Jos Kirps's Popular Science and Technology Blog

Software simulation of a human brain

October 07, 2009

The Blue Brain Project is an attempt to create a synthetic brain by reverse-engineering the mammalian brain down to the molecular level, and it is hoped that it will eventually shed light on the nature of consciousness.

In December 2006 the initial goal of the project had been completed: the simulation of a rat neocortical column, which is also responsible for conscious thought of rats. Henry Markram, director of the Blue Brain Project, then said that he thinks it will be possible to simulate a human brain, including aspects of consciousness, within only 10 years. The project has been launched in 2005 by scientists in Switzerland, Europe, and uses an IBM Blue Gene supercomputer running a software called NEURON. The NEURON software is capable of simulating an artificial neural network and offers a biologically realistic model of brain structures.

NEURON was primarily developed by Michael Hines, John W. Moore, and Ted Carnevale at Yale and Duke. The software simulates brain structures while it offers a shell, an own scripting language as well as a Python interface. The software is known for its extreme parallelism, which allows to take full advantage of massively parallel supercomputers. The project currently uses an IBM Blue Gene supercomputer, which is required to allow simultaneous simulation of thousands of neurons (and probably millions in future).

The IBM Blue Gene supercomputer is powered by thousands of PowerPC processors. Each compute node contains two CPUs as well as a cache sub-system and the logic to support multiple communication sub-systems. Compute nodes are packaged two per compute card, with 16 compute cards plus up to 2 I/O nodes per node board. There are 32 node boards per rack. Each Blue Gene/L node is attached to three parallel communications networks, including a 3D toroidal network for peer-to-peer communication between compute nodes.

Using this outstanding combination of software and hardware, the project leaders even expect that it may become possible to simulate a first mammmal brain as soon as in 2010.

What's next?