Nextbigfuture has covered the nanomechanical computer design created by Ralph Merkle, Robert Freitas, Tad Hogg, Thomas E. Moore,
Matthew S. Moses and James Ryley several times.
A team from UCLA and Lawrence Livermore National Labs has created a MEMS version of the nanomechnical computer design. They published the creation of actual logic gates created in different MEMS sizes. They are looking to implement a more complete system with 100-micron sized gates that with a chip operating at a megahertz. The UCLA and Lawrence Livermore team made some improvements on the Merkle, Freitas and team design by going through the process of actually building the parts.
Merkle, Freitas and team had the objective of designing a nanomechanical system that would be the easiest to fabricate using molecular nanotechnology when that capability arrives.
Electronic computers have ruled for the past 60 years but three-dimensional micro-additive manufacturing technology provides new fabrication techniques for complex microstructures which have rekindled research interest in mechanical computations.
They implemented the new digital mechanical computation approach based on additively-manufacturable micro-mechanical logic gates. The proposed mechanical logic gates (i.e., NOT, AND, OR, NAND, and NOR gates) utilize multi-stable micro-flexures that buckle to perform Boolean computations based purely on mechanical forces and displacements with no electronic components. A key benefit of the proposed approach is that such systems can be additively fabricated as embedded parts of micro-architected metamaterials that are capable of interacting mechanically with their surrounding environment while processing and storing digital data internally without requiring electric power.
A major advantage of micro-mechanical logic devices is that they utilize energy in a mechanical form and require no electrical power source or electronic components. As a result, such devices generate no electromagnetic signature and are highly insensitive to radiation damage.
If this was implemented with reversible computing, then Ralph and team calculated it would be billions of times more energy efficient than existing computers.
Nature Communications – Additively manufacturable micro-mechanical logic gates
The chief difference between the designs proposed in this paper and the existing mechanical logic gates is that the proposed designs achieve the following properties simultaneously:
* Functional completeness: All possible digital logic operations can be expressed by combining the designed logic gates. The functionally complete sets of binary logic gates are {AND, NOT}, {OR, NOT}, {NAND}, and {NOR}, which have all been demonstrated in this paper.
* Continuous operation: The proposed logic gates do not need to be reset to an initial state prior to the next logic operation. The presence of the input(s) will immediately trigger the operation.
* Scalable design: The proposed designs can function at all scales including the micro-scale. The flexure-based designs avoid sliding contact between surfaces and therefore avoid hysteresis and failure due to friction and wear. Designs that use sliding-contact bearings are not easy to fabricate and tend to bind due to intermolecular forces that become dominant on the micro-scale.
* Constant energy storage across different logic states: Each logic gate stores the same amount of total strain energy at different logic states. This allows for nearly zero-energy operation in theory.
The current 2PS/HOT system has a minimum micro-fabrication resolution of 800 nm and can generate optical trapping forces of up to 50 pN. It should be able to make logic gates of 100-micron size. 100 Micron logic gates would perform logic operations at Mhz frequencies. This estimation will be further investigated and verified in future work.
SOURCES – Arxiv, Nature Communications
Written By Brian wang
Brian Wang is a Futurist Thought Leader and a popular Science blogger with 1 million readers per month. His blog Nextbigfuture.com is ranked #1 Science News Blog. It covers many disruptive technology and trends including Space, Robotics, Artificial Intelligence, Medicine, Anti-aging Biotechnology, and Nanotechnology.
Known for identifying cutting edge technologies, he is currently a Co-Founder of a startup and fundraiser for high potential early-stage companies. He is the Head of Research for Allocations for deep technology investments and an Angel Investor at Space Angels.
A frequent speaker at corporations, he has been a TEDx speaker, a Singularity University speaker and guest at numerous interviews for radio and podcasts. He is open to public speaking and advising engagements.
We can probably handle the coupling issue, it will just make developing VLSI tools interesting. Our design tools might resemble something like topology optimization. We can specify ‘goals’ for components like the power transmission and optimize geometry from scratch rather than picking components from a library like we’ve done with electronic VLSI.
This is an initial exploratory implementation of an initial design. There’s still plenty of work until it reaches real-world applications. Ultimately you’d be looking at molecular-level implementations of much improved designs, built with nanotechnology tools. Both volume and cost will likely go down significantly by that point, and performance will go up. Don’t expect practical use at all for a while yet.
Good points, but keep in mind that this is proof-of-concept work, so of course there’s room for improvement. For the coupling issue, someone will probably come up with a decoupling method when it’ll become needed.
Making a device that uses 10 million times less energy per unit of compute work of current CPUs at the cost of making it take up 10 million times the volume is going to cost too much money to build in the first place.
I don’t see this getting much traction due to insurmountable economic reasons. Once you hit about 1 watt-hour per day of use for a device, there isn’t much economic reason for going lower. A 7 watt-hour mobile phone battery charged daily is like 50 cents per year of electricity. You can maybe go 10x more on the cpu package volume.
This sort of thing is going to be very niche for a very very long time.
These are not reversible like the gates proposed by Merkle et. al. In
addition, these gates also need some form of intermediate amplification,
whereas Merkle et al’s do not. Routing power for all the elements in a
mechanical computer, especially one using flexures, should prove to be
interesting. The same goes for VLSI design tools. In designing electronic computer chips we can decouple analysis of each of the logic units and routing signals is much simpler. In mechanical systems, things can be a lot more coupled. This is especially a problem with flexures which can store elastic energy and even deform out of plane.
It would be interesting to make a 4004 level mechanical computer, especially if one can be built out of all metal. Such a mechanical computer could work in environments with high temperatures and radiation. In fact NASA, has proposed making almost entirely mechanical rovers for Venus as electronics do not work well at 450 C.