Eden.fm

Source: Dailytech

Los Alamos researchers are putting this power to work with a program dubbed PetaVision. The program was created to model neuron and synapse interaction in the visual cortex of the human brain. The brain uses over a billion neurons and trillions of synapses alone to process the visual information it receives and is one of the most complicated processes known to exist in grey matter.

Supercomputers like Roadrunner bring new possibilities for modeling human recognition systems, and the advances are not likely to stop there. In the past, computers have been unable to flawlessly perform cognitive tasks that the human brain does easily; tasks like picking out a face in a crowd, or detecting oncoming vehicles in traffic. Such a large step up in processing power may enable scientists to breech this difficult wall in mimicry.

The researchers used PetaVision to set a processing record with Roadrunner, spinning up to an astonishing 1.144 petaflop/s. “Just a week after formal introduction of the machine to the world, we are already doing computational tasks that existed only in the realm of imagination a year ago,” explains Terry Wallace, associate director for Science, Technology and Engineering at Los Alamos.

The supercomputer’s architecture is based on a hybrid node system. Each node contains two AMD Opteron dual-core and four PowerXCell 8i processers. The PowerXCell CPUs are derived from the same Cell processor used in the Sony Playstation 3 and act as computational accelerators for the Opterons.