Nvidia GTC 2022: More than 60 updates for Cuda-X libraries The highlights of Cuda updates and enhancements
05.04.2022By Robert Brunner *
Nvidia has more than 60 updates for its “Cuda-X” libraries at its “GTC” conference on Tuesday, opening up new areas in the sciences and industrial application for “accelerated computing”, i.e. for applications running on computers equipped with accelerators, such as Nvidia GPUs and DPUs.
The new “Cuda” libraries make it easier to work in the fields of quantum informatics and 6G research, research on logistics optimization, robotics, cyber.Security, genomics and drug discovery, data analysis and more.
The new libraries or extensions are intended to boost the processes in the fields of quantum computing and 6G research, logistics optimization, robotics, cyber security, genomics and drug development, data analysis and also contain new functions that significantly improve the performance of the Cuda software computing platform.All in all, Nvidia is opening up new areas for accelerated computing.
New software stacks and updates are now available and are intended to strengthen Cuda’s position as the industry’s most comprehensive platform for developers to create accelerated applications to meet challenges in high performance computing (HPC) areas such as research in 6G, quantum computing, genomics, drug discovery and logistics optimization, as well as advanced work in robotics, cyber security, data analysis and more.
The Nvidia platforms for AI, HPC and graphics are based on Cuda and include software development kits and tools that form the core for higher performance and accelerated algorithms in various application areas. The SDKs make it easier for developers, researchers, and data scientists to take full advantage of the power of Nvidia’s advanced hardware and cope with the immense complexity at the intersection of computing, algorithms, and science.
selection
Greg Estes, vice president of developer programs at Nvidia, said, “AI and accelerated computing innovations are driving major scientific breakthroughs and the development of new applications and services in virtually every industry. With these updates, Nvidia is making it easier than ever for researchers and developers to harness the power of Cuda and get the highest performance out of our platforms.“
The Cuda platform has been downloaded more than 33 million times since its launch in 2008, 8 million times last year alone, a triple increase in three years. Recent highlights include:
- To accelerate the simulation of quantum circuits is now “CU-Quantum” generally available and available for the first time in the “CU-Quantum DGX” appliance, which offers HPC researchers a complete quantum simulation stack. It is optimized for use on the “Nvidia DGX” platform.
As a key element in a rapidly growing ecosystem, “CU-Quantum” is now integrated as a backend in popular simulators of “Google Quantum AI”, “Xanadu” and Oak Ridge National Laboratory and is also now available as part of the development platforms for quantum applications of Classiq and Zapata Computing or used by QC Ware, Xanadu and others to conduct quantum research in areas from chemistry to climate modeling on a large scale.
- „Sionna“, for research using 6G physical layers, is a new, GPU-accelerated open source library with native support for neural network integration and machine learning. Sionna enables rapid prototyping of complex communication system architectures and complements Nvidia’s investments in the wireless sector, which also includes “Aerial”, a unified software-defined platform for the deployment of AI-on-5G.
Christoph Studer, Professor at the Department of Information Technology and Electrical Engineering at ETH Zurich, is enthusiastic: “Sionna will change the way AI/ML-based communication systems are developed in the future. Even better, it runs incredibly fast on our Nvidia DGX station.“
- „Rapids“, the data science libraries for accelerating drug development, social connections, fraud detection and more, is one of Nvidia’s most popular SDKs with 2 million downloads and is used in over 5,000 projects on “Github”. The “Rapids Accelerator for Apache Spark” speeds up processing by more than three times without the need to change the code. With 80 percent of Fortune 500 companies using Apache Spark in production, data engineers can transparently accelerate Spark data frames and SQL operations.
- „CU-Opt“, formerly “RE-Opt”, for logistic optimization research focusing on vehicle routing, allows users to route thousands of packages to thousands of locations in a matter of seconds with record accuracy. This enables real-time redirection and potentially saves billions in delivery costs every year.
- „Morpheus“ for cyber security applications, it makes it possible to analyze up to 100 percent of the data in real time to detect threats more accurately and eliminate them faster. F5, a provider of security products, was able to increase the performance of its malware detection model by 200 times, from 1,013 messages per second to 208,333 messages per second, with only 136 lines of code.
- „DGL Containers“, for the training of graphical neural networks, adds improvements in the algorithm that allow 4x faster end-to-end ETL and training workflows (ETL = Extract, Transform, Load) than CPUs in areas such as drug discovery and fraud detection.
- „Nsight“Systems, for system-wide visualization of an application’s performance, allows developers to see how GPU-accelerated libraries use system resources, interact with the application, and identify optimization opportunities to reduce bottlenecks.
- Other updated libraries include “Monai” for medical imaging, “Flare” for federated learning with almost 300,000 downloads,”Maxine” for the reinvention of communication, “Riva” for voice AI,”Merlin” for recommendation systems and “Isaac“ for robotics.
More about the Cuda-X libraries in the GTC 2022 keynote by Nvidia CEO Jensen Huang.
The 3 million members of the Nvidia developer program will receive benefits such as access to training courses and materials through the “Nvidia Deep Learning Institute”, technical lectures and tools and much more.
