Building a Quantum Computer From Off-the-Shelf Parts

https://spectrum.ieee.org/tech-talk...le-qubits-quantum-computer-news-silicon-wafer

"A new technique for fabricating quantum bits in silicon carbide wafers could provide a scalable platform for future quantum computers. The quantum bits, to the surprise of the researchers, can even be fabricated from a commercial chip built for conventional computing."

Hope it works, we need a big breakthru in computing as Moore's Laws' demise looms...

So, Even if we do get quantum computers to be a thing, how long till the Unreal engine runs on em?

Seems like Quantum computers "work" completely different than the computers we're used to running.

Accursed said:

So, Even if we do get quantum computers to be a thing, how long till the Unreal engine runs on em?

Seems like Quantum computers "work" completely different than the computers we're used to running.

Click to expand...

probably all OS will have to be written in C again... or oh no there is no quantum C... well beyond what i understand... do we have to start ALL over??? im guessing there will be some kind of quantum -> 64bit stack to help out... things could get real annoying soon.

i am a lowly python dev. i wonder what the future holds. hopefully employment in some form :S

Hitachi built a few from room temperature ferrite shirt buttons back in the 50's...
Was a macroscopic almost quantum thing, a whole lot of quantum in parallel.
Superposition, yes. Error correction, yes (by majority vote). Entanglement, no.

Some qbits these days work the same way. Certainly most detectors do.
Quantum doesn't mean you can only do absurd quantum logic with them.
Folks are now building perfectly normal 8bit CPUs that run at 80GHz, etc.
Efficient too, if you don't count bill for running the refrigerator down to zero.

Nothing really useful, but proofs of concept.

purple_monster said:

probably all OS will have to be written in C again... or oh no there is no quantum C... well beyond what i understand... do we have to start ALL over??? im guessing there will be some kind of quantum -> 64bit stack to help out... things could get real annoying soon.

i am a lowly python dev. i wonder what the future holds. hopefully employment in some form :S

Click to expand...

The first iterations will likely be quantuum co-processors much like separate floating point co-processors in the 80's & 90's. Most code will be conventional and only computations that wildly benefit from quantum computing will be done on that processor.

Silicon carbide? Doubt they will ever integrate that into actual CPU dies. Could probably connect it as its own "quantum chiplet" with other chiplets though.

Generally things made with SiC are done for two reasons: you can push a ton of power through them (both amperage wise and voltage wise), and/or to remove transistor switching frequency transients. Hopefully they used this material for the later property, and not because these are massively power hungry.

If the current ones have to be cooled with liquid helium for the quantum entanglement to manifest, how are they going to translate this to one that runs for any appreciable amount of time using these carbide wafers?

Accursed said:

So, Even if we do get quantum computers to be a thing, how long till the Unreal engine runs on em?

Seems like Quantum computers "work" completely different than the computers we're used to running.

Click to expand...

Yes. They could probably run emulators for our current binary based CPU's though. I mean if they get fast enough, they could still be faster, even running an emulator for boolean math (binary CPU's).

But you are right that the way they compute is a different, perhaps it is an expanded boolean type of math.

Quantum computers bits have a third state when they are both on and off. Have not researched how that is mathematically useful, or how problems needing computed can be written/converted into whatever the new version of the math is. It's a good question. I believe I read that someone started working on the math in the 80's, but haven't yet read any good explanations.

What the heck is boolean? -> Current binary CPU's use transistors to represent the boolean math functions and compute them. AND OR NAND NOR NOT XOR XNOR. These are binary math operations we create with arrangements of transistors (simple explanation here http://www.ee.surrey.ac.uk/Projects/Labview/gatesfunc/ ). AND is binary multiplication, OR is binary addition, NOT is inverse the input. CPU's can be written out as a giant mathematical (boolean algebra) flowchart with many inputs and outputs.
About the math: https://en.wikipedia.org/wiki/Boolean_algebra

Util they figure out the error correction qubit scaling overhead issues in smaller, near-term setups with implementations that will lead to the large systems actually capable of demonstrating that fabled "quantum supremacy"....


.... I'll take a 3990x.

GoodBoy said:

Yes. They could probably run emulators for our current binary based CPU's though. I mean if they get fast enough, they could still be faster, even running an emulator for boolean math (binary CPU's).

But you are right that the way they compute is a different, perhaps it is an expanded boolean type of math.

Quantum computers bits have a third state when they are both on and off. Have not researched how that is mathematically useful, or how problems needing computed can be written/converted into whatever the new version of the math is. It's a good question. I believe I read that someone started working on the math in the 80's, but haven't yet read any good explanations.

