Peter Zhang
Nov 20, 2024 14:09
NVIDIA know-how now accelerates 384 of the world’s prime supercomputers, enabling breakthroughs in local weather forecasting, drug discovery, and quantum simulation, in accordance with the newest TOP500 listing.
NVIDIA’s dominance within the discipline of accelerated computing continues to develop, as highlighted by the newest TOP500 listing of the world’s strongest supercomputers. In response to NVIDIA, 384 of those methods now function on its know-how, facilitating important breakthroughs in areas corresponding to local weather forecasting, drug discovery, and quantum simulation.
Accelerated Computing’s Function in Scientific Progress
Because the introduction of CUDA in 2006, NVIDIA has been on the forefront of developments in synthetic intelligence and accelerated computing. The latest TOP500 listing underscores the corporate’s achievements, with 87% of the 53 new entries being accelerated methods, 85% of which make the most of NVIDIA Hopper GPUs. These methods are pivotal in driving analysis in essential areas, from local weather modeling to drug improvement.
On the SC24 convention, NVIDIA unveiled cuPyNumeric, a CUDA-X library that permits builders to scale their purposes seamlessly throughout computing clusters with out modifying their Python code. Moreover, updates to the NVIDIA CUDA-Q platform have been introduced, empowering researchers to simulate quantum units on an unprecedented scale.
Blended Precision and AI: New Instruments for Researchers
Blended-precision floating-point operations and AI are more and more turning into important instruments for contemporary scientific analysis. At SC24, two tasks nominated for the Gordon Bell Prize demonstrated how these applied sciences are advancing genomics and protein design. Utilizing combined precision, researchers like David Keyes and Arvind Ramanathan have achieved exceptional leads to genomics and protein design, respectively, leveraging NVIDIA’s highly effective computing platforms.
NVIDIA’s BioNeMo framework, now open supply, additional helps AI-driven drug discovery, enabling fast creation and deployment of AI fashions for pharmaceutical purposes. This displays a broader shift in direction of AI and mixed-precision operations throughout the TOP500, with 249 exaflops of AI efficiency now accessible to those methods.
Sustainability in Accelerated Computing
As computational calls for improve, so does the necessity for sustainability. NVIDIA’s options rank among the many most energy-efficient, with methods just like the JEDI at EuroHPC/FZJ setting benchmarks in power effectivity. The corporate additionally launched new NIM microservices for its Earth-2 platform, considerably accelerating local weather modeling and simulation.
In an period of heightened environmental consciousness, NVIDIA’s developments in accelerated computing provide a steadiness between efficiency and sustainability, paving the best way for a future the place high-performance computing can coexist with power effectivity.
For extra insights, go to the NVIDIA weblog.
Picture supply: Shutterstock