-DARPA’s Omnipresent High Performance Computing (OHPC) Program.

Posted: June 26, 2010 in 2010, Articles
Tags: ,

Network World:

Not known for taking the demure route, researchers at DARPA this week announced a program aimed at building computers that exceed current peta-scale computers to achieve the mind-altering speed of one quintillion (1,000,000,000,000,000,000) calculations per second.

Dubbed extreme scale computing, such machines are needed DARPA says to “meet the relentlessly increasing demands for greater performance, higher energy efficiency, ease of programmability, system dependability and security.”

10 hot energy projects that could electrify the world

DARPA says its Omnipresent High Performance Computing (OHPC) systems will involve all manner of new research and development.  Specifically the outfit is looking for:

  • Hardware, software and algorithms for reducing and managing power requirements for high performance computing systems, including the memory and storage hierarchy
  • Hardware, software and language design that enables highly programmable systems, which reduces the need for users to be aware of system complexity, including heterogeneous cores, the memory hierarchy
  • Improved hardware and software for bolster system dependability, managing the component failure rate, and security compromises including approaches for shared information and responsibility among the OS, runtime system, and applications
  • Scalable I/O systems, which may include alternatives to file systems
  • Self aware system software, including operating system, runtime system, I/O system, system management/administration, resource management and means of exposing resources, and external environments

According to DARPA: “Advances in Commercial Off-The-Shelf systems performances were enabled by increases in clock speed, decreases in supply voltage, and growth in transistor count. These technology trends have hit a performance wall where increasing clock speed results in unacceptably large power increases, and decreasing voltage causes increasing susceptibility to transient and permanent errors. Only increasing transistor count continues to drive performance increases, with value only if energy can be minimized while optimizing the ability to efficiently utilize available concurrency. Increasing density has not helped reduce the energy costs of data transport across a chip, between neighboring chips, or between chips on disparate boards. Current interconnect protocols are beginning to require energy and power budgets that rival or dwarf the cost of doing computation.”

The extreme machines are part of DARPA’s overarching Ubiquitous High Performance Computing (UHPC) program which is looking to develop low-energy architectures and protocols for logic, memory, data access, and data transport.  It also has as goals to develop dependable computer systems that place a high priority on resiliency and security at all system levels; concurrent management and the efficient use of massively parallel resources; and run a self-aware OS that manages real-time performance, dependability and system resources,  DARPA says.

DARPA says it expects prototype UHPC systems by 2018.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s