What the heck is boolean? -> Current binary CPU's use transistors to represent the boolean math functions and compute them. AND OR NAND NOR NOT XOR XNOR. These are binary math operations we create with arrangements of transistors (simple explanation here http://www.ee.surrey.ac.uk/Projects/Labview/gatesfunc/ ). AND is binary multiplication, OR is binary addition, NOT is inverse the input. CPU's can be written out as a giant mathematical (boolean algebra) flowchart with many inputs and outputs.
About the math: https://en.wikipedia.org/wiki/Boolean_algebra

Click to expand...

No need to emulate. SQUIDs, just like Parametrons of old, have no trouble ignoring superposition and doing normal logic.
Ones and zeros are replaced by sinewaves of opposite phase. Mix waves together, and whichever shouts loudest wins.
Its a majority vote thing. They force a superposition to decide, and thereby collapse to a conventional bit.

MAJORITY(A,B,1) is the same as OR
MAJORITY(A,B,0) is the same as AND
MAJORITY(A,B,C) is the same as CARRY
Flipping the phase is the same as NOT
Therefore you also get NAND, NOR, MINORITY
All three of which are universal logics that can make any other.
With no inputs, its also a random bit generator.

Already enough function in one device to claim to be an ALU.
An extra Parametron of five input MAJORITY(A,B,C,-CARRY,-CARRY)
provides an extremely simple way to compute both SUM and XOR.
A trio of Parametrons can also compute SUM and XOR if you are
only allowing a limit of three inputs. But wait, there's more...

These things are oscillators that amplify weak vote (or qubit, or signal from distant spacecraft) to a full scale oscillation.
Phase locked to a comnmon clock that pumps them like a kid on a swingset. But it doesn't care if they swing forward
on the odd or even clocks. Binary choice of phase lock gets you a swinging bit of dynamic memory to store a result.

I'm tellin you, it does about freaky everything, and with no transistor in sight.
If you wanna fake it with two transistors and no oscillation, a flip-flop can do
most of the same tricks except RND. You never gonna balance out a perfect
superposition. But for conventional logic as described above, that defect isn't
a problem. Just hook you up some mixer resistors and vote. Fully differential
I/O helps...

Either way latches up an answer and won't change with future inputs. Both
a very useful feature and huge pain in the ass. Have to turn one off to forget.
Three or more rolling waves of powered down forgetfullness are used to
draw new votes and logic forward into the empty wake they leave behind.
Set it and forget it...

Between those waves you have a fully isolated pipeline of latched logic.
If it takes 12 steps to do something, you can jam four operations into the
pipe. That pipeline is a freebie, might as well use it.

If not the Quantum Pocket Fisherman of logic thingamabobs, I dunno what is.

atp1916 said:

Util they figure out the error correction qubit scaling overhead issues in smaller, near-term setups with implementations that will lead to the large systems actually capable of demonstrating that fabled "quantum supremacy"....


.... I'll take a 3990x.

Click to expand...

Error correction could be a simple matter of not making the damn qubit too small.
A condensate of identical qubits oscillating together in one macroscopic device.
Like a laser for example. The parametrically pumped kind, though way too big...
https://www.rp-photonics.com/optical_parametric_oscillators.html
If a small percent go sideways, it ain't enough to mess up a majority vote.
Which is why the old Parametrons never suffered any such problem.

But if you can't reliably make smaller than an equivalent two transistor flip-flop,
whats the point? We are also not yet counting room for supporting mixers or
manifolds. Clock lines and biasing. And the optical example was obviously
ridiculous, only to make the point of many qubits oscillating together.....

At least Parametrons, you can make those yourself from off the shelf junk.
Get started down the path of understanding how oscillating logic works.
You have your choice of pumping with a magamp on the inductor side, or
varactor diodes on the capacitor side.

Lets pretend we wanted to do this as primitively as possible. To pump, we first
need some way to 2x periodically vary the tuning of a resonant circuit. Maybe
we hook up a motor to one of these, and spin it double fast...

Not showing one capcitor required in parallel for a complete resonant tank.

If you can tweak some variance to occur twice per cycle of resonance, like a kid
kicking out feet on a swing, thats all you need to pump up a random qubit to full
scale and find out which direction the kid was initially pushed. The output format
will be Binary Phase Shift Key'd. Of course you probably want to hook up three
inputs (maybe other kids with bungie cords) and do majority voting logic instead.

