NVIDIA DGX SYSTEMS
Artificial intelligence allows companies and the scientific community the discovery of new frontiers. Indeed, its adoption has so far been limited by CPU architectures, which do not meet the needs of today’s deep learning algorithms, which require the right power to parallel computing.
To meet these needs, NVIDIA has created the DGX Family solutions for Artificial Intelligence and Deep Learning.
Whether it’s creating quality customer experiences, delivering more accurate results in science , or simplifying the supply chain, companies need an infrastructure that can deliver AI-powered information. NVIDIA DGX ™ systems offer world-leading solutions for large-scale enterprise AI infrastructure.
DGX solutions are fully integrated and combine the world’s most powerful GPU technology, with innovative GPU-optimized software and simplified management, for revolutionary performance and results. NVIDIA DGX solutions are designed to provide data scientists with pervasive access to the exploration of artificial intelligence, ranging from the desk to the data center and the cloud.
NVIDIA DGX A100
THE UNIVERSAL SYSTEM FOR AI INFRASTRUCTURE
NVIDIA DGX ™ A100 is the universal system for all workloads related to Artificial Intelligence, offering unprecedented processing density, performance and flexibility, within a single system. Launched by NVIDIA in May 2020, it’s the new member of the DGX family and also the most powerful.
Based on the revolutionary NVIDIA A100 Tensor Core GPU, the new NVIDIA DGX ™ A100 performs training, inference and analysis workloads simultaneously and with ease. This solution also includes exclusive access to a global team of artificial intelligence experts who offer planning, deployment, and optimization expertise to help companies implement their digital transformation processes quickly.
GPU: 8x NVIDIA A100 Tensor Core
Performance: 5 petaFLOPS AI 10 petaOPS INT8
Total system GPU memory: 320 GB
CPU: Dual AMD Rome 7742, 128 cores total, 2.25 GHz (base), 3.4 GHz (max boost)
System memory: 1TB
System weight:123 kg
Temperature: 5 ° – 30 °
Software: Ubuntu Linux OS
NVIDIA DGX STATION
A plug-and-play solution, the world’s first personal supercomputer designed for Intelligence Artificial, built on the same software stack that powers all NVIDIA DGX ™ systems. This powerful workstation gives you a quick start to the world of deep learning and data science.
NVIDIA® DGX Station ™ is the only personal supercomputer with four NVIDIA® Tesla® V100 GPUs, next-generation NVIDIA NVLink ™ and the new Tensor Core architecture, based on the DGX software. It guarantees you silence and revolutionary performance, with the same power of 400 CPUs.
With NVIDIA® DGX Station ™, you get world-class support and access to NVIDIA deep learning expertise, a training library built by experts, software updates, and priority resolution of critical issues.
GPU: 4X Tesla V100
TFLOPS (FP16 GPU): 480
Total system GPU memory: 64 GB
NVIDIA Tensor Core: 2,560
NVIDIA CUDA® Core: 20,480
CPU: Intel Xeon E5-2698 v4 2.2 GHz (20 cores)
System memory: 256GB DDR4
LRDIMM Acoustics: <35 dB
System weight: 40 kg Maximum
Power Requirements: 1,500W
- Operating System Ubuntu
- Desktop Linux
- Recommended GPU driver for DGX
- Toolkit CUDA