BLUE GENE and confusion about flops and tops and whistles

From: turin (turin@hell.com)
Date: Fri Mar 31 2006 - 01:24:05 MST


This is at the new postscript at the end of "How Long Before Superintelligence" and I have some questions about this, so please answer them even if they might be stupid.

Postscript III
(30 October, 2005)

The fastest supercomputer today is IBM's Blue Gene/L, which has attained 260 Tops (2.6*10^14 ops). The Moravec estimate of the human brain's processing power (10^14 ops) has thus now been exceeded.

<So Blue Gene is doing 260 teraflops. I thought the brain was supposed to do 1-15 petaflops or am I wrong?
<

The 'Blue Brain' project was launched by the Brain Mind Institute, EPFL, Switzerland and IBM, USA in May, 2005. It aims to build an accurate software replica of the neocortical column within 2-3 years. The column will consist of 10,000 morphologically complex neurons with active ionic channels. The neurons will be interconnected in a 3-dimensional space with 10^7 -10^8 dynamic synapses. This project will thus use a level of simulation that attempts to capture the functionality of individual neurons at a very detailed level. The simulation is intended to run in real time on a computer preforming 22.8*10^12 flops. Simulating the entire brain in real time at this level of detail

<But here it says we do only 22.8 teraflops, I mean since the computer is running in "real time", that would clock us <at 22.8 teraflops yes? The NEC earth simulator does 40 teraflops. There's a beowulf cluster a few miles north of <where I live that does 26? Is this true? I'm not disappointed if we do 3 orders of magnitude less than I thought, <maybe we can hack ourselves faster, I jus thought we weren't as far along. The Cell chip in the new Play Station III <will do a teraflop, well if we had the right architecture and compilers for it we could

 (which the researchers indicate as a goal for later stages of the project) would correspond to circa 2*10^19 ops, five orders of magnitude above the current supercomputer record. This is two orders of magnitude greater than the estimate of neural-level simulation given in the original paper above, which assumes a cruder level of simulation of neurons. If the 'Blue Brain' project succeeds, it will give us hard evidence of an upper bound on the computing power needed to achieve human intelligence.

<I suppose it would give us an upper bound on the computing power. I dont know. I dont know that much about <connectionism and neural nets, what do you guys think?

Functional replication of the functionality of early auditory processing (which is quite well understood) has yielded an estimate that agrees with Moravec's assessment based on signal processing in the retina (i.e. 10^14 ops <for whole-brain equivalent replication).

<There it is, a teraflop. We do a teraflop? That's all? Well, hopefully processing speed of 3 magnititudes of my errour will not diminish the sublime algorithms which the mind uses. Hopefully hacking the human mind wil lgive us a shortcut to superintelligence........ hopefully. If we just have a few very effective learning algorithms and have to be in crucially correct environments for our cognitive architecture to develop properly, I don't know if hacking the human mind will help us with superintelligence that much. we would have to simulate the body and the environment as well........ The point is, no matter what our processing speed, I hope the code of the human mind is useful in making superintelligence. Because if our cognitive architecture is if not optimal a good way to go about making superintelligence, maybe our powers of prediction and interaction with it would be better.

I am thinking either an upload would be very very dangerous or very very safe. I cannot see how there would be a wishy washy upload unless it was incorrectly instantiated or kept on the equivalent of AI sedatives all day long.

EOF
<
<

EOT;



This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:56 MDT