AMD has demonstrated continued momentum in high performance computing, with NVIDIA’s announcement that 2nd Generation AMD EPYC™ 7742 processors will power their new DGX A100 dedicated AI and Machine Learning system. The new generation of AMD EPYC processors, along with AMD Radeon™ Instinct™ GPUs, will also power the world’s fastest two exascale-class supercomputers, Frontier and El Capitan.
AMD has demonstrated continued
momentum in high performance computing, with NVIDIA’s announcement that 2nd
Generation AMD EPYC™ 7742 processors will power their new DGX A100 dedicated AI
and Machine Learning system. The new generation of AMD EPYC processors, along
with AMD Radeon™ Instinct™ GPUs, will also power the world’s fastest two
exascale-class supercomputers, Frontier and El Capitan.
Each DGX A100 will feature dual 2nd
Gen AMD EPYC™ 7742 processors, for a total of 128 cores, running at 2.25 GHz
(base), 3.4 GHz (max boost). 2nd Gen AMD EPYC processors are
the first and only current x86-architecture server processor supporting PCIe
4.0, providing up to 128 lanes of I/O, per processor for high performance
computing and connections to other devices like GPUs.
In addition to the DGX-A100 systems,
AMD EPYC processors are powering the following supercomputers around the world:
- HLRS
‘HAWK’: The
High-Performance Computing Center Stuttgart (HLRS) selected DDN’s
EXAScaler with IME scale-out flash storage system for “Hawk,” its new
flagship supercomputer system. Hawk will empower scientists and engineers
to conduct research on larger and more complex phenomena while supporting
data-intensive workflows that combine big data analysis, deep learning and
simulation. Hawk is an HPE Apollo 9000 System and is among the fastest
supercomputers worldwide and the fastest general purpose system for
scientific and industrial computing in Europe. It is also supported by 2nd Gen
AMD EPYC processors. (Link)
- BIG
RED 200: Indiana
University acquired the fastest university-owned supercomputer in the
nation to support its advanced research in artificial intelligence,
machine learning, data analytics, and scientific and medical research, Big
Red 200. The system replaced the highly successful Big Red II Cray supercomputer
with speeds in excess of six quadrillion calculations per second. Big Red
200 is supported by 2nd Gen AMD EPYC processors and
next-generation NVIDIA Tensor Core GPUs. (Link)
- NEC
SX-AURORA TSUBASA: NEC was selected by the German weather forecasting service,
Deutscher Wetterdienst (DWD), to build the system valued at 50 million
Euro. Based on 2nd Gen AMD EPYC processors, the vector
supercomputer is expected to be six times faster than DWD's current
supercomputer, and more powerful than those used by the British and French
weather authorities. NEC SX-Aurora TSUBASA will support DWD with the
monitoring, analysis and prediction of meteorological and climatological
processes and drive sophisticated forecasts and warnings in case of severe
weather events. (Link)
- ETH
ZURICH EULER VI:
ETH Zurich is using 2nd Gen AMD EPYC 7742 processors in
its Euler VI system. (Link)
- DELL
EXPANSE: The
San Diego Supercomputer Center (SDSC) selected Dell Technologies to power
its next-generation supercomputer, Expanse. Expected to deploy in
mid-2020, Expanse will be powered by Dell EMC PowerEdge servers with 2nd Gen
AMD EPYC processors and is expected to deliver peak performance of five
petaflops. The system will enable tens of thousands of researchers to
speed time to discovery in fields of astronomy, molecular dynamics,
machine learning and artificial intelligence. (Link)
- CRAY
ARCHER2: Cray
and Hewlett Packard Enterprise (HPE) were awarded 48 million Euros to
expand UK Research and Innovation’s (UKRI) high-performance computing
capabilities with Cray’s next-generation Shasta supercomputer. The ARCHER2
supercomputer will be equipped with Slingshot interconnect, Cray
ClusterStor high-performance storage, the Cray Shasta Software platform,
and 2nd Gen AMD EPYC processors, giving it 11x higher
performance than its predecessor, ARCHER. ARCHER2 will be the first Shasta
system announced in EMEA and the second system worldwide used for academic
research. (Link)
- CRAY
VULCAN: United
Kingdom’s Atomic Weapons Establishment (AWE) selected Cray and Hewlett
Packard Enterprise (HPE) to support security and defense of the U.K. with
the Cray Shasta supercomputer. The Vulcan supercomputer will feature
a single Shasta supercomputer with expected performance of more than seven
petaflops. The architecture will include the Cray Slingshot interconnect,
2nd Gen AMD EPYC processors and Cray ClusterStor Lustre
storage. The high performance storage system will offer nearly 100
gigabytes per second of I/O performance. (Link)
- JOLIOT-CURIE: GENCI and Atos announced the
first operational supercomputer worldwide to include 2nd Gen
AMD EPYC 7H12 series processors. The Joliot-Curie supercomputer is
based on Atos’ BullSequana XH2000 solution and has a sustained revised
performance, thus far, of more than five petaflops. It also ranked No.59
on the TOP100 global ranking in 2019.(Link)
- ATOS
BULLSEQUANA XH2000 SUPERCOMPUTERS: Atos was selected to supply French
national meteorological service, Météo-France with two BullSequana XH2000
supercomputers based on 2nd Gen AMD EPYC processors. The
new systems will multiply Météo-France’s computing power by more than 5x,
compared to its current solution, enabling it to achieve several
scientific breakthroughs in weather forecasting. Each new supercomputer is
capable of processing more than 10 million billion operations per second
and are amongst the most powerful meteorological supercomputers in the
world. (Link)
- CRAY
HPC11: The Air
Force Life Cycle Management Center, in partnership with Oak Ridge National
Laboratory, tapped Cray to supply a Cray Shasta supercomputing system with
a contract valued at $25 million. The HPC11 supercomputer will enable
higher fidelity weather forecasts for U.S. Air Force and Army operations
worldwide. It will be one of the first supercomputing systems with 2nd Gen
AMD EPYC processors for use in operational weather forecasting. (Link)
COMMENTS