Intel has announced that it is now shipping its first Xeon Phi products utilising its Many Integrated Core (MIC) architecture to partners in the high-performance computing (HPC) and supercomputer segments, an attempt to challenge rivals Nvidia and Advanced Micro Devices with a powerful parallel processing engine of its own.
The first Intel Xeon Phi coprocessors, announced in June, will be made generally available on 28 January, 2013, the company said at the Supercomputing Conference (SC12) in Salt Lake City, Utah. The new chips, formerly codenamed Knight's Corner, are being packaged in discrete graphics-like cards and are built to work with Intel's Xeon E5 family of server and workstation CPUs.
"[The] Intel Xeon Phi coprocessor represents an achievement in Intel innovation that will help propel us to new heights in research and discovery, and reaffirms our commitment to Exascale-level computing," said Diane Bryant, vice president and general manager of Intel's Datacenter and Connected Systems Group.
"The combination of the Intel Xeon processor family and the Intel Xeon Phi coprocessor will change the scope and scale of what highly parallel applications can accomplish, by delivering unprecedented performance, efficiency and programmability. With this technology as a new foundation for HPC, solving real-world challenges from accurately predicting weather patterns 21 days in advance, to developing new cures for diseases will become increasingly possible," she added.
Intel introduced two Xeon Phi product lines at SC12. The Xeon Phi 5110P (pictured below), priced at $2,649 (£1,670), is already shipping to partners and will be officially released on 28 January, 2013. The Xeon Phi 3100 product family will be made available in the first half of 2013 carrying prices under $2,000 (£1,260) per coprocessor, Intel said.
The chip giant first offered a glimpse of Xeon Phi at the International Supercomputing Conference in Hamburg, Germany in June, saying it was working with Cray and others on a new family of powerful server and workstation products aimed at the supercomputing and HPC segments.
The Intel Xeon Phi line is the first to incorporate Intel's MIC architecture — which evolved from an attempt to develop an x86-based graphics processor code named Larrabee several years ago. Larrabee was considered a high-profile flop when Intel was unable to deliver GPUs based on the technology but the company stuck with the architecture, fine-tuning it and adapting it for highly parallel workloads.
Xeon Phi coprocessors first appeared in Cray's Cascade supercomputer and also play a part in seven systems on the latest supercomputing Top500 List, including the University of Tennessee's Beacon, which was labelled the most power-efficient supercomputer on the list with performance to the tune of 2.44 gigaflops per watt, while back at IDF 2012, Intel introduced Stampede – a new Petascale supercomputer built on Xeon E5 and Xeon Phi technology.
The new products are built on Intel's most advanced process technology, the 22-nanomter node utilising "3D" tri-gate transistors. The Xeon Phi 5110P, built for dense computing environments, achieves 1.01 teraflops (TFLOPS) of double-precision performance, supports 8GB of GDDR5 memory, and chugs along at 225 watts, Intel said. The forthcoming Xeon Phi 3100 family is more of a value proposition, offering greater than 1 TFLOP of double-precision performance, support for up to 6GB memory at 240GB/sec bandwidth, and thermals of 300 watts.
Partners with early access to Intel's new Xeon Phi products include the Texas Advanced Computing Center (TACC), as well as Professor Stephen Hawking and the Cosmos Lab at the University of Cambridge, where Xeon Phi coprocessors are currently being tested in the university's SGI supercomputer, Intel said.
"I am delighted that our new COSMOS supercomputer from SGI contains the latest many-core technology from Intel, the Intel Xeon Phi coprocessors. With our powerful and flexible SGI UV2000, we can continue to focus on discovery, leading worldwide efforts to advance the understanding of our universe," Hawking said in a statement.
Intel said that the "more than 50" OEM partners in the HPC and supercomputer segments currently building Xeon Phi systems include Acer, Appro, Asus, Bull, Colfax, Cray, Dell, Eurotech, Fujitsu, Hitachi, Hewlett-Packard, IBM, Inspur, NEC, Quanta, Supermicro, and Tyan.