Singapore’s National Supercomputing Center has announced that it will use AMD’s EPYC Milan and NVIDIA’s A100 GPUs to power their next 10 PFLOP supercomputers.
Singapore NSCC supercomputer powered by 100,000 AMD EPYC Milan processor cores and NVIDIA A100 GPUs will be packed with 10 PFLOPs of computing power!
The supercomputer will be used by the National Supercomputing Center (NSCC) for research institutes, government agencies and businesses, reports The Register. The supercomputer received a $ 200 million investment directly from the Singapore government (March 2019).
The next-generation national supercomputer for Singapore will be a hot-water-cooled green system – one of the first known deployments of such a system in a tropical environment. When operational, the supercomputer is expected to deliver a total of 10 PFLOPS of raw computing power and is eight times more powerful than the current ASPIRE1 supercomputer.
ASPIRE1, which went live in 2016, is running near full capacity to support local advanced search that requires high-end computing resources. The new system is the first in a series of supercomputers that will be phased in by 2025 to expand and upgrade Singapore’s high performance computing (HPC) capabilities for the research community here.
The new system, which will be 8 times faster than NSCC’s existing pool of HPC resources, will expand and augment ongoing research efforts by enabling tools such as artificial intelligence (AI) and deep machine learning. optimize modeling, simulation and even software simulation for quantum. computing.
The Supercomputer itself will be based on the latest CPU / GPU architectures from NVIDIA and AMD. It will be based on the HPE Cray EX product line and will feature a total of 900 compute nodes. The exact system has over 100,000 cores based on AMD’s Zen 3 core architecture. It includes the AMD EPYC 7003 “Milan” line of server processors that feature up to 64 cores and clock speeds up to nearly 3.70 GHz.
NSCC’s supercomputer will also ship 352 NVIDIA A100 Tensor Core GPUs which add an additional quarter of a million GPU cores to the system. We know that the Tensor A100 offers maximum computing power of 19.5 TFLOP, so the system gets at least 6.8 PFLOP of computing power from GPUs alone, while the rest of the 3.2 PFLOPs from computing power is driven by the AMD 3rd Gen EPYC Milan Fries. The supercomputer will also include several 100 Gbit / s links in order to be able to work with SingAREN-Lightwave Internet Exchange of the Singapore Advanced Research and Education Network.
There is no official name for the supercomputer yet, but it is expected to be commissioned and operational in early 2022.