THE QUANTUM
CONUNDRUM W H AT
AN ENTANGLED
MESS WE WEAVE...
Forget material science, feature density, and all that old-
fashioned stuff involving classical physics and chemistry.
Could quantum computing make Moore’s Law completely
redundant, but in a good way?
According to quantum computing advocates, it’s possible
to imagine a single quantum computer capable of not just
matching the combined number-crunching might of all
existing classical computers, but also of executing as many
calculations at once as is practically useful. In other words,
all the computing power any of us could ever need in a single
machine. Forget Bill Gates’ apocryphal claim that nobody
needs more than 640K, quantum computing aficionados
think we won’t need more than one PC between us.
If that sounds like a fringe theory, try this comment from
Android OS architect Andy Rubin. “If you have computing
as powerful as this could be, you might only need one
[computer].” So, how might that work? Conventional
computing operates in the binary realm of zeros and ones,
otherwise known as bits in computing vernacular. The
basic component in a classical computer, the transistor,
is therefore either off or on. There is no alternative.
With quantum computing? Not so much. Thanks to
a quantum phenomenon known as superposition, which
applies at the atomic and sub-atomic scale, it’s possible for
a quantum computing bit to be not just both on and off at the
same time, but also a huge array of hybrid superpositions
somewhere in between on and off. This is a qubit and it is the
basic building block of quantum computing.
The other important weirdness of quantum computing,
and arguably the factor that makes things exciting, is
entanglement. It’s a notion even tenured physics professors
struggle with. But the basics involve the idea that the
quantum properties of two or more particles can be
inextricably linked regardless of any distance between
them. Change the spin of one particle, for example, and
others instantly react.
The trick to achieving powerful quantum computing,
therefore, is to ‘entangle’ multiple qubits. Quantum-
mechanically link or entangle two qubits and you can
perform two calculations simultaneously. Link three qubits
and the math of quantum entanglement means you can do
two to the power of three—or eight—calculations. Link four
and you can perform 16 calculations simultaneously. Follow
that logic and by the time you have 300 entangled qubits, you
can perform more calculations in parallel than there are
atoms in the known universe. Handy.
The catch? There is doubt that a quantum computer
on that scale is possible due to the problem of quantum
decoherence. Some argue the quantum state of the machine
can’t be maintained long enough to do anything useful, but
others question whether quantum computers can be applied
to general computation at all, and instead will always be
limited to irrelevant, esoteric math of little practical benefit.
Unfortunately, we forgot to polish up on our applied physics
PhDs here on Maximum PC, so all we can say is time will tell.On the other hand, this powerful knob that we have been using is
not the only knob we have at our disposal.”
It’s debatable how many more nodes are available using the
conventional 2D planar approach. Certainly, TSMC expects to
have 3nm and then 2nm conventional planar semiconductor
nodes. Last year, TSMC said that the development of its 2nm
is “progressing nicely through the pipeline.” 1nm may also be
possible using a new approach that replaces FinFET transistors
with so-called GAA or Gate All Around alternatives.
But all that begs the question of what those additional ‘knobs’
Wong was referring to could be? Planar semiconductors are
fast approaching their physical limits and can’t possibly keep on
delivering nearly long enough to fulfill TSMC’s 30-year vision for
Moore’s Law. So, what other types of technologies is the industry
likely to turn to?
In practice, it looks likely to be a mix of the exotic and the
prosaic. If Moore’s Law is all about density, taking a slightly
different system-level view of that metric allows new chip-
packaging techniques to keep the show on the road. Fusing
multiple chiplets of both logic and memory together using
silicon interposers allows for much greater system density while
allowing for better production yields.
Trying to manufacture a single large chip is much harder and
more expensive than lots of little chips. But fusing them together
in a way that reduces latency and allows for performance akin to
a monolithic chip is a huge technical challenge that the likes of
TSMC and Intel have only recently begun.
If that’s an extension of the existing planar model, the other
obvious option is to go up instead of sideways. According to
Godfrey Cheng, Head of Global Marketing at TSMC, “one possible
path forward is the use of transistors made of two-dimensional
materials instead of silicon as the channel. We are raiding the
periodic table.
“By potentially using these new materials, one possible
future of great density improvements is to allow the stacking of
multiple layers of transistors in something we call Monolithic 3D
Integrated Circuits. You could add a CPU on top of a GPU on top
of an AI Edge engine with layers of memory in between. Moore’s
Law is not dead. There are many different paths to continue to
increase density.”Are quantum computers, such as the D-Wave Advantage, useful?Moore’s law
34 MAXIMU MPCMAR 2022
© D-WAVE