasts a 16MHz
clock, 32kiB of flash memory, 2kiB of
SRAM and a 10-bit, 6-input ADC.
In 1975, 16MHz would have sounded like lab-grade or military-grade
hardware; 32kiB of ROM would have
required an entire board full of chips
(and would have cost some serious
money); and having a 10-bit ADC integrated into the process would have
felt like something from NASA.
When I was a student in the late
1970s, even the humblest single-board
computer (SBC) was far outside my
means. I’m almost embarrassed to
say that, while rooting through some
boxes of bits at home the other day on
the hunt for a piezoelectric buzzer (for
use in my next Weird and Wonderful
column), I must have run across at
least five Arduino Uno boards lurking in various states of benign neglect.
In 1975, just one of these boards
would have had engineers flocking
around in wide-eyed disbelief!
Do you remember when the boy in The
Sixth Sense says, “I see dead people”?
I mistakenly remembered this as “I see
dead people everywhere”. The reason I
mention this is that I’m starting to see
Arduinos everywhere. As I pen these
words, I’m sitting at the desk in my
office at work, and I see at least three
Arduino Uno boards peeking out at
me from under piles of bits and pieces.
Furthermore, there’s a small cardboard
box sitting on a shelf in front of me that
has been niggling me for ages because I
couldn’t recall what it contained, but I
was too lazy to get up and see. I finally
forced myself to look inside, only to
discover a “Make Your Own Arduino”
kit (https://docs.arduino.cc/hardware/
make-your-uno-kit/).
Where on Earth did that come from?
And what are the odds that it would surface while I was writing this column?
Speaking of DIY Arduinos, have
you seen the giant version created
by a guy who goes by the name of
Uncle Stem? This really is a beauty to
behold. He talks us through the build
Max the Magnificent
in a YouTube video (https://youtu.be/
RZkv9WKkpGg).
This isn’t just for show. It actually
works, although paradoxically and/or
ironically, it’s powered by an Arduino
Nano. You can upload programs to it
and run them. The oversized onboard
light-emitting diodes (LEDs) light up,
and you can plug wires and external
components into the header pins.
Of course, Uncle Stem has also created
a set of large components, such as resistors and LEDs, that are to scale. Seeing
this leads me to think of two things.
First, Uncle Stem has too much time
on his hands. And second, I want one!
Feeling tensor?
It’s often said that you don’t realise
what you don’t know until you try to
explain something you think you do
know to someone else. I was reminded of this recently while listening to
a friend who was attempting to explain the differences between central
processing units (CPUs), digital signal
processors (DSPs), graphical processing units (GPUs), and neural processing
units (NPUs).
Of course, it’s easy to be a critic (spotting other people’s shortcomings is one
of the things I do best), but it struck me
that my friend might have been better
served if he’d begun by laying a little
groundwork by introducing the different
types of data that need to be processed.
For example, in computer science,
a scalar is a single value; a vector is
a one-dimensional array of scalars; a
matrix is a two-dimensional grid of
scalars (which can also be viewed as
a table of scalars, or as a collection of
row or column vectors); and a tensor
extends this to three or more dimensions (think of it as a stack of matrices,
or a higher-dimensional stack).
These terms are used somewhat differently in other disciplines such as
mathematics and physics (we had to
overload their meanings, didn’t we?).
You might find it helps to visualise
this as a progression as follows:
• 0D: Scalar → one number
• 1D: Vector → a line of numbers
• 2D: Matrix → a table of numbers
• 3D (and higher): Tensor → a stack of
tables (and possibly stacks of stacks).
Beyond that, it gets a bit complicated.
continued on page 75...
67
