The Cray supercomputer, the iPhone and blood

When I was in high school, the most powerful computer money could buy was the Cray 2. The thing would require a room all by itself. I don’t know how much it sold for, but an old MIT article says that it cost $140 per central-processing unit hour. In other words, if you had to ask how much it cost, you could not afford it.

It provided 1.9 GFLOPS peak performance. It had 1 GB of memory.

In a recent iPhone, the GPU alone has more number crunching power than the Cray 2 had and just as much memory.

We can play a little thought experiment. What if I had gone to the Cray engineers and told them that, one day, I would be able to fit their entire computer in my pocket. What would they have said?

And what if, today, I were to tell you that in 40 years, we will be able to fit all the computational power of your phone into nanobots that can live in your blood stream?

9 thoughts on “The Cray supercomputer, the iPhone and blood”

  1. I guess I’d say that a human red blood cell has 10 cubic micrometers, so putting a gigabyte in there leaves about a cubic nanometer per bit, which can fit about 50 silicon atoms, which seems like not quite enough for a RAM.

    I’d further say that progress is often in unexpected directions, e.g. we don’t have cars that are a million times faster than before, and probably if we have nanobots, their value will come from being networked.

    1. I was not thinking in terms of storage capacity, but it is a good insight.

      The human genome uses about 1.5 GB (http://bitesizebio.com/8378/how-much-information-is-stored-in-the-human-genome/). White cells carry our DNA so, in a practical sense, we already have “nanobots” carrying around more than 1GB of data in our blood, for some unfair definition of “nanobot”.

      According to Wikipedia, Church’s lab managed to store 5.5 petabits in each cubic millimeter of DNA. Or, in other words, they store one bit per 180 cubic nanometers. It is somewhat larger than one bit per cubic nanometer, but that’s a practical result we can achieve today.

      So, in any case, I am not worried about storage capacity.

      I’d further say that progress is often in unexpected directions, e.g. we don’t have cars that are a million times faster than before, and probably if we have nanobots, their value will come from being networked.

      Right. So imagine that you could get a computer the size of a red blood cell with the computational power of an entire brain… what could it do? It would be severely limited by bandwidth. And, of course, it takes energy to compute, and computation tends to generate heat. So a single cell has serious physical limitations in how much computation it can run.

  2. Judging from the sort of discussions that are going on in the realms of astrophysics, subatomic physics and even the “soft sciences”, I rather doubt if the supercomputer in the cell will be the big deal in 40 years.

    I rather think it will come from an expansion of consciousness and of awareness, and it will open doors that are today not only unimaginable but are beyond the power of human speech.

    Contact with extra-terrestrials is one obvious possibility (or rather, disclosure of contact, if you are convinced they are already operating merrily under the radar).

    Cloning of oneself to have a “close companion”? It’s late Wednesday night between volcanoes in the west Java mountains so I won’t attempt any more silly projections.

    Liked the comparison, though. I heard a similar one from a woman who worked for the IRS in 1960, in New Orleans. The government went to work on one square city block, strengthening floors for the installation of heavy equipment. According to this lady, the entire second floor of the building – one block square – was packed with mainframe computer hardware. I would imagine it was mostly doing donkey work like mailing lists.

    The capacity of this behemoth? 40 megabytes.

  3. In answer to the post-ending question: “You want to put some money on that?”

    Many smart people who seriously know what they’re talking about (and know all the failed predictions of the past) believe that Moore’s Law (and it’s many related “laws”) will be a thing of the past in less than 10 years. The insane exponential trends of computing over the last half century have to end at some point.

    1. In answer to the post-ending question: “You want to put some money on that?”

      In this post, I have made no prediction, only asked a question.

      Yes, I am willing to make bets regarding the progress we shall make in computing though not for 40 years in the future. I cannot be reasonably certain that I will be alive then.

      Many smart people who seriously know what they’re talking about (and know all the failed predictions of the past) believe that Moore’s Law (and it’s many related “laws”) will be a thing of the past in less than 10 years.

      In the context of Moore’s law, most of the failed predictions were smart people predicting that Moore’s law would end. It did not. The pessimists were wrong.

      I have covered some of my thoughts on the end of Moore’s law elsewhere:

      http://lemire.me/blog/2015/07/17/going-beyond-our-limitations/

      The insane exponential trends of computing over the last half century have to end at some point.

      We have been on an exponential curve ever since life started on this planet millions of years ago. Computing is just the latest embodiment of evolution.

      Will there be an end to our flat silicon processors? Of course. But you are mistaken if you think it means that things will stop evolving ever faster…

  4. I think that quantum computers will change the world and our concepts of computing. With one qubit, you can test 2 things simultaneously (and not sequentially). With 8 qubits, you can carry out 256 simultaneous tests. Once we discover how to harness this power, it will be mindblowing. It will severly compromise traditionnal cryptography based on secret keys. Imagine that a 16 qubits quantum computer can crack instantaneously a 65k crypto key. The excellent book “The Code book” by Simon Singh explains the basics on quantum cryptography.

    It may seems like fiction but I believe that the Darpa is more advanced in this field than we think. If they come up with some working quantum computer, they wont tell us. As seen in the past, if you can crack someone else’s code, do not let him know. This way, he will continue to use his compromised code.

    1. I think that quantum computers will change the world and our concepts of computing. With one qubit, you can test 2 things simultaneously (and not sequentially). With 8 qubits, you can carry out 256 simultaneous tests.

      Chances are that the computer in front of you does much the same. Modern CPUs are superscalar, multicore and vectorized.

      We have outgrown scalar computing about a decade ago (more or less).

      State-of-the-art GPUs can run thousands of operations simultaneously.

      It may seems like fiction but I believe that the Darpa is more advanced in this field than we think. If they come up with some working quantum computer, they wont tell us.

      The American military has quantum computers, like Google… but I don’t think that’s what they use.

      Nvidia and Intel are probably the guys providing the hardware.

  5. I love the discussion – my bet is inside of 20 years… Really we can already print DNA base pairs and are learning more and more how the genome is folded and the enzymes it is creating. Once these processes are completely understood we may not even need the nanobots…

Leave a Reply

Your email address will not be published. Required fields are marked *