Supermicro Launches New GPU Server Powered by NVIDIA A100, Delivering Up to 5 PetaFLOPS Performance

Supermicro has announced the launch of its latest GPU server powered by NVIDIA A100, capable of delivering up to 5 PetaFLOPS of performance in a compact 4U form factor—designed specifically for Artificial Intelligence (AI) and Deep Learning applications.

NVIDIA HGX A100 Architecture

Supermicro, a leader in high-performance server solutions, introduced its new GPU server lineup built on the latest NVIDIA HGX A100 architecture. The lineup includes two models:

A 2U system equipped with 4 NVIDIA HGX A100 GPUs

A 4U system equipped with 8 NVIDIA HGX A100 GPUs

The 4U model delivers up to 5 PetaFLOPS of AI processing performance. The GPUs are connected in an all-to-all topology via NVIDIA NVSwitch, providing up to 600 GB/s of bandwidth between GPUs.

Server Specifications

In addition, the server supports: Up to 2 AMD EPYC processors, Up to 8TB of system memory and PCIe Gen 4 for high-speed expansion. This enables support for high-performance networking such as GPUDirect RDMA and InfiniBand HDR, offering up to 200 Gbps bandwidth—making it ideal for workloads including AI, Deep Learning, High-Performance Computing (HPC), and Data Analytics.

About Throughwave Thailand

Throughwave Thailand is a distributor of comprehensive enterprise IT solutions, including server, storage, networking, and security products, along with VMware and Microsoft solutions. The company serves leading organizations with tens of thousands of users. Throughwave's team of highly experienced engineers provides trusted, 24/7 support in collaboration with partners across Thailand.

For more information, please contact Throughwave Thailand:

Tel: +66 2-210-0969

Email: info@throughwave.co.th

Or visit our website: https://www.throughwave.co.th