Skip navigation

AMD Business

3 Posts authored by: mark.papermaster Employee

At AMD, we are excited to celebrate Exascale Day along with Oak Ridge National Laboratory and Cray, a Hewlett Packard Enterprise Company, as our research and development teams are hard at work to change the world of computing with the groundbreaking Frontier supercomputer.

 

Frontier is expected to be the most powerful supercomputer of all time when it goes live, with an expected performance upwards of 1.5 exaFLOPS, or 1.5 billion, billion calculations per second. Powered by AMD EPYC™ CPUs, Radeon™ Instinct GPUs, Radeon Open eCosystem (ROCm) and EPYC open source software, Frontier targets more than five times faster performance than the world’s current reigning fastest supercomputer.

 

We are optimizing AMD Radeon Instinct GPUs and AMD EPYC CPUs in a 4:1 GPU to CPU configuration which will allow us to achieve high throughput of data. AMD’s Infinity Fabric will support high-speed connections between processors and allow Frontier to hit historic, sustained high-performance computation across the system.

 

As we approach and pass the barrier of exascale computing, the Frontier supercomputer opens up new possibilities for scientific research. Oak Ridge National Labs, Cray and AMD have created the Center for Accelerated Application Readiness (CAAR) program to develop applications designed for problems which only Frontier can help solve:

 

  •       Princeton University: to simulate future states of the Milky Way galaxy using massive amounts of satellite and telescope data in an astrophysical simulation code called Cholla.
  •       ORNL: to use a codebase known as Combinatorial Metrics (CoMet) to study the genetics of opioid addiction and toxicity, chronic pain, Alzheimer’s, and autism.
  •      Georgia Institute of Technology: to run GPUs for Extreme-Scale Turbulence Simulations (GESTS) to simulate turbulence with nearly 35 trillion grid points in order to better understand fluid turbulence as it relates to pollution, ocean currents and astrophysics.
  •      Virginia Polytechnic Institute and State University: to study the Lattice Boltzmann Methods of Porous Media (LBPM) code to understand the volumetric maps of mineral composition in order to train neural networks to predict future geometric configuration of fluids.
  •      ORNL: to run calculations of realistic condensed matters from first principles (FP) calculations, previously inaccessible before Frontier, through the Locally Self-Consistent Multiple Scattering (LSMS)
  •      University of Illinois at Urbana-Champaign: to use Frontier in conjunction with codebase Nanoscale Molecular Dynamics (NAMD) to understand viruses like Zika and pave the way for new drugs and vaccines to prevent future outbreaks.
  •      Michigan State University: to study complex-time dependent phenomena at the particle level such as nuclear reactions and fission through symmetry-projection techniques on a code called Nuclear Coupled-Cluster Oak Ridge (NuCCOR).
  •      University of Delaware: to develop advanced particle accelerators for radiation therapy of cancer, high energy physics, and photon science using code Particle-in-cell on Graphics Processing Units (PIConGPU).

 

This list inspires the work we do every day, as it takes the millions of hours of work that has gone into the latest AMD processors and brings it to life in tangible ways that will truly change the future. AMD is proud to be at the forefront of innovation and discovery through our collaboration with Cray and Oak Ridge National Laboratory. Working together with these exceptional technology partners and the researchers Frontier aims to empower, we can redefine the future of high-performance data centers and  have a profound effect on advancing science and technology.  

 

Cautionary Statement

This blog contains forward-looking statements concerning Advanced Micro Devices, Inc. (AMD) including, but not limited to, the expectations and benefits of the Frontier supercomputer, which are made pursuant to the Safe Harbor provisions of the Private Securities Litigation Reform Act of 1995. Forward-looking statements are commonly identified by words such as "would," "may," "expects," "believes," "plans," "intends," "projects" and other terms with similar meaning. Investors are cautioned that the forward-looking statements in this blog are based on current beliefs, assumptions and expectations, speak only as of the date of this blog and involve risks and uncertainties that could cause actual results to differ materially from current expectations. Such statements are subject to certain known and unknown risks and uncertainties, many of which are difficult to predict and generally beyond AMD's control, that could cause actual results and other future events to differ materially from those expressed in, or implied or projected by, the forward-looking information and statements. Investors are urged to review in detail the risks and uncertainties in AMD's Securities and Exchange Commission filings, including but not limited to AMD's Quarterly Report on Form 10-Q for the quarter ended June 29, 2019.

