AMD EPYC™ Processors and New AMD Instinct™ MI100 Accelerator Redefine Performance for HPC and Scientific Research

AMD EPYC™ Processors and New AMD Instinct™ MI100 Accelerator Redefine Performance for HPC and Scientific Research

Support for next generation AMD EPYC processors, codenamed ‘Milan,’ in new HBv3 Digital Machines introduced by Microsoft

AMD Instinct™ MI100 accelerator is first accelerator to use new AMD CDNA architecture dedicated to superior-performance computing (HPC) workloads

SANTA CLARA, Calif., Nov. sixteen, 2020 (Globe NEWSWIRE) — Throughout this year’s SC20 digital tradeshow, AMD (NASDAQ: AMD) is showcasing its leadership in the significant effectiveness computing (HPC) field. It released the new AMD Instinct™ MI100 accelerator with ROCm™ 4. open up ecosystem assist and showcased a expanding checklist of AMD EPYC™ CPU and AMD Intuition accelerator based deployments, and highlighted its collaboration with Microsoft Azure for HPC in the cloud. AMD also remains on keep track of to start out volume shipments of the 3rd Gen EPYC processors with “Zen 3” main to select HPC and cloud consumers this quarter in progress of the envisioned public start in Q1 2021, aligned with OEM availability.

The new AMD InstinctMI100 accelerator, is the world’s quickest HPC GPU accelerator for scientific workloads and the first to surpass the 10 teraflops (FP64) efficiency barrier one. Created on the new AMD CDNA architecture, the AMD Instinct MI100 GPU allows a new course of accelerated techniques for HPC and AI when paired with twond Gen AMD EPYC processors. Supported by new accelerated compute platforms from Dell, HPE, Gigabyte and Supermicro, the MI100, put together with AMD EPYC CPUs and ROCm 4. program, is created to propel new discoveries ahead of the exascale era.

“No two clients are the exact in HPC, and AMD is giving a path to today’s most superior systems and abilities that are vital to aid their HPC perform, from modest clusters on premise, to virtual devices in the cloud, all the way to exascale supercomputers,” said Forrest Norrod, senior vice president and common supervisor, Details Heart and Embedded Alternatives Company Group, AMD. “Combining AMD EPYC processors and Intuition accelerators with critical application program and advancement equipment enables AMD to provide leadership general performance for HPC workloads.”

AMD and Microsoft Azure Ability HPC In the Cloud

Azure is utilizing 2nd Gen AMD EPYC processors to electric power its HBv2 digital devices (VMs) for HPC workloads. These VMs offer you up to 2x the performance of initial-generation HB-series digital equipmenttwo, can help up to 80,000 cores for MPI work3, and acquire advantage of twond Gen AMD EPYC processors’ up to forty five% more memory bandwidth than similar x86 alternate optionsfour.

HBv2 VMs are utilized by a lot of prospects which include The College of Illinois at Urbana-Champaign’s Beckman Institute for Sophisticated Science & Technology which applied 86,four hundred cores to product a plant virus that previously necessary a management course supercomputer and the U.S. Navy which fast deploys and scales enhanced weather and ocean pattern predictions on need. HBv2 driven by twond Gen AMD EPYC processors also provides the bulk of the CPU compute power for the OpenAI environment Microsoft declared before this calendar year.

AMD EPYC processors have also helped HBv2 get to new cloud HPC milestones, these types of as a new report for Cloud MPI scaling final results with NAMD, Top rated twenty benefits on the Graph500, and the initial 1 terabyte/sec cloud HPC parallel filesystem. Throughout these and other application benchmarks, HBv2 is offering 12x better scaling than uncovered somewhere else on the public cloud.

Introducing on to its existing HBv2 HPC digital device powered by 2nd Gen AMD EPYC processors, Azure introduced it will make the most of subsequent generation AMD EPYC processors, codenamed ‘Milan’, for long term HB-sequence VM goods for HPC.

You can see far more about the AMD and Azure collaboration in this video clip with Jason Zander of Azure and Lisa Su of AMD.

AMD Proceeds to Be the Selection for HPC

AMD EPYC processors and Instinct accelerators have the functionality and capabilities to guidance quite a few HPC workloads across a assortment of implementations. From small clusters at analysis centers, to commercial HPC, to off premise and in the cloud, to exascale computing, AMD proceeds to provide general performance and choice for HPC alternatives.

Hewlett Packard Organization (HPE), CSC Finland and EuroHPC recently released a new pre-exascale procedure, LUMI. Dependent on the HPE Cray EX supercomputer architecture, LUMI will use next era AMD EPYC CPUs and Intuition accelerators and is predicted to offer a peak effectiveness of 552 petaflops when it comes on-line in 2021, building it 1 of the speediest supercomputers in the planet.

