Home > News content

AMD: 64 core second generation Xiaolong fully supports NVIDIA Amps

via:驱动之家     time:2020/5/19 13:49:07     readed:583

NVIDIA unveiled a new GPU architectureAI Artificial Intelligence and ml machine learning system DGX A100, A100 GPU with up to eight amp architectures, and two amd 64 core Xiaolong processors in a very low profile.

At one time, we guessed that NVIDIA uses Xiaolong 7H12, 64 core 128 threads, frequency 2.6-3.3ghz, and thermal design power consumption 280W.

AMD:64核心二代霄龙全力支持NVIDIA安培

Today, the AMD official revealed the answer to two of the top 7742, the same 64 core 128 threads, the reference frequency is slightly lower than 2.25 GHz, but the maximum acceleration can reach 3.4 GHz, while thermal design power control at 225 W.

Amd notice emphasizes,The second generation of Xiaolong processor is the world's first and only x86 architecture server processor supporting PCIe 4.0 busEach processor can provide up to 128 PCIe 4.0 channels to support high-performance computing and high-speed connection with other devices such as GPU.

According to the data, NVIDIA DGX A100 system is also equipped with PCIe 4.0 nvme SSD with 15tb capacity, but the specific brand and product information is still unclear.

With its superior specifications and performance, and relatively low price, AMD's second generation Xiaolong is making great strides in many high-performance computing systems, especially Xiaolong 7742, which is a star product, especiallyThe world's two super fast 10 billion secondary supercomputers, frontier and EI captian, have the figure of the second generation Xiao Long, as well as the AMD radeon insight acceleration card.

AMD:64核心二代霄龙全力支持NVIDIA安培

Other supercomputers of AMD's second generation Xiaolong include:

HLRS -

Stuttgart High Performance Computing Centre (HLRS) has selected DDN EXAScaler and its IME lateral extended flash memory system to support the new flagship supercomputer system

Hawk enables scientists and engineers to study larger and more complex phenomena, and supports data intensive workflow combining big data analysis, deep learning and simulation.

Hawk belongs to HPE Apollo 9000 system. It is one of the super fast supercomputers in the world and one of the super fast general systems used in scientific and industrial computing in Europe.

- BIG RED 200:

Indiana University has deployed the top supercomputer, big red 200, to support advanced research in artificial intelligence, machine learning, data analysis, science and medical research. This system has replaced the very successful big red 200 II Cray supercomputer, with an operation speed of more than 600 million times per second. In addition to AMD's second generation Xiaolong, it is also equipped with NVIDIA amp architecture tensor core GPU.

- NEC SX-AURORA TSUBASA:

Deutsche wetterdienst (DWD), a German weather forecast service company, chose NEC to build the 50 million euro vector supercomputer, which is expected to be six times faster than DWD's current supercomputer and more powerful than the supercomputers used by the British and French meteorological bureaus.

NEC SX Aurora Tsubasa supports the monitoring, analysis and prediction of meteorological and climate change processes, and provides complex prediction and early warning services in case of severe climate events.

- ETH ZURICH EULER VI:

ETH Zurich uses the second generation AMD epyc 7742 processor in its Euler VI system.

- DELL EXPANSE:

The San Diego Supercomputer Center (SDSC) chose Dell to build the next generation supercomputer expansion. It is expected to deploy in the middle of 2020, which is expected to bring the highest performance of 5 petaflops (billions of floating-point operations per second).

The system will help thousands of researchers in astronomy, molecular dynamics, machine learning and artificial intelligence to speed up the process of exploration and discovery.

- CRAY ARCHER2:

Cray and HPE received 48 million euros to expand ukri's high-performance computing capabilities with Cray's next-generation Shasta supercomputer.

Archer2 supercomputer will be equipped with slingshot interconnection, Cray clusterstor high-performance storage, Cray Shasta software platform and the second generation AMD epyc processor, and the performance will be 11 times that of the former archer.

Archer2 will be the first Shasta system announced in EMEA and the second in the world for academic research.

- CRAY VULCAN:

The British Atomic Weapons Research Institute (AWE) chooses Cray and HPE to maintain the national defense security of the UK with the help of Cray Shasta supercomputer.

Awe Vulcan supercomputer will be equipped with a Shasta supercomputer, whose performance is expected to exceed 7pflops. Its architecture includes Cray slingshot interconnection, second generation AMD epyc processor, Cray clusterstor luster storage, and I / O performance is close to 100GB / s.

- JOLIOT-CURIE:

French National High Performance Computing Organization (genci) and ATOS announced the world's first operational supercomputer Joliot Curie. Based on ATOS bullsequana xh2000 solution, its performance has been continuously improved. So far, its running speed has exceeded 5pflops, ranking 59th in the world.

ATOS BULLSEQUANA XH2000:-

French Meteorological Service (M)

Each new supercomputer can handle more than 100 million operations per second. It is one of the most powerful supercomputers for meteorological research in the world.

CRAY HPC11:-

The United States Air Force life cycle management center (aflcmc) and Oak Ridge National Laboratory (ORNL) purchased Cray hpc11, a Cray Shasta supercomputing system provided by Cray, with a contract value of $25 million.

Hpc11 supercomputers will provide more accurate weather forecasts for U.S. Air Force and army operations around the world. HPC11 will be one of the first supercomputing systems equipped with the second generation AMD EPYC processor for military weather forecast.

AMD:64核心二代霄龙全力支持NVIDIA安培

China IT News APP

Download China IT News APP

Please rate this news

The average score will be displayed after you score.

Post comment

Do not see clearly? Click for a new code.

User comments