Inspur Blog

AI-boosting servers with NVIDIA A100 80GB PCIe GPUs

An “industrial HPC revolution” is coming, predicts NVIDIA, catalyzed by a convergence of growing industry needs for high-performance computing, and breakthroughs in deep learning and compute acceleration. As an AI technology forerunner NVIDIA is paving the way with advanced computing infrastructures, like the A100 80G PCIe GPU. It delivers ultra-fast GPU memory bandwidth up to 2TB/s, with each Multi-Instance GPU (MIG) achieving a memory capacity of 10GB and 1.2 times higher throughput than the previous iteration, A100 40GB.

Keeping in lockstep, Inspur is leading the charge in AI hardware with three servers supporting the NVIDIA® A100 80G PCIe GPU: NF5468M6, NF5468A5, and NF5280M6. These servers are designed and positioned to meet the needs of evolving and increasingly diversified large-scale AI and HPC computing scenarios. Backed by robust R&D and agile production capabilities, Inspur is one of the few AI computing manufacturers to fully support NVIDIA® A100 Tensor Core GPUs at the earliest time. In so doing, Inspur provides customers with the most comprehensive, high-performance, and flexible AI computing products that adapt to various computing scenarios.

Leveled-up AI compute power for evolving needs

The NF5468M6, ultra-flexible for AI workloads, can be loaded with 2x 3rd Gen Intel® Xeon Scalable Processors and 8x NVIDIA® A100/A30/A40 or 16x A10 GPUs in a 4U chassis. It is widely used in AI public cloud and enterprise AI cloud platforms, video analysis, and other scenarios with rich and diversified configurations. The NF5468M6 also supports one-click switching of three GPU topologies including Balance, Common, and Cascade to meet the performance optimization requirements of different AI application scenarios.

The NF5468A5 is a versatile AI server that can be equipped with 2x AMD® EPYC 7003 series processors and 8x NVIDIA® A100/A30/A40 GPUs in a 4U chassis. It supports the most advanced AI computing technology and adopts CPU-to-GPU non-blocking design, providing excellent performance. Featuring a high-quality and high-compatibility design, NF5468A5 offers a reliable computing platform for clients specializing in various AI applications.

The NF5280M6 is purpose-built for all scenarios. With 2x 3rd Gen Intel® Xeon Scalable Processors and 4x NVIDIA® A100/A40/A30/A10 GPUs or 8x T4 Tensor Core GPUs in a 2U chassis, it is capable of long-term stable operation at 45°C. The NF5280M6 is equipped with the latest PFR/SGX technology and a trusted security module design suitable for demanding AI applications.

Leave a Reply

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