Foresight Institute Logo
Image of nano

Cellular automata used for 700-bit parallel processing

We’ve received an update on work by our friend Anirban Bandyopadhyay at the National Institute for Materials Science in Tsukuba, Japan.  Here’s the abstract of his recent Nature Physics paper:

Modern computers operate at enormous speeds—capable of executing in excess of 1013 instructions per second—but their sequential approach to processing, by which logical operations are performed one after another, has remained unchanged since the 1950s. In contrast, although individual neurons of the human brain fire at around just 103times per second, the simultaneous collective action of millions of neurons enables them to complete certain tasks more efficiently than even the fastest supercomputer. Here we demonstrate an assembly of molecular switches that simultaneously interact to perform a variety of computational tasks including conventional digital logic, calculating Voronoi diagrams, and simulating natural phenomena such as heat diffusion and cancer growth. As well as representing a conceptual shift from serial-processing with static architectures, our parallel, dynamically reconfigurable approach could provide a means to solve otherwise intractable computational problems.

He explains:

…we have realized 700 bits parallel processing using cellular automaton for the first time in the world. This is a significant advancement from our 16 bit parallel processing which you highlighted in your website (…This invention may be in coherence with the Feynman’s vision…We can solve some problems which computers will take more than the age of this universe. We did it in 6-10 minutes (in the Nature Physics paper).

Some coverage:
Anirban writes, “Hope you may like this.”  We do indeed!  —Christine Peterson

4 Responses to “Cellular automata used for 700-bit parallel processing”

  1. miron Says:

    The claim that “We can solve some problems which computers will take more than the age of this universe” makes it hard to take this guy seriously.

    It makes no sense to claim that his method is not a computer. It is one, with many more processing elements.

    It would also be good to know if his processing elements are “Turing Complete” – i.e. if they can execute arbitrary programs while maintaining the performance gains.

  2. Gopal Nair Says:

    Well indiscriminate use of terms like ‘computer’ capability to elemental processing at nodes amounts to superfluous. Computers are a different system. One must understand there are several processing units working in any molecule as they form the molecule itself but the difficulty is to use it in a cohesive way and then make it scalable to use for a beneficial mode. Achieving some thing like a state change can be claimed as processing but claiming to the level of ‘computer’ ‘computer’ blinds the logic and that too challenging the billion years for a single computation. One must understand the automata what we use today in logic circuits switch, doesn’t work exactly in the cognitive paradigm of brain where massive state space structure keeps and connects the entity and information. Hence, claims of reality could be admitted but imaginative predication with today’s gate computing logic unsuitably imposed on molecular potentials and claiming processing, is too tall. It may be suitable for weekly, break fast dose magazines who starve for revolutionary exposure, not for true scientific spirit. Nature should have been more careful. However, let them develop and scale it for meaningful coherent multifunction operation and not make ‘computers’ because in that sense each cell is a computer. It is years away and understanding the brain with massive neuronal structural state properties, is still under exploration level only. Still I encourage them to look in to refining multilevels of link and perfect it.

  3. unanimous Says:

    Intractable problems like P=NP are fundamental problems of computer science. Scaling up of the processor built here will enable proving one of the biggest problem of this century.
    The comment whether elements are “Turing complete” is rediculus when it is said that they “can solve some problems” not all. Can we ask them about those some problems?
    Looking for universality in the computing constructs of above work for “arbitrary programs” clearly shows lack of understanding of the work. The basic text books of CA teaches us about random attempts of replicating natural events.

    It would be very interesting if some scientist in the world invents “arbitrary programming” in CA by theory. Then after that invention we may ask experimentalists to realize that. Till then it is better not to prove our weakness in the fundamentals.

  4. Duodecahedron Says:

    I have made some calculation to verify “more than the age of this universe” claim.

    YouTube video of the above paper ( shows that for a moving cell state computer needs to calculate 2^n area. They used 500 cells to solve cancer cell problem on molecular layer. For a computer to solve this, it should calculate 2^500 ~ 10^150 areas for every update. Say, one area means one bit.

    If a computer operates 10^30 bits per second it will take 10^120 seconds which is more than 10^15 years (age of this universe!) to solve the problem (following the way CA solves in the above paper).

    The molecular CA presented in the above paper simply handles one intractable situation, which they have used for two problems.

Leave a Reply