darkmighty 283 days ago [-]
> A key performance metric of computers is their energy dissipation. One contribution to dissipation is friction at the rotary joints in each logic gate. Due to the joint’s small frictional drag, mechanical computers constructed from them can, in principle, dissipate orders of magnitude less power than conventional semiconductor computers, while still operating at relatively high speeds.

This claim seems dubious, perhaps someone has more expertise to comment. The justification

> Operating this lock involves rotation at the joints by up to ∆θ≈1 rad. The model system analyzed in [11] is an excerpt of the links and joints shown in the closeup on the right of Figure 24. From [11, Eq. 2], this rotation dissipates bout 2.4×10−27J per rotary joint when operating at f = 100 MHz.

Seems akin to saying "We operate our microchip at 1 microvolt / 1 pico ampere at f=100MHz, giving 10-26J per operation." -- which seems like a silly aspiration without carefully analyzing noise and quantum mechanical constraints. (without which it would seem almost any computing device could operate at arbitrarily low power).

pjc50 283 days ago [-]
However it would have a nice property: not operating would consume zero power. Useful for low-leakage cases or energy harvesting.
sigstoat 283 days ago [-]
it is possible to design microprocessors which can just utterly stop. https://en.wikipedia.org/wiki/Static_core power consumption doesn't drop to exactly zero, but it can come into competition with the quiescent current of the power supply.
deepnotderp 283 days ago [-]
That's possible for conventional CMOS as well, one example is power gating. Not exactly zero leakage, but pretty damn close.
QAPereo 283 days ago [-]
The keyword there is, “in principle,” which is even less than “on paper.” Essentially if you had elements which asymptotically approached frictionless interactions, and you had an idealized power source... sure, less entropy than firing electrons through Si.

Things ignored to make that work:

Materials which are likely to ever exist.

A degree of mechanical perfection on the order of one of NASA’s fused quartz spheres.

An idealized means of transferring energy to the device.

A perfect vacuum.

Perfect shielding from external influence, especially vibration.

The faster you crank this up, the more energy is in the form of Ke in the device, and like a flywheel, it may come to resemble a bomb.

fernly 283 days ago [-]
Did no one else notice the name of the lead author? Ralph C. Merkle [1] is a rather well known computer scientist, involved in public-key cryptography, hashing algorithms, and inventor of the Merkle Tree [2] that is at the core of the blockchain.

[1] https://en.wikipedia.org/wiki/Ralph_Merkle

[2] https://en.wikipedia.org/wiki/Merkle_tree

joshu 283 days ago [-]
I did. Iirc Merkle is big into nanotech too.
kurthr 282 days ago [-]
Yes, he's collaborated with Eric Drexler, the author of Engines of Creation at the Forsight Institute.


userbinator 283 days ago [-]
It somewhat reminds me of https://en.wikipedia.org/wiki/Z1_(computer) although this mechanism is definitely superior in terms of friction. I guess the majority of research into mechanical computers stopped shortly after electrical/relay ones started becoming more interesting, which is why there's definitely better mechanical designs possible.

In fact, people probably use a lot of mechanical devices which can serve as logic gates without knowing it --- although everyone focuses on electronic digital computers, the mechanisms on which computers can be built are surprisingly vast and simple.

Also, the PDF is rather bloaty for the content, because all the figures are extremely high-resolution bitmap images instead of vector graphics, despite looking like they were created with a vector graphics program.

whitten 284 days ago [-]
this is intriguing. since the links and rotary joints will be susceptible to friction I wonder if you could pump oil through the machine to remove any shavings that might be created.

I wonder if the fact that the gates etc. are in a single plane could be used to make it function more effectively as well.

It certainly is using a different mechanical approach than the Differential Machine.

fernly 283 days ago [-]
They make the point that rotary joints are for conceptual illustration, and it could equally use hinge or flex joints. This would be relevant at the nano scale,

> In particular, a molecular version of this architecture could use stiff covalently-bonded nanotubes for the links and single bonds for the joints.

In that case joint friction would not be a consideration, or at least I would imagine things like electrostatic or molecular bonding forces would be larger.

godelmachine 284 days ago [-]
I was going to submit the same paper, and had the same question, along with another one. The abstract says that the system can be scaled down to molecular level, much like electronics has been scaled to sub mm or nm level. But at the sub nanometer level, we face quantum effects which are hard to overcome. Would mechanical based computing systems give us the same headache as we try to scale them further down?
mhb 283 days ago [-]
quantum effects which are hard to overcome

This is discussed thoroughly in Nanosystems (Table of Contents - http://e-drexler.com/d/06/00/Nanosystems/toc.html)

sp332 283 days ago [-]
I think we know how to do links and rotary joints at that scale, and that's all you need for these machines.
jakewins 283 days ago [-]
Heh, one step closer to a real self-replicating rep-rap :) Now to implement a PID controller..
bowmessage 283 days ago [-]
This is a very scary idea!
sitkack 282 days ago [-]
If people like reading outlandish papers like this, see the full queue.

2018 https://arxiv.org/list/cs.ET/18

2017 https://arxiv.org/list/cs.ET/17

sytelus 283 days ago [-]
Billiard Ball Computer is just made out of billiard balls :).


It's turing complete. I'm wondering if the mechanism described in paper does something above and beyond...

colordrops 283 days ago [-]
I believe that this is the idea behind reversible computing, which I would assume would require superconducting circuits needing energy only to drive inputs, and the CPU would expend negligible heat.
xg15 282 days ago [-]
Technical question: Could someone explain why, in the shift register, the output lock is needed at all?

Couldn't you simply connect the output of each holding lock to the respective input of the next cell and get the same results?

VectorLock 283 days ago [-]
It looks like you could really easily prototype and make toy projects with this system using a peg board, and maybe a half dozen parts: pins, bell cranks, the locks, spacers/washers, joints and links a variety of lengths.
narrator 283 days ago [-]
Isn't this how the technology was described in Stephenson's book "The Diamond Age" in which computing was done with nanoscale Babbage like machines?
abecedarius 283 days ago [-]
Pretty much. See Nanosystems, linked in another comment, for a technical treatment. Its logic design used sliding rods, so friction could be more of an issue, especially if you can't make it atomically precise. (With atomic precision you could design the surfaces for https://en.wikipedia.org/wiki/Superlubricity) Quick intro here: http://www.halfbakedmaker.org/blog/58

Merkle wrote a later paper on a different approach, buckling-spring logic: http://www.zyvex.com/nanotech/mechano.html I'm not sure how this latest design is supposed to be better still. I haven't really dug into either of them.

BTW this is the same Merkle who's always getting mentioned in blockchain articles.

282 days ago [-]
bane 283 days ago [-]
At large scale, I wonder what kind of horsepower would be required to run a "useful" implementation made at a scale that an normal workshop could produce?
random_user456 283 days ago [-]
i want to build a few these at 3d print level, but would love to see someone prototype some of these at micro level, anyone have access to lithograph? Another basic question is potential clock speed based on size and material used. curious if anyone has done some basic ballpark theoreticals
grondilu 282 days ago [-]
Can't someone make a simulation in algodoo or something? That'd be useful for a proper illustration of how it works.
283 days ago [-]