Supermicro Launches New Servers and GPU-Accelerated Systems for AI-Ready Data Centers

Publisher:EE小广播Latest update time:2024-10-17 Source: EEWORLDKeywords:Supermicro Reading articles on mobile phones Scan QR code
Read articles on your mobile phone anytime, anywhere

Supermicro Launches New Servers and GPU Acceleration Systems for AI-Ready Data Centers, Powered by AMD EPYC™ 9005 Series CPUs and AMD Instinct™ MI325X GPUs


New Supermicro systems help customers upgrade their data centers to run AI workloads more smoothly


SAN JOSE, Calif., Oct. 17, 2024 /PRNewswire/ -- Supermicro, Inc., a provider of comprehensive IT solutions for AI/ML, HPC, cloud, storage and 5G/edge, today announced a new line of servers, GPU-accelerated systems and storage servers powered by AMD EPYC™ 9005 Series processors and AMD Instinct™ MI325X GPUs . The new H14 product line is one of the industry's broadest server families, including Supermicro's Hyper Systems, Twin multi-node servers and AI inference GPU systems, all available with air or liquid cooling. The new "Zen5" processor core architecture uses AVX-512 vector instructions with a full data path, supports CPU-based AI inference, and delivers 17% more instructions per cycle (IPC) than previous 4th Gen EPYC processors, resulting in increased performance per core.


Supermicro's H14 series uses the latest fifth-generation AMD EPYC processors, with up to 192 cores per CPU and a thermal design power consumption (TDP) of up to 500W. The new H14 series includes Hyper and FlexTwin™ systems to cope with higher cooling requirements. The series also includes three systems for AI training and inference workloads, supporting up to 10 GPUs and using the AMD EPYC 9005 series CPU as the main processor. In addition, the H14 series also has two systems that support the AMD Instinct MI325X GPU.


"The Supermicro H14 server with the EPYC 9005 64-core CPU delivers 2.44x the performance of the Supermicro H11 system based on the 2nd Gen EPYC 7002 Series CPUs on SPECrate®2017_fp_base1," said Jianhou Liang, President and CEO of Supermicro . "This significant performance improvement enables customers to reduce their total data center footprint by at least two-thirds2, while adding new AI computing performance and further improving data center energy efficiency. The H14 server series delivers exceptional performance, density, and energy efficiency with Supermicro's liquid and air cooling technologies, diverse system designs, and proven building block solutions."


"Supermicro's Building Block Solutions enable them to continue to bring AMD architecture-based solutions to market quickly across a wide range of compelling system designs," said Forrest Norrod, executive vice president and general manager of AMD's Datacenter Solutions Group . "We are collaborating with Supermicro to leverage their global in-house engineering and manufacturing capabilities, combined with their rack-scale integration of air- and liquid-cooled systems, to enable customers to quickly realize value with AMD EPYC CPUs and Instinct GPUs at any scale."


Supermicro's H14 server lineup includes the following product series:


Hyper - Supermicro's flagship enterprise server supports two EPYC 9005 CPUs, each with up to 192 cores and 500W of power, while supporting up to 9TB of memory through 24 DIMM slots for maximum performance. Available in a 1U chassis with up to 12 2.5" NVMe/SATA drive bays or a 2U chassis with up to 24 2.5" NVMe/SATA drive bays, Hyper features an advanced thermal design to accommodate the highest performance CPUs for demanding AI inference, enterprise or cloud workloads.


CloudDC - This versatile system is optimized for cloud data centers and features a single EPYC 9005 CPU in a 1U chassis with up to 12 2.5" NVMe/SATA drive bays. This system uses the OCP (Open Compute Platform) DC-MHS specification (Data Center Modular Hardware System) technology to ensure compatibility with OCP standards.


GrandTwin™ - GrandTwin is a 4-node computing platform that packs a single EPYC 9005 CPU into a dense 2U form factor. This system is often used in multi-server cluster applications such as object storage, virtualization, and high-performance computing applications.


FlexTwin™ - FlexTwin is a 2U 4-node high-performance, high-density computing system with dual EPYC 9005 CPUs per node. This system uses advanced liquid cooling technology to achieve high energy efficiency and run the highest performance EPYC 9005 CPUs smoothly for HPC, EDA and other demanding workloads.


5U GPU System - Supermicro's 5U PCIe GPU system with EPYC CPU supports up to 10 double-width accelerators for design and visualization applications.


4U GPU System (Liquid Cooling) - High-density 8-way accelerator platform with EPYC CPU, using advanced liquid cooling technology to support OAM accelerators in the most compact 4U form factor, designed for high-performance AI and HPC applications.


8U GPU System - This 8-way accelerator system uses AMD Instinct MI325X GPU and EPYC 9005 CPU for large-scale LLM AI training. Its 8U chassis can be deployed in any air-cooled data center.


Supermicro H14 products powered by AMD EPYC CPUs are now available for customer testing through the Supermicro JumpStart program.


image.png


Supermicro demonstrated the new H14 solution at the AMD Advancing AI Day on October 10, 2024 at the Moscone Center in San Francisco.


1 Supermicro A+ Server 2023US-TR4 with two AMD EPYC 7702 64-core CPUs, SPECrate®2017_fp_base score of 485 (https://www.spec.org/cpu2017/results/res2020q1/cpu2017-20200204-20866.html, data obtained on October 2, 2024), compared to Supermicro AS-2126HS-TN with two AMD EPYC 9555 64-core CPUs, SPECrate®2017_fp_base score of 1670. AS-2126HS-TN with 9555 CPUs results published on October 10, 2024 at www.spec.org/cpu2017/results.

2Based on the SPECrate®2017_fp_base comparison described in footnote (1), the number of systems using the AS-2126HS-TN with a 9555 CPU was reduced by 70.9% compared to the 2023US-TR4 with a 7702 CPU.


Keywords:Supermicro Reference address:Supermicro Launches New Servers and GPU-Accelerated Systems for AI-Ready Data Centers

Previous article:Akamai Announces Native API Traffic Analytics Interface to Simplify and Accelerate Enterprise API Protection
Next article:Liquid-cooled AI server market faces difficulties? UQD components are the key

Latest Network Communication Articles
Change More Related Popular Components

EEWorld
subscription
account

EEWorld
service
account

Automotive
development
circle

About Us Customer Service Contact Information Datasheet Sitemap LatestNews


Room 1530, 15th Floor, Building B, No.18 Zhongguancun Street, Haidian District, Beijing, Postal Code: 100190 China Telephone: 008610 8235 0740

Copyright © 2005-2024 EEWORLD.com.cn, Inc. All rights reserved 京ICP证060456号 京ICP备10001474号-1 电信业务审批[2006]字第258号函 京公网安备 11010802033920号