Hibernation Mode: Calculating the energy needs of post-human uploads

Anders Sandberg tries to calculate the amount of energy it would take to sustain an uploaded human consciousness
How efficient could a postbiological civilization be? The current IBM roadrunner does 376 million calculations per watts. If we take my mid-range estimates of computing needs, 10^22 to 10^25 FLOPS, then a single emulation would need 10^13 to 10^16 watts. The total insolation of Earth is about 10^17 watts, so this won't do - there would be space for just a few minds on the entire planet. But current research on zettaflops computing suggest we can do much better. A DARPA exascale study suggests we can do 10^12 flops per watt, which means "just" a dozen Hoover dams per mind. Quantum dot cellular automata could give 10^19 flops per watt, putting the energy needs at 200-2000 watts. That is between 2 and 20 times the current wattage of a current human. However, we bio-humans get our energy through the inefficient method of having plants collect sunshine (at about 3%) efficiency, then we either harvest them and eat a small part of them (expending a lot of agricultural energy) or have animals eat them (at a few percent efficiency) and finally we eat the result, again with a few percent efficiency. A brain emulation of this type would just need a few square meters of solar panels (plus night-time energy storage). In terms of area and energy required, these postbiological humans would have far smaller material requirements than we do. They could also run slower to save energy.
[via Charlie Stross] Photo: sporkwrapper
This entry was posted in Uncategorized. Bookmark the permalink.

5 Responses to Hibernation Mode: Calculating the energy needs of post-human uploads

  1. Nelson.C says:

    Arthur: I’d notice the difference.

    Zaphod: No, you wouldn’t, you’d be programmed not to.

  2. CALCULON says:

    @ Nelson.C: The perfect quote. I tip my hat to you.

    The singularity and mind transfer both terrify and excite me in equal measures. Sure, they’re not mutual, but if we were ever going to live forever, either as a species or as individuals, one or both would probably need to take place.

  3. mellon says:

    A watt is a measurement of power, not a measurement of power consumption. Calculations happen over time. So all of these numbers are nonsense. I suspect someone misquoted the original source. It should be 376 million calculations per watt/hour, or something like that. Sorry to be pedantic, but the article just doesn’t make sense as quoted.

  4. Dewi Morgan says:

    how many flops you want to use to emulate a brain depends on what level you want to emulate it at: are you simulating all the chemistry, or are you just emulating the logic?

    Sort of the difference between emulating the logic of a Z80 CPU (can be done in javascript), or simulating the current flow and silicon doping (would take a supercomputer).

    Speaking of saving bits… anyone else noticed that recaptcha’s got a lot lower res lately?

  5. jimkirk says:

    I’ve read several estimates of human brain processing power at around 100 million megaflops, or 10^14 flops. Anybody know where these “mid-range” estimates of 10^22 to 10^25 come from? That’s 8 to 11 orders of magnitude higher.

    I even have my doubts about 10^14. Architecture and algorithm are fundamental elements that can greatly improve performance.

    Even a 4 bit processor might do, if you’re programmed not to notice.

Leave a Reply

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

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>


More BB

Boing Boing Video

Flickr Pool




Displays ads via FM Tech

RSS and Email

This work is licensed under a Creative Commons License permitting non-commercial sharing with attribution. Boing Boing is a trademark of Happy Mutants LLC in the United States and other countries.

FM Tech