ISC High Performance: Inspur NF5468M6 and NF5468A5 AI Servers Supporting Nvidia A100 PCIe Gen 4 GPU for AI

Inspur Electronic Information Industry Co.,Ltd releases NF5468M6 and NF5468A5 AI servers supporting the Nvidia A100 PCIe Gen 4 GPU at ISC High Performance 2020.

Nvidia A100 PCIe Gen 4 GPU

It will provide AI users with the ultimate AI computing platform with performance and flexibility.

Thanks to its agile and strong product design and development capabilities, Inspur supports the Nvidia A100 Tensor Core GPU and build up a comprehensive and competitive next-gen AI computing platform. The A100 GPU brings versatility by accelerating a full range of precisions, from FP32 to FP16 to INT8 and all the way down to INT4. This includes the TF32 precision, which works like FP32 while providing 20x higher FLOPS for AI without requiring any code change. In addition, the A100 offers multi-instance GPU technology, which enables a single GPU to be partitioned into 7 hardware-isolated instances to work on multiple networks simultaneously. At present, The company’s NF5488M5-D and NF5488A5 with the A100 are in mass production.

They present innovative designs and strike a balance between performance and flexibility, which meets increasingly complex and diverse AI computing needs. They can offer computing performance for HPC and cloud application scenarios.

They accommodate 8 double-width A100 PCIe cards in a 4U chassis. Both support the PCIe Gen4 of 64GB/s bi-directional bandwidth, delivering a 100% increase in bandwidth compared to PCIe Gen3 with the same power consumption. Its performance will meet the requirements of complex challenges in data science, HPC, and AI. Besides, 40GB of HBM2 memory increases memory bandwidth by 70% to 1.6TB/s, allowing users to train larger deep learning models. The NVIDIA NVLink bridge design can provide P2P performance of up to 600GB/s between two GPUs, resulting in increases in training efficiency

Another 2 AI servers of the company, NF5468M5 and NF5280M5, also support Nvidia A100 PCIe Gen 4.

Read also:
Inspur: M6 Four-Socket Servers
Supporting 3rd gen Intel Xeon scalable processor and Optane persistent memory 200 series
June 26, 2020 | Press Release