Over and above LUMI, AMD powered HPC units continue on to mature in volume. Because SC19, there have been additional than fifteen supercomputing units declared working with AMD EPYC CPUs, Instinct GPUs, or both. A emphasize of the units includes

  • Chicoma – Los Alamos Nationwide Laboratory – this process is based mostly on the HPE Cray EX supercomputer architecture and utilizes twond Gen AMD EPYC CPUs, mixed with 300 terabytes of program memory for COVID-19 analysis,
  • Corona – Lawrence Livermore National Laboratory – this program was not too long ago upgraded with funding from the Coronavirus Aid, Reduction and Financial Protection (CARES) Act, adding approximately one,000 AMD Instinct MI50 accelerators, pushing peak general performance to more than eleven petaFLOPS,
  • Mammoth – Lawrence Livermore Nationwide Laboratory – the “big memory” cluster employs twond Gen AMD EPYC Processors to complete genomics investigation and graph analytics needed by experts functioning on COVID-19.
  • Northern Info – a distributed computing method in Europe that is working with AMD EPYC CPUs and Instinct accelerators for massive scale HPC purposes such as rendering, artificial intelligence and deep mastering,
  • Pawsey Supercomputing Centre – Working with the HPE Cray EX supercomputer architecture and long run AMD EPYC CPUs and AMD Instinct accelerators, the supercomputer at Pawsey will be Australia’s most highly effective supercomputer.

In addition, AMD is also powering the subsequent supercomputers: Anvil and Bell – Purdue College, Major Red two hundred – Indiana College, Bridges 2 – Pittsburgh Supercomputing Centre, CERN, European Centre for Medium-Variety Weather Forecasts, Expanse – San Diego Supercomputer Centre, Goethe University Frankfurt, IT4Innovations Nationwide Supercomputing Center, Jetstream 2 – Indiana University, Mahti – CSC, Mangi – University of Minnesota, National Oceanic and Atmospheric Administration, Crimson Raider – Texas Tech College, TinkerCliffs – Virginia Tech.

“With the Expanse supercomputer, our purpose is to give scientists and scientists cloud-like access to a substantial-general performance machine that can handle every little thing from astrophysics to zoology,” stated Michael Norman, Director of the San Diego Supercomputer Middle. “The 2nd Gen AMD EPYC processors have assisted us accomplish wonderful functionality with Expanse, enabling our researchers to do far more science than before. We also have a excellent collaboration with AMD and have worked together to generate a discussion board for AMD HPC shoppers to share ordeals, information and far more, to greater gain HPC study.”

Paving the Route to Exascale Computing

To support researchers start off on the path to exascale, AMD has offered Oak Ridge Nationwide Labs entry to the new AMD Intuition MI100 accelerator, which delivers a huge leap in compute and interconnect efficiency. The Intuition MI100 accelerator permits a new class of accelerated systems and provides real heterogeneous compute capabilities from AMD for HPC and AI. Made to enhance the 2nd Gen AMD EPYC processors, and designed on the AMD Infinity Architecture, the AMD Intuition MI100 provides real heterogeneous compute abilities from AMD for HPC and AI.

“Frontier, powered by AMD, represents a massive boost in computational ability in contrast to today’s systems. It’s likely to allow scientists to answer queries that we failed to have the answer to in advance of,” reported Bronson Messer, director of science, Oak Ridge Management Computing Facility. “The means to operate molecular simulations that are not just a several million atoms, but a several billion atoms, provides a additional reasonable representation of the science, and to be equipped to do that as a subject of class and more than and in excess of once again will lead to a important total of critical discoveries.”

AMD carries on to give the general performance, abilities and scale needed to energy latest and future HPC workloads, no make any difference if they are encouraging pupils at a study middle, improving upon aerodynamic efficiency for an auto manufacturer, or furnishing useful insights for crucial health-related breakthroughs. Study extra about the AMD presence at SC20 and its HPC capabilities here.

Supporting Methods

About AMD
For a lot more than 50 a long time AMD has pushed innovation in significant-efficiency computing, graphics and visualization systems ― the creating blocks for gaming, immersive platforms and the facts center. Hundreds of thousands and thousands of buyers, primary Fortune 500 enterprises and slicing-edge scientific research amenities about the environment depend on AMD engineering day by day to make improvements to how they dwell, get the job done and play. AMD staff members all over the world are concentrated on constructing great goods that press the boundaries of what is probable. For more details about how AMD is enabling nowadays and inspiring tomorrow, visit the AMD (NASDAQ: AMD) websiteblogFacebook and Twitter pages.


