Jaquar chases Roadrunner for fastest supercomputer

Summary:Two systems, including one based on x86 chips, have become the first ever to break the petaflop/s barrier in the latest Top500 supercomputer list.

Cray's XT5 Jaguar supercomputer has narrowly missed displacing IBM's Roadrunner system as the world's fastest supercomputer, according to the results of the bi-annual 'Top500' supercomputing list announced on Friday.

Roadrunner (left), which is located at the Los Alamos National Laboratory, was enhanced earlier this year and in June became the first to break the petaflop barrier, reaching 1.105 petaflops, according to Top500.

Jaguar, which operates at the Oak Ridge National Laboratory, became only the second-ever to pass the petaflop mark last week, as the result of a $100m (£70m) upgrade. However, Top500 officials placed Jaguar's performance at 1.059 petaflops, narrowly trailing that of Roadrunner.

One petaflop represents one quadrillion floating point operations per second.

AMD pointed out that Jaguar is based on its quad-core Opteron processors, making Jaguar the first entirely x86-based system to break the petaflop barrier. Roadrunner is based on IBM QS22 blades, which are powered by the PowerXCell 8i, an advanced version of the chip found in the Sony PlayStation 3.

Oak Ridge and Los Alamos are both operated by the US Department of Energy, which runs seven of the top 10 systems on the list. Nine of the top 10 systems are located in the US. The most powerful system outside the US is China's Dawning 5000A, which is also the largest system based on Windows HPC 2008.

IBM and AMD may lead the list, but about three-quarters (75.8 percent) of the other systems on the list are based on Intel chips, slightly up from 75 percent on the last list, issued in May.

IBM Power and AMD Opteron chips nearly tied for the second most-used processors, with 60 and 59 systems each respectively, or 12 percent and 11.8 percent. Quad-core chips have taken over the Top500 list rapidly, and are now to be found in 336 systems. Seven of the list's systems are using nine-core PowerXCell chips.

HP built the highest number of systems, at 209, followed by IBM with 188.

The entry-level of the list increased to 12.64 teraflops from 9.0 teraflops in May, and the 500th system on the current list would have placed at 267 on May's list, the Top500 organization said.

The US accounts for the majority of the Top500-listed supercomputing systems, with 291 systems, up from 257. Europe follows with 151 systems, followed by Asia with 47 systems. Within Europe the UK leads, accounting for 45 systems on the list, followed by Germany with 24 systems. Within Asia, Japan leads with 18 systems, followed by China with 16 and India with eight.

The Top500 list is issued twice a year and is compiled by the University of Mannheim, the NERSC/Lawrence Berkeley National Laboratory and the University of Tennessee, Knoxville.

Topics: Processors, Hardware, Networking

Kick off your day with ZDNet's daily email newsletter. It's the freshest tech news and opinion, served hot. Get it.

Related Stories

The best of ZDNet, delivered

You have been successfully signed up. To sign up for more newsletters or to manage your account, visit the Newsletter Subscription Center.
Subscription failed.