The second phase of the build-out of HLRN-IV – the planned 16 peak-petaflops supercomputer serving the North-German Supercomputing Alliance (HLRN) – is underway at Zuse Institute Berlin (ZIB) and the University of Göttingen. The upgraded ZIB system Lise, named in honor of esteemed physicist Lise Meitner, is now operational.
Built in partnership with Atos and Intel, Lise consists of nearly 1,200 compute nodes powered by Intel Xeon Platinum 9242 48-core (Cascade Lake-AP) processors connected with Intel’s Omni-Path interconnect architecture. The system ranked 40th on the November 2019 Top500 list with 5.4 Linpack petaflops. A similar complex will be deployed at Göttingen in the first half of the year.
HLRN operates a massively parallel supercomputing system that serves over 200 universities and research institutions in Germany. The beefed-up system was designed to allow “researchers and scientists will be able to run compute-intensive projects involving mathematical simulations, machine learning and artificial intelligence (AI)” according to Intel which now has a brief Lise project profile on its site. Among areas of focus are:
- Earth System Sciences – Including climate, oceans, rain forests, glaciers, Antarctic phytoplankton (microalgae), mineral dust cycles and the stratosphere.
- Fluid Dynamics – Including turbulence models for ship turbines, wind turbines and aircraft wings.
- Healthcare – Including computer-aided drug design, improving medical care at home, and a better understanding of illness and treatment of diseases.
Interestingly, “Although we looked at accelerators, including GPUs, as part of the procurement process, there was no advantage with regards to obtaining the highest performance in using GPUs or other accelerators in the system,” according Thomas Steinke, head of ZIB Supercomputing.
Professor Alexander Reinefeld (ZIB) is quoted, “By entering into the second phase of our efforts with Atos and Intel, our users will benefit right away from the more powerful system without needing to change their code.”
He added, “We are expecting the highest performance for all benchmark applications. Our benchmark suite was carefully chosen so that each code challenges specific parts of the system: CPU, communication network, and parallel I/O. We are not looking for peak theoretical performance—we demand real system performance which makes it more complicated for vendors to optimize their infrastructure for our applications. That meant that our selection of the right processor and the right interconnect are all crucial for the overall performance.”
Intel, Atos and HLRN started work on the HLRN-IV Lise supercomputer in early 2019 and the system entered production in December. This year the organizations are kicking off the final phase of the HLRN-IV system, to be installed at Göttingen. According to HLRN’s website, the Göttingen upgrade will be finalized this June.
The phase-one HLRN-IV complex “Emmy” — comprising 448 Intel Skylake Gold 6148 CPU nodes — was deployed at Göttingen in late 2018.