Intel Corp., imec and five Flemish universities on Tuesday officially opened the Flanders ExaScience Lab at the imec research facilities in Leuven, Belgium. The lab will develop software to run on Intel-based future exascale computer systems delivering 1000 times the performance of today's fastest supercomputers, using up to 1 million cores and 1 billion processes to do so.
The ExaScience Lab will be the latest member of Intel's European research network – Intel Labs Europe – that consists of 21 labs employing more than 900 R&D professionals.
Designing exascale computers using current technology and design methodologies would mean the systems would become extremely hot and require a power plant to deliver the power needed to run them. When building a system consisting of millions of cores, getting all of them to work together for an extended period of time also represents a challenge. Hence, completely new computer programming methods and software will be required to bring power consumption to acceptable levels and to make the system fault tolerant. Power and reliability will be key challenges that will need to be understood to turn the vision of exascale computing into reality.
According to Intel, reaching the next level of supercomputing performance is about more than just reaching an arbitrary milestone, but it is about crossing many different thresholds of possibility that reside in the exascale domain, to provide scientists and doctors new tools to draw new insights from of massive amounts of data. It’s about, along the way, developing the technologies that will one day allow the cloud to scale to level where massive distributed computers can simulate reality and synthesize “holodeck” like science-fiction experiences, Intel said. Over the long term, high performance technologies become personal technologies.
The exascale labs begin research as Intel unveils new plans for the Intel many-core x86 architecture for high-performance computers, which build upon Intel’s history of many-core related research including Intel’s “Larrabee” program and single-chip cloud computing.
Breakthroughs in exascale computing could mean the ability to simulate very complex systems, impossible to replicate today like the human body or Earth's climate. The result, if the computing industry is successful, could mean finding cures for diseases or better predicting natural disasters. The Flanders ExaScience Lab will be focused at enabling scientific applications, beginning with the simulation and prediction of "space weather," or electromagnetic activity in the space surrounding Earth's atmosphere. Solar flares – large explosions in the Sun's atmosphere – can cause direct damage to Earth. Damage can be to electric power networks, pipeline systems and the quality of wireless communication, as examples. To accurately predict and understand the effects, exascale computing power is needed. Chosen for its extremely complex nature, the software findings are expected to be used and extended to address many other problems.
Tags: Intel, Exascale, HPC, SCC, Larrabee
Comments currently: 1
Discussion started: 11/18/13 11:18:53 AM
Latest comment: 11/18/13 11:18:53 AM
Columbia University has designed the first FM transmitter using a NEMS device from graphene. Graphene has the potential to move the per bit transaction costs from the pico joules per bit down to the nano joules per bit. Notwithstanding, the voltage required to power NEMS vs MEMS devices cannot be directly correlated so a 3-6 orders of magnitude improvement could put a graphene device in the atto joule per bit range of power consumption.
Looking at Greenarraychips.com the GA144F18a is a chip which demonstrates 1 nano watt power consumption at 345 GIPS. This 27 instruction processor schools RISC and CISC offerings demonstrating that sub-processors could be incorporated in chip topologies in such a way as to save the world ExaWatts of power in the future. Combing what I have just described will bring us to the next generation of wearable computers and implants.
I have read that Lithium Niobate chip implants which use longitudinal wave transmission as opposed to transverse wave transmission systems can forgo the issues associated with the losses during the transition through dissimilar mediums. That being said implantation and connectivity no longer are an issue. I have read about an implant which when attached to the optic never can either send or receive information via the optic nerve. The computer monitor or someone who monitors you might be the obvious uses for said chips.
For One Light
11/18/13 11:18:53 AM]
Add your Comment
Enter your username and e-mail address. Password will be sent to you.