Supercomputer Vs Human Brain

Question

With supercomputers doing calculation in petaflops ($10^{15}$ Calculations per Second), have we crossed the speed of Human Brain?

Answer 1

I will just show the statistics of last attempt to mimic the brain process.

In 2011 fastest computer in Japan was launched:

K computer OR SPARC64 VIIIfx 2.0GHz

Features:

• Manufacturer: Fujitsu
• Cores: 705,024
• Linpack Performance (Rmax) 10,510 TFlop/s
• Theoretical Peak (Rpeak) 11,280.4 TFlop/s
• Power: 12,659.89 kW
• Memory: 1,410,048 GB (16GB RAM per CPU)
• Processor: SPARC64 VIIIfx 8C 2GHz (88,128 CPUs, 8-core each)
• Operating System: Linux

It's currently world's 4th fastest supercomputer.

Source: Top 500 Supercoputers

Image Source: http://wondergressive.com/wp-content/uploads/2014/01/WG-k-computer.jpg

Now,

An 83,000-Processor Supercomputer Can Only Match 1% of Your Brain

...The most accurate simulation of the human brain to date has been carried out in a Japanese supercomputer, with a single second’s worth of activity from just one per cent of the complex organ taking one of the world’s most powerful supercomputers 40 minutes to calculate. Researchers used the K computer in Japan, currently the fourth most powerful in the world, to simulate human brain activity. The computer has 705,024 processor cores and 1.4 million GB of RAM, but still took 40 minutes to crunch the data for just one second of brain activity...

Source: http://gizmodo.com/an-83-000-processor-supercomputer-only-matched-one-perc-1045026757

Human Brain:

Do we have brain to spare?

by David A. Drachman, MD

Within the liter and a half of human brain, stereologic studies estimate that there are approximately 20 billion neocortical neurons, with an average of 7,000 synaptic connections each.1 The cerebral cortex has about 0.15 quadrillion synapses—or about a trillion synapses per cubic centimeter of cortex. The white matter of the brain contains approximately 150,000 to 180,000 km of myelinated nerve fibers at age 20, connecting all these neuronal elements. Despite the monumental number of components in the brain, Szentagothai estimated that each neuron is able to contact any other neuron with no more than six interneuronal connections—“six degrees of separation.”

Image Source: http://www.nature.com/polopoly_fs/7.2933.1329907514!/image/far-to-go.jpg_gen/derivatives/fullsize/far-to-go.jpg

Why Brain beats Supercomputer?

Why is it so hard for computers to reproduce what your grey matter does as a matter of course? Volume. The human brain consists of about 200 billion nerve cells (neurons) that are linked together by trillions of connections called synapses. As the tiny electrical impulses shoot across each neuron, they have to travel through these synapses, each of which contains about 1000 different switches that route that electrical impulse. In total, one human brain could contain hundreds of trillions of these neural pathways. It's like a Choose Your Own Adventure book that stretches from here to Jupiter.

Answer 2

There's a very big difference between doing the calculations needed to simulate the human brain (or any animal brain - we can do a fairly decent job on C. elegans), and doing computations. While a basic leaky integrate & fire model is fairly simple, to accurately simulate a single neuron in real time takes a pretty fast computer. See e.g. these links for more, http://www.neuron.yale.edu/neuron/, http://bluebrain.epfl.ch/ http://www.artificialbrains.com/darpa-synapse-program.

Brains do well what computers do poorly, and vice versa. The HP12C calculator on my desk, made in the early '80s, can do many sorts of mathematical calculations faster than I can, because brains aren't really optimized for doing, for example, compound interest calculations. But my brain, or my dog's brain for that matter, can do all the 'calculations' needed to process a stream of images, recognize that they show a ball, predict its path, and catch it - while simultaneously processing visual, auditory, olfactory, and kinesthetic inputs for obstacle avoidance and threat recognition, and (in my case, anyway) perhaps thinking about how to solve complex programming problems.

Maybe I should add that there are basically three different things that a computer can do in this context:

1. Complex calculations, whether compound interest, weather modeling, seismic tomography or whatever.
2. Simulating the actual behavior of a brain, or some subset of one, starting from a single neuron.
3. Trying to replicate what a brain does through algorithms, or Artificial Intelligence.

(1) is what we mostly use computers for, and they're way better at it than human brains. (2) is a special case of (1), which we do because we don't have much of an idea how to do (3), and having a simulated brain we can alter at will might help in learning how real brains do what they do.

Answer 3

A simple calculator (which is a computer) calculates 298465+46547 much faster than us. a PC forecast possible events faster than us. But even the fastest computer can't recognize a new pattern faster than us (the computer may never do that). So each one has its relative advantages.