This press release is made up of ahead-hunting statements relating to Innovative Micro Devices, Inc. (AMD) such as the features, features, efficiency, availability, timing and anticipated rewards of AMD merchandise together with the next generation AMD EPYCTM Processors and AMD IntuitionTM MI100 Accelerator and the predicted timing and positive aspects of AMD partnerships, which are manufactured pursuant to the Secure Harbor provisions of the Non-public Securities Litigation Reform Act of 1995. Ahead-wanting statements are usually identified by words these as “would,” “may,” “expects,” “believes,” “designs,” “intends,” “assignments” and other terms with similar indicating. Traders are cautioned that the ahead-searching statements in this press release are based mostly on present-day beliefs, assumptions and expectations, converse only as of the day of this push launch and involve hazards and uncertainties that could induce genuine effects to vary materially from recent anticipations. This sort of statements are subject to selected known and unidentified hazards and uncertainties, quite a few of which are hard to predict and normally over and above AMD’s handle, that could cause true final results and other future situations to differ materially from individuals expressed in, or implied or projected by, the forward-on the lookout info and statements. Content factors that could induce genuine results to vary materially from present-day expectations contain, with no limitation, the adhering to: Intel Corporation’s dominance of the microprocessor marketplace and its aggressive small business techniques the means of third celebration companies to manufacture AMD’s merchandise on a timely foundation in ample portions and working with aggressive systems anticipated production yields for AMD’s items the availability of essential gear, resources or production procedures AMD’s skill to introduce products and solutions on a well timed foundation with features and general performance levels that provide worth to its consumers world financial uncertainty the decline of a considerable shopper AMD’s capability to make earnings from its semi-custom made SoC products  the impact of the COVID-19 pandemic on AMD’s business, monetary issue and benefits of operations political, legal, financial pitfalls and normal disasters the impression of governing administration actions and laws such as export administration restrictions, tariffs and trade security steps the influence of acquisitions, joint ventures and/or investments on AMD’s small business, together with the announced acquisition of Xilinx, and the failure to integrate obtained companies AMD’s means to total the Xilinx merger the effect of the announcement and pendency of the Xilinx merger on AMD’s organization opportunity protection vulnerabilities possible IT outages, data decline, facts breaches and cyber-attacks uncertainties involving the purchasing and cargo of AMD’s products quarterly and seasonal product sales styles the limitations imposed by agreements governing AMD’s notes and the revolving credit rating facility the aggressive markets in which AMD’s merchandise are bought market place circumstances of the industries in which AMD products are offered AMD’s reliance on 3rd-social gathering intellectual home to layout and introduce new solutions in a well timed method AMD’s reliance on third-bash businesses for the design and style, manufacture and supply of motherboards, application and other computer system platform factors AMD’s reliance on Microsoft Corporation and other software program vendors’ assistance to design and style and develop computer software to run on AMD’s goods AMD’s reliance on 3rd-social gathering distributors and insert-in-board partners the potential dilutive result if the 2.a hundred twenty five% Convertible Senior Notes because of 2026 are converted foreseeable future impairments of goodwill and technological innovation license buys AMD’s skill to draw in and keep qualified staff AMD’s capability to crank out enough income and working cash stream or receive exterior financing for analysis and development or other strategic investments AMD’s indebtedness AMD’s means to deliver sufficient funds to services its debt obligations or fulfill its operating cash demands AMD’s potential to repurchase its exceptional financial debt in the celebration of a improve of control the cyclical mother nature of the semiconductor business the affect of modification or interruption of AMD’s internal business enterprise procedures and details techniques compatibility of AMD’s merchandise with some or all marketplace-standard computer software and components costs linked to defective products and solutions the performance of AMD’s offer chain AMD’s capability to rely on 3rd occasion offer-chain logistics capabilities AMD’s stock rate volatility globally political conditions unfavorable currency exchange fee fluctuations AMD’s means to effectively regulate the product sales of its merchandise on the grey market AMD’s skill to sufficiently safeguard its technological know-how or other mental property recent and future claims and litigation probable tax liabilities and the effects of environmental regulations, conflict minerals-related provisions and other legal guidelines or polices. Traders are urged to critique in depth the hazards and uncertainties in AMD’s Securities and Exchange Commission filings, which include but not minimal to AMD’s Quarterly Report on Type 10-Q for the quarter finished September 26, 2020.

©2020 State-of-the-art Micro Devices, Inc. All rights reserved. AMD, the AMD Arrow emblem, EPYC, AMD Instinct, and combinations thereof are logos of Superior Micro Equipment, Inc. PCIe is a registered trademark of PCI-SIG Company. Other merchandise names applied in this publication are for identification reasons only and may be logos of their respective organizations.


one Calculations executed by AMD Performance Labs as of Sep 18, 2020 for the AMD Instinct™ MI100 (32GB HBM2 PCIe® card) accelerator at 1,502 MHz peak improve motor clock resulted in eleven.fifty four TFLOPS peak double precision (FP64), forty TFLOPS peak solitary precision matrix (FP32), 23.1 TFLOPS peak solitary precision (FP32), 184.six TFLOPS peak fifty percent precision (FP16) peak theoretical, floating-place efficiency. Released outcomes on the NVidia Ampere A100 (40GB) GPU accelerator resulted in TFLOPS peak double precision (FP64). 19.five TFLOPS peak single precision (FP32), 78 TFLOPS peak 50 percent precision (FP16) theoretical, floating-position efficiency. Server producers may perhaps differ configuration choices yielding distinct benefits. MI100-03
2 Source:
3 Supply: site/azure-hbv2-virtual-devices-eclipse-80000-cores-for-mpi-hpc/
four AMD EPYC™ 7002 Series processors have forty five% more memory bandwidth than Intel Scalable processors in the similar class.

Gary Silcott
AMD Communications
+1 512-602-0889

Laura Graves
AMD Investor Relations
+one 408-749-5467

Major Emblem

Supply: Sophisticated Micro Equipment, Inc.

Add Comment

Your email address will not be published. Required fields are marked *