Supercomputing is one of the most important events for the supercomputing community. The 2024 edition of this international conference dedicated to high-performance computing, networking, storage and analysis is currently taking place for the first time in Atlanta, Georgia, from November 17 to 22. NVIDIA kicked off SC24 with a series of technological innovations aimed at transforming industries such as biopharma and climate science with AI and supercomputers.
The announcements are rooted in the company’s long history of transforming IT.
Jensen Huang said:
“Supercomputers are among humanity’s most vital instruments, driving scientific breakthroughs and pushing the frontiers of knowledge. Twenty-five years after the creation of the first GPU, we have reinvented computing and sparked a new industrial revolution.”
Huang highlighted CUDA-X libraries as the engines of accelerated computing. These libraries, with more than 400 components, including cuDNN for deep learning and cuQuantum for quantum circuit simulations, are essential for advancements in various fields, including AI-based healthcare and circuit simulations quantum.
NVIDIA also announced the new library cuPyNumic, a GPU-accelerated implementation of NumPy, designed to power data science, machine learning, and digital computing applications. It allows scientists and researchers to run their Python code, written with NumPy’s familiar interface, on laptops, workstations, cloud servers, or massive supercomputers without having to modify their code.
NVIDIA’s announcements focused on next-generation drug discovery tools, real-time climate predictions and quantum simulations.
Health and biopharmacy
BioNeMo Framework : NVIDIA announces the open source launch of this collection of programming tools, libraries and AI models optimized for drug discovery, doubling the training speed compared to traditional tools.
NVIDIA also revealed DiffDock 2.0, a tool to predict the 3D binding structure of a drug to a target protein, which is essential for drug discovery.
It can predict the orientation of a molecule 6.2 times faster and with 16% greater accuracy than its predecessor.
Climate forecasting with Earth-2
NVIDIA announced two new NVIDIA NIM microservices for NVIDIA Earth-2, a digital twin platform for simulating and visualizing weather and climate conditions. Microservices CorrDiff NIM et FourCastNet NIM can accelerate climate change modeling and simulation results by up to 500 times, providing more accurate predictions to anticipate natural disasters.
Quantum computing
NVIDIA also revealed significant updates to the NVIDIA CUDA-Q development platform, which enable larger-scale simulations and explore more scalable qubit designs. Google Quantum AI uses the solution to perform advanced simulations of quantum processors, reducing complex calculations from weeks to minutes.
Digital twins and real-time simulation
With Omniverse Blueprint for Real-Time CAE Digital Twins NVIDIA provides interactive workflows for creating digital models in diverse sectors including energy, aerospace, automotive, manufacturing, and other industries. This allows rapid optimization and reduction of industrial costs.
Hardware news
NVIDIA announced the availability of GPU PCIe NVIDIA H200 NVLthe newest addition to the Hopper family, ideal for energy-efficient air-cooled enterprise rack designs.
It delivers up to 1.7x faster language model inference and 1.3x higher performance on HPC applications, making it ideal for flexible data center setups.
The Superpuce Grace Blackwell, GB200 NVL4, planned for 2025, doubles the performance of previous generations.
Related News :