DGX2

 

 

 

 

 

NVIDIA DGX-2

The world's most powerful AI system
for the most complex AI projects.

 

THE MOST POWERFUL IA SYSTEM IN THE WORLD

 

Experience 10x deep learning performance with NVIDIA® DGX-2 ™, the world's first 2-petaFLOPS system that combines 16 interconnected GPUs for NVIDIA's highest levels of speed and scale. Powered by NVIDIA® DGX ™ software and NVIDIA NVSwitch's scalable architecture, the DGX-2 is the platform of choice for the most challenging AI challenges in the world.

Download technical sheet

 

 

 

 

DISCOVER THE BENEFITS

Unparalleled computing power for optimizing training procedures

Artificial intelligence (AI) -based workflows are becoming more complex and require more computing power. NVIDIA® DGX-2, which leverages the industry's most advanced GPU architecture, can interconnect up to 16 GPUs to accelerate previously untraceable AI models. This solution offers exceptional scalability to help you drive models up to four times more complex on a single node, with performance times greater than with an 8 GPU system. NVIDIA DGX-2 is now available via two models including the all-new DGX-2H, specifically designed to deliver maximum performance in the most intensive applications. Find out how DGX-2H is the cornerstone of DGX-2 POD, the first-ever IA supercomputing infrastructure to join the Top500 list of the fastest supercomputers in the world.

Revolutionary network structure

With NVIDIA DGX-2, the complexity and size of models are no longer limited by traditional architectures. You can now parallelize the training of your models by taking advantage of the NVSwitch network structure. This innovative technology - which equips the first ever 2-petaflop GPU accelerator with 2.5TB / sec interconnect bandwidth - delivers up to 12x more performance than previous-generation solutions and offers up-to-date performance. access to solutions up to 5 times faster.

AI projects on an unprecedented scale

DGX-2 is a turnkey solution that allows you to quickly evolve AI projects. Implement flexible network options that facilitate the configuration of advanced Deep Learning clusters and scalable virtualization features that enhance user experience and workload isolation in shared infrastructure environments. This rapid deployment model and scalable architecture enable your IT teams to accelerate deployment procedures to focus on your projects.

Enterprise IA infrastructure with permanent availability

AI is essential to your business, so you need to be able to work with peace of mind with a reliable, available and functional platform. The NVIDIA DGX-2 solution has been specifically designed to reduce downtime, streamline network operation, and maintain business continuity. With a business infrastructure that draws on all of NVIDIA's expertise, DGX-2 helps you implement reliable, ongoing operations for your most important AI initiatives.

BENCHMARKS

Nvidia DGX-2 APY EUROPE benchmarcks

The new standard of backtesting

NVIDIA DGX-2 and Python GPU acceleration have processed more than 20 million trading simulations, setting a new standard through the latest STAC-A3 backtesting report.

The reference of AI infrastructures

NVIDIA DGX-2 has established six MLPerf-approved records, a set of benchmarks for machine learning that measure training performance in a wide variety of use cases.

Unprecedented performance

Based on DGX systems, NVIDIA DGX SuperPOD is the 22nd fastest supercomputer in the world. Discover how it is used to accelerate the development of autonomous vehicles.

 

 

APY IS ATP PARTNER IN FRANCE
CONTACT OUR SPECIALISTS IN NVIDIA DGX-2 SOLUTION

CONTACT