I’m pleased to announce AMD has joined other industry leaders in the Compute Express Link (CXL) Consortium. 

 

Compute Express Link (CXL) is an open industry standard interconnect offering high-bandwidth, low-latency connectivity between host processors, systems and devices such as accelerator cards, memory buffers, and smart I/O devices. Designed to address the increasing demands of high-performance computational workloads, CXL targets heterogeneous processing and memory systems across a range of high-performance computing applications by enabling coherency and memory semantics between processors and systems. This is increasingly important as processing data in Artificial Intelligence and Machine Learning requires a diverse mix of scalar, vector, matrix and spatial architectures across a range of accelerator options.

 

Since 2016 AMD has played a leadership role in driving three other new bus/interconnect standards, CCIX, OpenCAPI and Gen-Z.  Like CXL, these three efforts are driven by the need to create tighter coupling and coherency between processors and accelerators, and better exploit new and emerging memory/storage technologies in open, standards-based solutions.

 

While these different groups have been working to solve similar problems, each approach has its differences. As a long-standing supporter of open standards, we’re excited to join CXL and the possibilities presented as we work with other ecosystem leaders to address challenges we face as an industry.

 

If you’d like more information on CXL.  Please go to https://www.computeexpresslink.org/

 

MARK PAPERMASTER  

Executive Vice President and  Chief Technology Officer

Our strong momentum continues to grow! Last week, we celebrated a solid first quarter and our company’s 50th anniversary. Today , Lisa joined U.S. Department of Energy (DOE) Secretary Rick Perry, Oak Ridge National Laboratory (ORNL ) Director Thomas Zacharia and Cray CEO Peter Ungaro to announce that AMD EPYCTM CPUs, Radeon Instinct GPUs, and Radeon Open Ecosystem (ROCm) and EPYC open source software will power what is expected to be the world’s fastest supercomputer, called “Frontier.”

 

ORNL’s Frontier supercomputer aims to deliver more than 1.5 exaflops of performance. That level of performance exceeds the combined performance of today’s 160 fastest supercomputers! Scheduled to be online in 2021, the Frontier system will allow researchers to advance science in such applications as systems biology, materials science, energy production, additive manufacturing and health data science. Visit the Frontier website  to learn more about the areas of exploration.

 

AMD is delighted to partner with the DOE, Cray and ORNL to push the frontiers of HPC and lead the industry into the exascale era. This is one of the finest examples of a public and private partnership. AMD has a long history of working on HPC and is a strong partner for ORNL and DOE. In 2012, the DOE started a series of programs to push research that would enable future exascale compute systems. As part of this, AMD Research collaborated on FastForward, DesignForward, and PathForward – DOE programs that include public/private partnerships focused on key areas of chip, system, and software development to push the boundaries of high-performance computing. As the collaboration progressed, it evolved from research to product development, with exascale technologies incorporated into both our EPYC CPUs and Radeon Instinct GPUs for the datacenter. This partnership and shared vision around exascale computing drove our selection as the CPU and GPU provider for the Frontier project. We are incredibly excited to be a part of this program.

At the heart of the Frontier system is a future AI and HPC optimized AMD EPYC processor and a purpose-built AMD Radeon Instinct GPU in a 4:1 GPU to CPU configuration. The Radeon Instinct GPU also contains high-bandwidth memory allowing it to achieve very high throughput of data, while keeping power low and saving space. AMD’s Infinity Fabric will support high-speed connections between processors. The joint optimization of all these elements together enables us to hit this historic performance in each node of the system. Cray’s Slingshot scalable interconnect will handle the complex processing and communication of HPC and AI applications between nodes and between cabinets. The Frontier system will be supported by an enhanced version of AMD’s ROCm programming environment for AMD CPUs and GPUs.

 

Our strategy at AMD is about enabling high performance computing and we have an incredibly strong CPU and GPU roadmap, now more than ever optimized together. We believe supercomputing is at the forefront of the computer architecture innovation. Technology that is driving systems like Frontier is re-defining the standard for future high-performance datacenters.

 

Thank you to the thousands of dedicated employees who had a hand in this project. The science that the Frontier system enables will help solve some of the world’s toughest and most important challenges. This is another very proud day for AMD and demonstrates to the world that we have the technologies, vision and talent to lead the high-performance computing industry forward.

 

MARK PAPERMASTER  

Executive Vice President and  Chief Technology Officer

 

FORREST NORROD

Senior Vice President and General Manager

Datacenter and Embedded Solutions Group