AMD EPYC CPUs, AMD Radeon Instinct GPUs to power Cray Supercomputer
AMD and Cray unveil what is expected to be the world’s fastest supercomputer – defining a new standard for high-performance computing and pushing the technology boundaries of computational science to take the industry into the Exascale era.
The new system, called Frontier, is planned to come online in the U.S. in 2021 with over 1.5 exaflops of processing power. The total system contract award is valued at more than $600M USD for the system and technology development. Together with the Cray Shasta architecture, AMD is excited to build on leadership hardware and software with the singular vision of solving the toughest computing challenges in the world today. AMD innovations in Frontier include:
- High Performance Computing (HPC) and Artificial Intelligence (AI) optimized, custom AMD EPYC CPU and purpose-built Radeon Instinct GPU processors
- High Bandwidth Memory (HBM)
- Tightly integrated 4:1 GPU to CPU ratio
- Custom, high-speed coherent Infinity Fabric connection
- Enhanced, open ROCm programming environment for AMD CPUs and GPUs support.
With this announcement, AMD and Cray are driving a new HPC paradigm to support the complex compute, interconnect, software and storage requirements that Exascale computing demands.
SANTA CLARA, Calif. — May 7, 2019 — AMD (NASDAQ: AMD) today joined the U.S. Department of Energy (DOE), Oak Ridge National Laboratory (ORNL) and Cray Inc. in announcing what is expected to be the world’s fastest exascale-class supercomputer, scheduled to be delivered to ORNL in 2021. To deliver what is expected to be more than 1.5 exaflops of expected processing performance, the Frontier system is designed to use future generation High Performance Computing (HPC) and Artificial Intelligence (AI) optimized, custom AMD EPYC™ CPU, and AMD Radeon™ Instinct GPU processors. Researchers at ORNL will use the Frontier system’s unprecedented computing power and next generation AI techniques to simulate, model and advance understanding of the interactions underlying the science of weather, sub-atomic structures, genomics, physics, and other important scientific fields.
“AMD is proud to partner with Cray and ORNL to deliver what is expected to be the world’s most powerful supercomputer,” said Forrest Norrod, senior vice president and general manager, AMD Datacenter and Embedded Systems Group. “Frontier will feature custom CPU and GPU technology from AMD and represents the latest achievement on a long list of technology innovations AMD has contributed to the Department of Energy exascale programs.”
AMD innovations to be used in the Frontier system include:
- Future-generation High Performance Computing (HPC) and Artificial Intelligence (AI) optimized, custom AMD EPYC CPU, and Radeon Instinct GPU processors supported by High Bandwidth Memory (HBM) and extensive mixed precision ops for optimum deep learning performance;
- A custom high-bandwidth, low-latency coherent Infinity Fabric, connecting four AMD Radeon Instinct GPUs to one AMD EPYC CPU per node;
- An enhanced version of the open source ROCm programming environment, developed with Cray to tap into the combined performance of AMD CPUs and GPUs.
“We are excited to work with the team at AMD to deliver the Frontier system to Oak Ridge National Laboratory,” said Steve Scott, senior vice president and CTO at Cray. “Cray’s Shasta supercomputers are designed to support leading edge processor technologies and high-performance storage, all tightly interconnected by Cray’s new Slingshot network. The combination of Cray and AMD technology in the Frontier system will dramatically enhance performance at scale for AI, analytics, and simulation, enabling DOE to further push the boundaries of scientific discovery.”
AMD has a proud supercomputing history and a long-standing engagement with DOE, starting with the Jaguar supercomputer in 2005 and Titan supercomputer in 2011. The Frontier system leverages years of exascale technology investments by DOE. The contract award includes technology development funding, a center of excellence, several early-delivery systems, the main Frontier system and multi-year systems support.
