Today, AMD (NASDAQ: AMD) showcased its high performance computing (HPC) leadership at ISC High Performance 2023 and celebrated, along with key partners, its first year of breaking the exascale barrier. AMD EPYC processors and AMD Instinct accelerators continue to be the solutions of choice behind many of the most innovative, green and powerful supercomputers in the world, powering 121 supercomputers on the latest TOP500 list.
“AMD’s mission in high-performance computing is to enable our customers to tackle the world’s most important challenges,” said Forrest Norrod, executive vice president and general manager, Data Center Solutions Business Group, AMD. “Our industry partners and the global HPC community continue to leverage the performance and efficiency of AMD EPYC processors and Instinct accelerators to advance their groundbreaking work and scientific discoveries.”
AMD Powered Frontier Moves into Full User Operations
The Frontier supercomputer at Oak Ridge National Laboratory, powered by AMD EPYC processors and AMD Instinct accelerators, remains the fastest computer in the world – for the third list in a row. Frontier’s performance increased to 1.194 exaflops, adding 92 petaflops of performance since the November 2022 ranking. The additional 92 petaflops would place eighth on the latest Top 500 list. As well, Frontier still remains one of the most energy efficient systems in the world. The Frontier Test and Development System is number two on the Green 500 with the full system placing sixth.
This year marks the one-year anniversary of the Frontier officially breaking the exascale barrier for the first time, a feat that remains unmatched. Since then, Frontier entered full user operations and is now supporting users from numerous scientific disciplines. A sample of studies currently using the power of the Frontier supercomputer include:
- Cancer Distributed Learning Environment (CANDLE) which seeks to develop predictive simulations that could help identify and streamline trials for promising cancer treatments, reducing years of lengthy, expensive clinical studies.
- WarpX which looks to simulate smaller, more versatile plasma-based particle accelerators and enable scientists to design particle accelerators for everything from radiation therapy to making semiconductor chips. The team’s work won the 2022 Gordon Bell Prize, which recognizes outstanding achievement in high-performance computing.
- ExaSky which plans to expand the size, scope and accuracy of simulations for complex cosmological phenomena such as dark matter to uncover new insights into the dynamics of the universe.
“WarpX is a first-of-its-kind application that enables 3D simulations of laser-plasma physics. As the team worked on the application as part of the DOE Exascale Computing Project, we knew that it needed to be optimized from the get-go to run on the world’s fastest supercomputer, Frontier,” said Axel Huebl, lead software architect of WarpX at Berkeley Lab. “What made our jobs easy was the fact that the GPU kernels that support the AMD Instinct accelerators and the ROCm open software platform are written in modern C++ with a fully open software stack. This and close iteration on early software and hardware previews gave the WarpX team the confidence and ease to get up and running quickly on Frontier, with fantastic results, that ultimately helped us win the Gordon Bell prize.”
Driving a Robust HPC and AI Ecosystem
AMD is also providing the hardware and software portfolio needed to meet the rapidly growing demand for AI applications within the HPC industry.
For hardware, the AMD Instinct MI250X and AMD EPYC processors take the top two spots in the latest HPL-MxP mixed-precision benchmark, which highlights the convergence of HPC and AI workloads, with the Frontier and LUMI systems. Frontier posted a score of 9.95 exaflops of mixed precision performance, adding two exaflops to its score from the previous list. While Lumi posted a score of 2.2 exaflops in the HPL-MxP benchmark.
Beyond hardware, AMD works with the HPC and AI community to drive the next wave of innovation and support new applications through open software. The AMD ROCm open software platform is utilized across multiple large scale HPC systems and has demonstrated stability, robustness and use at this massive scale. ROCm support of large AI models is also rapidly expanding on industry-leading TensorFlow and PyTorch frameworks to help users accelerate AI workloads on AMD Instinct hardware.
For example, the LUMI Supercomputer at CSC Finland is being used to train and power a GPT-3 large language model (LLM) with 13 billion parameters based entirely on Finnish. As well, LUMI is also being used by the Allen Institute for AI to train a new LLM for scientific discovery called Open Language Model (OLMo). OLMo is a uniquely open language model intended to benefit the research community by providing access and education around all aspects of model creation.
Accelerating Adoption of AMD-Processor Powered Systems
AMD demonstrated new collaborations that are advancing what’s possible for solving the world’s most complex problems. Systems leveraging AMD products include:
- A new supercomputer from Atos for the Max Planck Society powered by 4th Gen AMD EPYC processors and AMD Instinct MI300 accelerators for research around astrophysics, life science research, materials research, plasma physics and AI.
- The Adastra system, one of the fastest and greenest in Europe, managed by GENCI and housed at CINES was commissioned, installed and is now available to French and European researchers.
- LUMI, one of the fastest and most energy efficient supercomputers in the world. Leveraging AMD EPYC and AMD Instinct processors, LUMI is being used to power new research around climate change and cancer.
- A new supercomputer launched in April 2023 from Oak Ridge National Laboratory (ORNL) and the U.S. National Oceanic and Atmospheric Administration dedicated to climate science research.
- Nautilus, a new supercomputer from the U.S. Navy DoD Supercomputing Resource Center will support development of weather and ocean products is powered by AMD EPYC CPUs.
Enabling More Energy Efficient Computing
In HPC, energy efficiency is a top priority for the industry to meet sustainability commitments and enable the next generation of supercomputers. AMD is driving innovation and setting new industry standards around energy efficiency. On the latest Green 500 list, AMD powers seven of the top 10 systems on the including spots two through seven with AMD EPYC CPUs and AMD Instinct Accelerators.
A highlight of energy efficiency is the AMD powered LUMI supercomputer which is the both the fastest supercomputer in Europe and one of the most efficient. LUMI also takes this the next step by running on 100 percent carbon neutral hydroelectric power and capturing waste heat to power the surrounding city.