Peter Jang
November 20, 2024 14:09
According to the latest TOP500 list, NVIDIA technologies now accelerate 384 of the world’s leading supercomputers, enabling breakthroughs in climate prediction, drug discovery, and quantum simulations.
NVIDIA’s dominance in accelerated computing continues to grow, as highlighted in the latest TOP500 list of the world’s most powerful supercomputers. According to NVIDIA, 384 of these systems are currently powered by NVIDIA’s technology, fueling breakthroughs in areas such as climate prediction, drug discovery, and quantum simulations.
The role of accelerated computing in scientific advancement
NVIDIA has pioneered advances in artificial intelligence and accelerated computing since launching CUDA in 2006. The recent TOP500 list highlights the company’s achievements, with 87% of the 53 new entries being accelerated systems, 85% of which utilize NVIDIA Hopper GPUs. These systems play a pivotal role in driving research in important areas from climate modeling to drug development.
At the SC24 conference, NVIDIA unveiled cuPyNumeric, a CUDA-X library that allows developers to seamlessly scale their applications across compute clusters without modifying their Python code. Additionally, updates to the NVIDIA CUDA-Q platform were announced, allowing researchers to simulate quantum devices at unprecedented scale.
Mixed Precision and AI: New Tools for Researchers
Mixed-precision floating-point arithmetic and AI are increasingly becoming essential tools for modern scientific research. Two projects nominated for the Gordon Bell Prize at SC24 demonstrated how these technologies are advancing genomics and protein design. Using mixed precision, researchers such as David Keyes and Arvind Ramanathan have leveraged NVIDIA’s powerful computing platforms to achieve remarkable results in genomics and protein design, respectively.
NVIDIA’s BioNeMo framework, now open source, further supports AI-based drug discovery, enabling rapid creation and deployment of AI models for pharmaceutical applications. This reflects the broader shift to AI and mixed-precision workloads across the TOP500, with 249 exaflops of AI performance now accessible on these systems.
Sustainability in Accelerated Computing
As computing demands grow, so does the need for sustainability. NVIDIA’s solution is one of the most energy efficient, with systems such as EuroHPC/FZJ’s JEDI setting the benchmark for energy efficiency. The company also introduced new NIM microservices for its Earth-2 platform, significantly accelerating climate modeling and simulations.
In an era of increasing environmental awareness, NVIDIA’s advancements in accelerated computing provide a balance between performance and sustainability, paving the way for a future where high-performance computing can coexist with energy efficiency.
Visit the NVIDIA blog for more insights.
Image source: Shutterstock