Announcing a Revolutionary Universal GPU System Ideal for Supermicro, AI, Machine Learning and HPC | Hebei Shinpo Online News / ONLINE NEWS

AsiaNet 95080

Achieves ultimate modularity and customization that allows flexible configuration of various CPU and GPU architectures

San Jose, Calif., March 24, 2022 /PRNewswire/ –Super Micro Computer, Inc. (Nasdaq: SMCI) is a global leader in high-performance computing, storage, networking and green computing technology. The company announced that it will be an innovative product that can simplify the deployment of large-scale GPU systems, “Universal GPU” ( 3478122-1&h=3807433449&u=). has been announced. The Universal GPU Server is a state-of-the-art design that also heralds support for future technologies in the future, providing ultimate flexibility along with the enterprise’s resource-efficient architecture.


Universal GPU system architecture can be configured with multiple GPU form factors, processor choices, storage and networking options, optimized by combining the latest technologies and delivered as a uniquely configured and highly scalable system . This scalable system can be optimized to meet the unique artificial intelligence (AI), machine learning (ML), and high-performance computing (HPC) application requirements of each enterprise. Ideal for customers looking for a next-generation computing environment capable of meeting the temperature demands of next-generation CPUs and GPUs.

Charles Liang, President and CEO of Supermicro, said, “Supermicro’s Universal GPU Server is a revolutionary product that sets new standards for industry standards and data center infrastructure. powerful and scalable modular form for enterprise CTOs and IT managers It transforms the ground rules of planning when deploying a system This product has a flexible and innovative system design that when combined with the best technology, enables rapid and large-scale deployment.”

The Universal GPU Platform starts with a system that combines a 3rd Gen AMD EPYC(TM) 7003 processor with an AMD MI250 GPU or an NVIDIA A100 Tensor Core 4-GPU, then an Intel(R) Xeon(R) processor from 3rd generation. ) We plan to support systems that combine a Scalable CPU with an NVIDIA A100 Tensor Core 4-GPU. These systems are designed to support up to 700W GPU thermal capacity.

The Supermicro Universal GPU Platform is designed to work with a variety of GPUs based on an open standard design. Compliant with hardware design standards such as Universal Baseboard (UBB), OCP Accelerator Module (OAM), PCI-E, and platform-specific interfaces, it is ideal for HPC and AI workloads You can choose the GPU architecture. This can meet the stringent requirements of many enterprises and simplifies GPU solution deployment, validation, operations, and upgrades. Additionally, IT managers can easily choose the right combination of CPU and GPU to create the best system for their customers.

4U or 5U universal GPU servers will be sequentially available with UBB and PCI-E 4.0 or PCI-E 5.0 compliant accelerators. Additionally, 32 DIMM slots and a wide range of storage and networking options are available to connect to the latest PCI-E standards. Supermicro Universal GPU Servers use SXM or OAM form factor base cards that support very fast GPU-to-GPU interconnects, such as NVIDIA NVLink(TM) and AMD Infinity Fabric(TM)/xGMI, or you can also connect the PCI-E GPU card. directly through the PCI-E slot. It supports all major CPU and GPU platforms currently on the market, allowing you to choose the one that suits your workload and application.

Designed for maximum airflow, this server can support current and future higher TDP (Thermal Design Power) CPUs and GPUs that are essential for maximizing application performance. Supermicro Universal GPU servers also offer a liquid cooling option if the CPU and GPU require additional cooling capacity. Plus, the modular design allows you to replace or upgrade specific subsystems of your server, extending the life of your entire system and reducing the amount of electronic waste generated with each new change in technology. CPU or GPU.

Paresh Kharya, Senior Director of Data Center Computing Product Management at NVIDIA, said, “The NVIDIA A100 Tensor Core GPU accelerates all AI applications and over 2,700 HPC applications. Supermicro’s new server platform integrates the NVIDIA A100 for a wide range of work including HPC, AI inference, training, and machine learning, providing businesses with a powerful system that delivers exceptional performance and productivity on the go.”

Brad McCredie, Corporate Vice President of Data Center GPUs and High-Speed ​​Processing at AMD, said, “AMD appreciates our partnership with Supermicro. Innovations in our advanced AMD EPYC processors and AMD Instinct(TM) GPUs for meet the needs of the HPC industry and the demand for data center workloads from accelerated computing. We look forward to a Supermicro Universal GPU system that will be the ideal platform to demonstrate.”

For more information on Universal GPUs, please see the links below.

In addition, a webinar (in English) will be held at 2:00 a.m. on April 22, 2022, Japan time. Pre-registration (free) is required on the link below to participate.

About Supermicro (Super Micro Computer, Inc.)

Supermicro (NASDAQ: SMCI) is a global leader in application-optimized hardware and complete computing solutions. Founded and headquartered in San Jose, CA, Supermicro is a leading innovator in the IT infrastructure market, including enterprise, cloud, AI, HPC, and IoT/Edge. As a complete IT solutions provider that provides servers, AI, storage, IoT, network switches, software and services, we constantly focus on transforming our products, including various types of motherboards, chassis and power supplies. Designed, manufactured and supplied by. Our products are designed and manufactured in the USA, Taiwan and the Netherlands, leveraging global operations for production scale and efficiency, optimal for green computing with improved TCO and environmental impact reduced. Our unique and award-winning Server Building Block Solutions(R) are a combination of various form factors, processors, memory, GPUs and other accelerators, storage, networking, power supplies and cooling methods (air cooling and liquid cooling). , it is possible to build the optimal configuration for the customer and realize the optimization of the application and the workload.

Supermicro, Server Building Block Solutions and We Keep IT Green are trademarks or registered trademarks of Super Micro Computer, Inc.

Intel, the Intel logo, and other Intel marks are trademarks of Intel Corporation or its subsidiaries.

All other brands, names and trademarks belong to their respective owners.

・ This press release is an abbreviated version of the press release issued in the United States on March 21, 2022 (local time).

Contact: Super Micro Computer Inc. Press Contact National Public Relations Agent Hoffman Japan Co., Ltd. Contact: Kojima / Hashimoto Email:

(Japanese version: provided by the customer)

Leave a Comment