Supercomputer Cray XC40 will build a map of neurons
Researchers at the Argonne National Laboratory use the Cray XC40 supercomputer to build a circuit for the interaction of neurons in the brain - the so-called connection . In the 20th century, scientists made up a nematode worm connection — this took 12 years. Current technology allows you to systematize large amounts of information much faster.
/ Flickr / Ivan / CC
The team of scientists needed a supercomputer to process the data obtained as a result of research using X-rays and electron microscopes. The subject of study - the brain shrew. Scientists plan to generate about exabyte data.
')
After the shrew, the next step is the mouse brain, which is ten times larger (75 million neurons). Ultimately, scientists want to explore the human brain, which contains 100 billion neurons connected by 100 trillion connections.
The process of making a connectom begins with an analysis of the brain at the submicron level — for this purpose, it is studied using x-ray microtomography at the Advanced Photon Source (APS) synchrotron radiation source of high brightness. As a result, scientists will receive images of the thinnest "layers" of material, on which blood vessels and cellular bodies are visible. The whole brain scan in the Argonne laboratory takes about an hour.
To then “see” the smaller details, use an electron microscope - due to the short-wave beams, it allows you to zoom to nanometer. Thus, it is possible to capture the synaptic connections between individual neurons of the brain. Next, scientists use a special set of computing tools that will transform the collected data into a neural map. For example, machine learning will help form predictive models and identify “points of interest”.
The Cray XC40 supercomputer, which will be used for this purpose, has a theoretical peak performance of 9 thousand teraflops - it is provided by 200 thousand Intel Xeon Phi 7230 64C cores with a frequency of 1.3 GHz. At the moment, it is the 16th most powerful computer in the world.
“Thanks to the capabilities of the Argonne supercomputer, we will be able to get truly revolutionary images that reveal the details of the functioning of the brain that we could not observe before,” said Kamel Fezzaa, a participant in the project.
Developing and effectively applying the methods of rapid analysis: data collection, analytics and the use of graphs and machine learning, scientists intend to promote the use of supercomputers in their research. The Cray supercomputer is already being used in the Human Brain Project (HBP) initiative - a ten-year project to study the human brain, founded in 2013.
Studies of the human brain can provide insight into how even small changes can cause the onset and development of neurological diseases - Alzheimer's disease or autism. In the future, it will help determine how to effectively treat these diseases.
The Argonne National Laboratory is a national research center for the US Department of Energy. Founded in 1946. Engaged in conducting fundamental research in physics, biology and environmental studies.
PS A couple of materials about high performance from our blog on Habré:
PPS A few more materials from the “First Corporate IaaS Blog”:
/ Flickr / Ivan / CC
The team of scientists needed a supercomputer to process the data obtained as a result of research using X-rays and electron microscopes. The subject of study - the brain shrew. Scientists plan to generate about exabyte data.
')
After the shrew, the next step is the mouse brain, which is ten times larger (75 million neurons). Ultimately, scientists want to explore the human brain, which contains 100 billion neurons connected by 100 trillion connections.
The process of making a connectom begins with an analysis of the brain at the submicron level — for this purpose, it is studied using x-ray microtomography at the Advanced Photon Source (APS) synchrotron radiation source of high brightness. As a result, scientists will receive images of the thinnest "layers" of material, on which blood vessels and cellular bodies are visible. The whole brain scan in the Argonne laboratory takes about an hour.
To then “see” the smaller details, use an electron microscope - due to the short-wave beams, it allows you to zoom to nanometer. Thus, it is possible to capture the synaptic connections between individual neurons of the brain. Next, scientists use a special set of computing tools that will transform the collected data into a neural map. For example, machine learning will help form predictive models and identify “points of interest”.
The Cray XC40 supercomputer, which will be used for this purpose, has a theoretical peak performance of 9 thousand teraflops - it is provided by 200 thousand Intel Xeon Phi 7230 64C cores with a frequency of 1.3 GHz. At the moment, it is the 16th most powerful computer in the world.
“Thanks to the capabilities of the Argonne supercomputer, we will be able to get truly revolutionary images that reveal the details of the functioning of the brain that we could not observe before,” said Kamel Fezzaa, a participant in the project.
Developing and effectively applying the methods of rapid analysis: data collection, analytics and the use of graphs and machine learning, scientists intend to promote the use of supercomputers in their research. The Cray supercomputer is already being used in the Human Brain Project (HBP) initiative - a ten-year project to study the human brain, founded in 2013.
Studies of the human brain can provide insight into how even small changes can cause the onset and development of neurological diseases - Alzheimer's disease or autism. In the future, it will help determine how to effectively treat these diseases.
About Argonne National Laboratory
The Argonne National Laboratory is a national research center for the US Department of Energy. Founded in 1946. Engaged in conducting fundamental research in physics, biology and environmental studies.
PS A couple of materials about high performance from our blog on Habré:
- IBM has opened access to a new 16-qubit quantum processor
- The first supercomputer DGX-1 based on Tesla V100 will be used in medicine
PPS A few more materials from the “First Corporate IaaS Blog”:
Source: https://habr.com/ru/post/338618/
All Articles