NVIDIA presents news for the future of high-performance computing

A NVIDIA announced this Monday a series of news for its platforms HPC (High performance computer), used mainly for activities that require more computational power to solve problems or render software than those available on desktops or servers, even corporate ones. Check out the news below:

What's New in NVIDIA HPC

The first news announced by NVIDIA was adopted by Microsoft Azure, the cloud computing solution from the company founded by Bill Gates, from the NVIDIA Quantum-2 400G Infiniband – which will allow users to have greater monitoring, management and detection of anomalies in their work developed in this aspect.

In addition, the implementation of the platform will also allow for greater scalability thanks to its Quantum Infiniband, which will make past processing power limitations no longer an issue.

Another novelty announced was related to the NVIDIA OMNIVERSE, the company's platform used for rendering 3D simulations on the most diverse fronts: it, together with the US space and air defense company Lockheed Martin, were chosen by the US National Ocean and Atmospheric Monitoring Agency (NOAA) to be responsible for monitoring terrestrial planetary activities. 

The project involves the construction of satellites by Lockheed Martin which, together with other physical terrestrial data collection platforms, will pass the information to the omniverse which will then build models representing the package of elements, serving as a complex and visual indicator for anomaly detection on these world fronts. Everything will be stored both in solutions DGX e hgx from NVIDIA servers and also in repositories of the Amazon Web Services.

Digital Twin simulations also arrive for HPC Data Centers

In addition to the above news, the NVIDIA also announced the arrival of simulations Digital Twin for the planning and construction of HPC Data Centers.

Simulations Digital Twin (digital twins) are virtual models that recreate real structures and projects in situations that perfectly simulate their functioning - thus creating a testing environment in which possible modifications and changes can be tested before being adopted in the physical model.

O Digital Twin for Data Centers HPC will use the omniverse, NVIDIA Air and simulation solutions 6Sigma DCX e Patch Manager for creating the virtual models.

The future of HPC

Finally, the NVIDIA also presented some future challenges for the HPC, especially with the start of implementations in Edge Computing, that is, solutions in which high-performance computers will be closer to the origin of the data they process, even if the conditions of the environment in which it is installed are not the best.

For that, it will be necessary that the NVIDIA create structures of HPC more compact and that connect in a more compact way and adapted to the environment, and which will allow an increase in data capture and processing between 10 and 1000 times.

Data migration between computers HPC used for collection and those used for processing will also be important, with the company seeing that high-performance information streaming solutions will also be important for the future of this type of computing.

For this, the NVIDIA announced its solution to Edge Computing on HPC, which will involve all these characteristics mentioned above in environments that may adopt programming languages ​​such as C++, Python, JAX and graphing for a better understanding of experiments and a larger testing surface.

All news announced by NVIDIA is expected to hit the market in 2023. Price details and other features were not presented by the company, but it is possible that they will be presented in the future. 

Source: NVIDIA

See also:

NVIDIA launches Omniverse Cloud, solution for projects in the metaverse