4U Dual Processor (3rd Gen Intel® Xeon®) GPU System with NVIDIA HGX A100 8-GPU 40GB/80GB, NVLink, NVSwitch, Redundant 3000W Titanium Level Power Supplies.
High Performance Computing AI/Deep Learning Training
Key Features
Dual Socket P+ (LGA-4189) 3rd Gen Intel® Xeon® Scalable Processors
Intel® C621A Chipset
32 DIMM Slots; Up to 8TB DRAM; Up to 8TB Intel® Optane™ Persistent Memory (up to 12TB with DRAM); 3200/2933/2666 ECC DDR4 LRDIMM;RDIMM; Intel® Optane™ Persistent Memory 200 series
10 PCI-E Gen 4.0 X16 LP Slots AIOM/OCP 3.0 Support
Flexible networking options
2 M.2 NVMe and SATA 6x 2.5" Hot-swap NVMe/SATA/SAS drive bays
4 heavy duty fans with optimal fan speed control
4x 3000W redundant Titanium level power supplies
Putting the HGX A100 8-GPU server platform together
With the GPU baseboard building block, the NVIDIA server-system partners customize the rest of the server platform to specific business needs: CPU subsystem, networking, storage, power, form factor, and node management. To deliver the highest performance, we recommend the following system design considerations:
Select two highest-end server CPUs to pair with the eight A100 GPUs, to keep up with the A100.
Use plenty of PCIe links. Use a minimum of four PCIe x16 links between the two CPUs and eight A100 GPUs, to make sure there is enough bandwidth for the CPUs to push commands and data into the A100 GPU.
For the best AI training performance at scale (many nodes together running a single training job), networking performance between nodes is critical. Use a ratio of up to 1:1 A100 GPUs to network interface cards (NICs). The Mellanox ConnectX-6 200-Gb/s NIC is the best option.
Attach NIC and NVMe storage to the PCIe switch and place it close to the A100 GPU. Use a shallow and balanced PCIe tree topology. The PCIe switch enables the fastest peer-to-peer transfer from NIC and NVMe in and out of the A100 GPU.
Adopt GPUDirect Storage, which reduces read/write latency, lowers CPU overhead, and enables higher performance.
Titanium Level 
