NVIDIA DGX Station A100 claims to deliver data center performance without a data center.
NVIDIA announces DGX Station A100 with workstation form factor. The product comes with 4 80 GB A100 GPUs and offers a total of 320 GB of GPU memory.
The NVIDIA DGX Station A100 is a product developed for science and business. So it’s suitable for research and engineering teams working on huge data sets and investing heavily in machine learning and artificial intelligence.
According to Nvidia, the DGX Station A100 is capable of delivering” data center performance without a data center.” This means that it is plugged into a standard wall outlet and does not require a data center-level cooling system. Inside the gold case is a 64-core AMD CPU with 512 GB of memory and a 7.68 TB NVME SSD. NVIDIA announced that the product is the only workgroup server that supports Multi-Instance GPU (MIG) technology. This means that a single DGX Station A100 can provide 28 separate GPU instances for parallel jobs or multiple users to access. BMW, Lockheed Martin, NTT Docomo, the Pacific Northwest Laboratory have already started using DGX stations for some artificial intelligence and research-oriented work.
NVIDIA announced that the standard DGX A100 will be sold with a new 80 GB GPU and increase its memory capacity to 640 GB per system. An upgrade option will also be available for customers who have purchased the DGX A100 with four 40GB A100 cards (320GB). The company did not disclose any price for its new enterprise-grade hardware. Let us remind you that the DGX A100 was launched in May with a starting price of $ 199,000.
