Intel is bulking up its supercomputing capabilities by buying Whamcloud, a startup that is a proponent of the open-source Lustre parallel distributed file system.
Intel bought the company to help in its push toward exascale computing, according to Brent Gorda, former president and CEO of Whamcloud and now general manager of Intels High Performance Data Division.
The Whamcloud acquisition extends Intels software and service portfolio in the high-performance computing space in addition to reinforcing Intels position in the open-source community, Gorda said in a July 13 blog post. Working as one company, we are now in a stronger position to advance our mutual goals and continue providing vendor-neutral solutions, delivering greater value to our customers and moving the industry to exascale performance.
Whamcloud was formed in 2010 by a group of Lustre engineers soon after Oracle bought Sun Microsystems. Sun had overseen Lustre development for several years, but after the acquisition, Oracle officials seemed to have little interest in pushing the open-source technology forward, though it still supports commercial deployments.
Lustre, a massively parallel file system, can scale to tens of thousands of compute notes and petabytes of storage, and is used in large-scale cluster-computing environments. It was first developed at Carnegie Mellon University in Pittsburgh.
IBMs Sequoia supercomputer at the Lawrence Livermore National Laboratory, a Blue Gene/Q system with more than 1.5 million Power cores, runs the Lustre file system. Sequoia in June was named the fastest supercomputer in the world, placing first on the Top500 list that was released at the ISC 2012 supercomputing show with a performance of 16.32 petaflops (quadrillion floating-point operations per second).
Intel officials have said that the high-performance computing (HPC) space is an increasingly important one for the company, promising as much as 20 percent annual growth for Intel.
This projects to be a major growth segment for the business, Rajeeb Hazra, vice president of the Intel Architecture Group and general manager of technical computing at the company, said in a June conference call with analysts and journalists before the chip maker announced it was branding its Many Integrated Cores (MIC) coprocessors as Xeon Phi.
Xeon Phi is a key effort in the larger HPC strategy at Intel, which already has a strong presence in the industry. According to the June Top500 list, Intel chips ran more than 74 percent of the worlds 500 fastest supercomputers, including four of the top 10.
In March, Intel released its Xeon E5-2600 processors, which offer up to eight cores and 80 percent more performance than the previous generation. They are also 50 more energy-efficient, and Intel officials have said the HPC space will be a key market for them. Forty-four of the supercomputers on the Top500 list run Xeon E5-2600 processors.
The company also has taken other steps to build up its supercomputing capabilities, including buying QLogics InfiniBand products and supercomputer maker Crays networking assets.
On July 13, Intel was one of several chip makers to receive money from the Department of Energy as part of its Extreme-Scale Computing Research and Development “FastForward,” which was created to help drive advancements in exascale computing. Intel netted $19 million in subcontracts, while rival Advanced Micro Devices received $12.6 million and Nvidia $12.4 million.