Just need a few slave boat drummers to keep them all kicking in phase lock.
Bring your own weapons. Paid after we get back.

-edit-

Or you could use a fixed coil and vary the capacitor side of things. Condensor
microphone might work. Then all them orphans would be out of logic jobs and
have to pick oakum instead. Maybe a few will have the elite timekeeping skills
to make it as drummers. Tune the logic resonators one octave below whatever
note drums are tuned to.

This calls for a round-robin of at least three drums. Every drum has to remain
in phase lock, even when muted. Adjacent sequential beats must also slightly
overlap to pull logic forward. Forgetful rest should trail behind each obsolete
bit to recycle its ability to accept a new vote. Also to isolate against backward
logic flow. 50% duty cycle per drum seems about right...

^^ But that's how you get...

GoodBoy said:

https://spectrum.ieee.org/tech-talk...le-qubits-quantum-computer-news-silicon-wafer

"A new technique for fabricating quantum bits in silicon carbide wafers could provide a scalable platform for future quantum computers. The quantum bits, to the surprise of the researchers, can even be fabricated from a commercial chip built for conventional computing."

Hope it works, we need a big breakthru in computing as Moore's Laws' demise looms...

Click to expand...

Quantum computing is not linear computing. You cant play games on quantum computers and you shouldn't treat them like it's a game. Moores law is not a scientific law.

Quantum computing should be literally fully controlled by governments and not corporations. It's an extremely, very extremely dangerous form of computing. I already have zero trust for governments but I have zero times infinity trust for a big tech firm and a quantum computer.

I literally cringe at the blind zealous support some of you show for technology. Zero regard for the capacity of those controlling the tech to be absolutely nefarious.

tangoseal said:

Quantum computing is not linear computing. You cant play games on quantum computers. Moores law is not a scientific law.

Quantum computing should be literally fully controlled by governments and not corporations. It's an extremely, very extremely dangerous form of computing. I already have zero trust for governments but I have zero times infinity trust for a big tech firm and a quantum computer.

Click to expand...

Such as Google and their Sycamore quantum computer.

I don't understand the Sun. Maybe we shouldn't trust it, or Santa.
Magnetic reconnection, too quantum...

Quantum computing will never replace conventional computing and vice versa. Each do certain things well in their own spheres which don't overlap. Being afraid of new technology is rather infantile and paranoid which only stymies development.

The future is coming whether we like it or not. There's always downsides to new technology, but we learn to deal with it. Fire, the Wheel, Gunpowder...

Never say never.

For example circuit, lets imagine an 8bit shift register that can rotate either direction.
Our playground equipment requires three drums "ABC", 24 swingsets, 24 bungies.
The logic function shall be copy (or buffer), as decided by one uncontested vote.

Perhaps two extra votes weigh in and overrule when we want to set an inital state.
Not required to show how this register rotates...

ABCABCABCABCABCABCABCABC
0..1..2..3..4..5..6..7..
00.11.22.33.44.55.66.77.
.0..1..2..3..4..5..6..7.
.00.11.22.33.44.55.66.77
..0..1..2..3..4..5..6..7
..00.11.22.33.44.55.66.7
7..0..1..2..3..4..5..6..
77.00.11.22.33.44.55.66.

Muted drums and resting swings advance to the right. Votes act only upon adjacent
waking seats. Swings resting "." abstain, which prevents any contest by backflow.
Rotate the opposite way by drumming CBA. Reverse bungie wiring is identical.

Primitive ECC for orphans who ignore drums or drummers who can't keep time.


If we duplicate each qubit 3 times and vote a majority of nine, no single error can
throw off the result. Of couse we don't need to do that unless we are dealing with
quantas of one. Single parametrons can swing a gazillion electrons, photons, flux
lines, spins, or convenient whatever might be found around the lab. Just mix three
crowds of aproximately equal influence and you get ECC benefits. Condensate is
also good for fanout. Cause how you split single qubits anyway?

Anything that superconducts (like Josephson Junctions in a SQUID), does so by
Cooper pairs attracted over long range interactions within the surrounding matrix.
Electrons normally repel ! Pairings can have interger spin and act as virtual bosons
that somehow pass through each other without collision. Such pairings and matrix
interactions don't sound like bosons or quanta of one to me, but maybe they are?

Normal logic with ECC, that works at room temperature, won't need to mess with
counting individual coup, Coops, or q-peen. Bragging rights still go to the smallest.

My bet is on analog computing.

My bet is on logic in memory, though I understand that even less...