1. Home
  2. Computing & Technology
  3. Wireless / Networking
IBM Blue Gene
The world's most advanced network supercomputer from International Business Machines will tackle Grand Challenge problems
An Article by your Guide Bradley Mitchell
 Join the Discussion
"I was at an IBM seminar today, and one of the speakers asked the audience:

'Do you have any idea what the size of the Internet is these days? You know, you've got bytes, kilobytes, gigabytes, terabytes... and do you know what comes after that?

'Someone in the audience piped up - Trilobites!'"
-XAVIANA
 
 Related Resources
• Network Clustering
• Bandwidth and Latency
• Network Topologies
• Availability Concepts for Networks and Systems
 
 Elsewhere on the Web
• IBM Research - Blue Gene
 

Why Supercomputers are Fast

Several elements of a supercomputer contribute to its high level of performance:

  • Numerous high-performance processors (CPUs) for parallel processing
  • Specially-designed high-speed interconnects (internal networks)
  • Specially-designed or tuned operating systems

Supercomputer Processors

Supercomputers utilize either custom or mainstream commercial microprocessors. Small supercomputers may contain only a few dozen processors, but today's fastest supercomputers incorporate thousands of processors. The table below summarizes the processor configuration of today's top supercomputers.

SystemProcessor Configuration
ASCI Red9,472 Intel Pentium II Xeon
ASCI Blue Pacific5,856 IBM PowerPC 604E
ASCI White8,192 IBM Power3-II
NEC Earth Simulator5,104 NEC vector processors

Some supercomputer designs feature network co-processors. When sending and receiving data at the rate necessary for high-performance networking, it's common for a single processor to become heavily loaded with communications interrupts that take away too many cycles from primary computing tasks. To solve this problem, the IBM Blue Gene system will utilize cells. Each cell contains a primary processor, a network co-processor, and shared on-chip memory.

In total, the IBM Blue Gene system will contain one million custom IBM processors. So that the system will fit within a reasonably-sized room and not consume too much power, the processors are engineered so small that 32 of them will fit on a single microchip.

Supercomputer Interconnects

In order for a large number of processors to work together, supercomputers utilize specialized network interfaces. These interconnects support high bandwidth and very low latency communication.

Interconnects join nodes inside the supercomputer together. A node is a communication endpoint running one instance of the operating system. Nodes utilize one or several processors and different types of nodes can exist within the system. Compute nodes, for example, execute the processes and threads required for raw computation. I/O nodes handle the reading and writing of data to disks within the system. Service nodes and network nodes provide the user interface into the system and also network interfaces to the outside world. Special-purpose nodes improve overall performance by segregating the system workload with hardware and system software configured to best handle that workload.

Supercomputer nodes fit together into a network topology. Modern supercomputers have utilized several different specialized network topologies including hypercube, two-dimensional and three-dimensional mesh, and torus. Supercomputer network topologies can be either static (fixed) or dynamic (through the use of switches).

More on Supercomputer Interconnects

One of the most critical elements of supercomputer networking is routing. Supercomputers that utilize message passing require routing to ensure the individual pieces of a message are routed from source to destination through the topology without creating hotspots (bottlenecks). Advanced routing techniques like wormhole and virtual cut-through routing are employed by today's ASCI supercomputers.

Supercomputers utilize various network protocols. Application data communications generally take place at the physical and data link layers. I/O and communications with external networks utilize technologies like HIPPI, FDDI, and ATM as well as Ethernet.

Supercomputer interconnects involve large quantities of network cabling. These cables can be very difficult to install as they often must fit within small spaces. Supercomputers do not utilize wireless networking internally as the bandwidth and latency properties of wireless are not suitable for high-performance communications.

Supercomputer Operating Systems

Many supercomputers run multiple copies of a UNIX-based operating system. The ASCI White and Blue Pacific systems, for example, run IBM AIX. In the 1990s, research into high-performance network operating systems led to the development of so-called "lightweight" operating systems (O/Ses) that consist of a small, simple kernel with many of the capabilities of a general-purpose O/S removed. The ASCI Red system runs the PUMA O/S on its compute nodes.

Next page > Beyond Performance - RAS > Page 1, 2, 3, 4, 5


» Today's Commmunity Discussions
» More Feature Articles

Subscribe to the Newsletter
Name
Email

Explore Wireless / Networking
About.com Special Features

Stay connected and entertained with reviews on tips on the latest HDTVs, cellphones and more. More >

Easy ways to connect two computers for networking purposes. More >

  1. Home
  2. Computing & Technology
  3. Wireless / Networking

©2009 About.com, a part of The New York Times Company.

All rights reserved.