“Frontier represents the state-of-the art in high-performance computing. Designing and standing up a machine of its scope requires working closely with industry, partnerships which not only enable breakthrough science but also ensure American scientific and economic competitiveness on the global stage,” said Jeff Nichols, associate laboratory director for Computing and Computational Sciences, ORNL. “We are delighted to work with AMD to integrate the CPU and GPU technologies that enable this extremely capable accelerated node architecture.”
Additional Resources
Follow AMD datacenter developments on Twitter @AMDServer
Rumor: AMD Epyc2 processors could get 64 cores over 8+1 dies - 10/31/2018 02:48 PM
That AMD has been going insanely strong with many-core processors is not a surprise, you've read all our Threadripper reviews and have learned that the top tier processors (e.g. 32-core versions) hav...
AMD and Oracle Collaborate to Provide AMD EPYC Processor-Based Offering in the Cloud - 10/24/2018 08:51 AM
Today at Oracle OpenWorld 2018, AMD (NASDAQ: AMD) announced the availability of the first AMD EPYC processor-based instance on Oracle Cloud Infrastructure. With this announcement, Oracle becomes the...
Gigabyte Adds New GPU & Storage Servers to AMD EPYC Server Line-Up - 08/02/2018 09:16 AM
Gigabyte is proud today to officially add three new single-socket GPU and Storage Servers to our AMD EPYC family: the 2U G291-Z20 and G221-Z30 GPU Servers, and the 4U S451-Z30 Storage Server....
Cray will supply CS500 clusters for supercomputers with AMD Epyc processors - 04/18/2018 04:21 PM
AMD has further strengthened its commitment to the HPC community by announcing that AMD EPYC processors have been added to the Cray CS500 Cluster Supercomputer product line....
Gigabyte refreshes AMD EPYC server line-up with more choices for NVMe storage - 04/11/2018 09:07 AM
Gigabyte refreshed its AMD EPYC 1U and 2U server line-up with a range updated options supporting different storage device combinations, with increased NVMe connectivity to integrate denser, high bandw...
Senior Member
Posts: 441
Joined: 2005-04-04
After Watching this about Milan and the possibility that it will include 15 Chiplets and maybe SMT4, I think I Have an Idea where AMD is going with it's future(Maybe Custom design?) HPC EPYC design on 7nm+:
1)Each CPU chiplet will be 6C/24T to save space/power while giving similar or better then 8c/16t performance.
2)Adding 4 custom Instinct GPU chiplets.
3)Adding 2 custom AI accelerator chiplets.
4)1 I/O chiplet with HBM memory stack.
So the final EPYC Milan(?) can be HPC beast with:
48C/192T Zen CPU cores.
4 custom Instinct GPUs.
2 AI accelerator Asics.
1 I/O Chiplet with HBM 3D staking .

EDIT: I see that there was already great article on such HPC APU design:
https://www.overclock.net/forum/225-...lops-200w.html
So after reading some of it I changed my illustration:

And 8 Milans could be installed in Cray’s Shasta 1U with Direct Liquid Cooling:

https://www.anandtech.com/show/13616/managing-16-rome-cpus-in-1u-crays-shasta-direct-liquid-cooling
Do you think that such chiplet design could be beneficial for HPC clients?
Senior Member
Posts: 8230
Joined: 2010-11-16
No bro. Doesn't work like that
According to the the USA exascale project strategy the DOE has put in writing the requirement that the project MUST NOT rely on the single best and the greatest hw, and instead it has to be built on several different architectures.
Is why you're seeing AMD GPU's there, being heavily subsidized by what seems to be a significant budget ramp up for exascale. Undoubtedly fueled by China's rising aspirations in the field and the whole bruhaha between the two countries.
That's why all of them (NV, intel and AMD) are getting a piece of the cake, regardless who has the best hw. Intel already fukd up once in delivery, that's why the need for redundancy and several contractors delivering different architectures.
Senior Member
Posts: 9633
Joined: 2006-10-29
Very interesting built , nice to see AMD chosen for such " Epyc " ( pun intended ) supercomputer