​IBM, Nvidia, U.S. Dept. of Energy outline supercomputing centers, collaboration

Two supercomputing centers of excellence revolve around making sure applications can take advantage of compute horsepower and gathering feedback from developers, engineers and scientists.

sierra.png

IBM will team up with Nvidia and the U.S. Department of Energy to launch two new supercomputer centers of excellence. These efforts primarily revolve around making sure applications can take advantage of supercomputing horsepower and gathering feedback from developers, engineers and scientists.

The centers, housed at the U.S. Department of Energy's Oak Ridge National Laboratory and the Lawrence Livermore National Laboratory will focus on advancing supercomputing and developing code innovation.

Under the partnership, outlined at the International Supercomputing Conference in Frankfurt, Germany, technology from IBM-led OpenPOWER effort, Nvidia and Mellanox will be showcased. IBM already has a supercomputing contract with the U.S. Department of Energy and is deploying two new supercomputer systems for operation in 2018.

Related: IBM launches SuperVessel, a free OpenPOWER cloud service

The IBM-Nvidia collaboration rhymes with a pact announced by HP and Intel. The goal is the same: Put HPC workloads and systems into more enterprises and use supercomputing deployments to showcase possibilities. IBM launched a POWER acceleration and design center in Montpellier, France and launched its SuperVessel service recently.

IBM said that the new HPC centers will also optimize research in areas such as energy, climate, biophysics and medicine before the two DoE supercomputer systems, dubbed Sierra and Summit, are deployed.

The IBM-Nvidia supercomputing centers will include technologies to ensure application readiness, collaboration between hardware engineers and application developers and designs that incorporate POWER processors from Big Blue and Nvidia's GPUs and interconnects. Teams at the centers will focus on programming models, algorithms, applications and compute performance.

Overall, the supercomputing systems being deployed will follow OpenPOWER designs with Nvidia and Mellanox technology.