Top 500 Highlights Supercomputing Trends
June 2nd, 2010 By: John Rath
The 25th annual International Supercomputing Conference opened Monday with a record 1,797 attendees. Held in Hamburg Germany the conference for high performance computing systems was represented by 50 countries and 151 exhibitors this year.
The Top 500 list was released Monday morning and the top ten had some changes. Most notable was China’s move to number two on the list with the Nebulae supercomputer, with a sustained computing speed of 1.271 Peta FLOPS (1 PetaFLOP is 1015 Floating point Operations Per Second). This is the first time Nebulae has been listed on the Top500. Housed in the National Supercomputing Centre in Shenzhen (NSCS) the Nebulae is a Dawning TC3600 blade system with Intel X5650 processors and a Nvida Tesla C2050 GPU. With120,640 cores the supercomputer tops #1 Jaguar with a theoretical performance of 2.98 petaFLOPS.
According to a Top500 sub-list for June 2010 there are 207 supercomputers utilizing Infiband. Data center connectivity solution provider Mellanox (MLNX) claimed Tuesday that adoption of its InfiniBand-connected CPU cores has grown 73 percent, representing more than 41 percent of the TOP500 list. “Advancements in multicore CPU architectures and the rise in GPU computing are placing greater demands on server and storage interconnects for higher bandwidth and lower application latencies,” said Jie Wu, Research Director, Technical Computing at IDC. Mellanox 40Gb/s InfiniBand connectivity products, including ConnectX-2 adapters and IS5600 switches with FabricIT fabric management software, provided the high-performance server and storage networking for the #2 Nebulae supercomputer.
On Tuesday Voltaire (VOLT) and Platform Computing announced a partnership to deliver optimized and automated management of data center resources, in virtualized and cloud computing environments. “Combining the intelligence of Voltaire’s UFM software with the dynamic provisioning capabilities of Platform LSF, enables customers to fully automate network configuration management and optimize the network. This meets the heavy needs of HPC data centers and cloud computing environments that may be running tens to hundreds of applications,” said Asaf Somekh, vice president of marketing at Voltaire.
Some other interesting highlights of the June 2010 Top500 list include:
- The entry level to the list moved up to the 24.7 teraFLOPS mark.
- Quad-core processor systems have saturated the list with 425 systems using them.
- The U.S. is the leading consumer of high performance computing systems with 282 of the 500, with Europe having 144 systems.
The Cray Jaguar supercomputer, owned by the Oak Ridge National Laboratory in Tennessee held on to the number one position with 1.75 petaFLOPS. Jaguar is a Cray XT5-HE with AMD Opteron six core 2.6 GHz processors. Those 224,162 cores reached a theoretical peak of 2.3 petaFLOPS. The Register has some additional information about the number one and two spots:
” Jaguar takes just under 4 watts to deliver a megaflops at a cost of $114 per megaflops for the iron, while Nebulae consumes 2 watts per megaflops at a cost of $39 per megaflops for the system.”
The BBC also reports on the supercomputers and shows how long it takes the average PC to match the performance of 1,750 trillion calculations per second.
The 500 fastest supercomputers listed on the June 2010 list combined are 32.4 petaFLOPS of computing power. The International Supercomputing Conference will run through June 3. InsideHPC has extensive coverage of the conference and has been reporting on a number of items coming from it. The conference web site also has video blogs and some live streaming.