NVIDIA expands CUDA-X libraries for agent-assisted scientific work
Original title: NVIDIA CUDA-X Libraries Give Agents the Tools to Accelerate Scientific Discovery
New and expanded GPU libraries speed black-hole imaging, fusion, molecular dynamics, materials and single-cell analysis, some by tens to hundreds of times.
NVIDIA said expanded CUDA-X libraries, including cuPhoton, cuPotential and cuSpa, let researchers and local AI agents process larger scientific datasets. cuPhoton.xBeam cut Event Horizon Telescope correlation from more than a month to eight minutes, and Pacific Fusion used cuPhoton.xLens with a local Nemotron agent during plasma simulations. cuPotential is claimed to speed molecular simulation software by up to 3x; Daviator in ALCHEMI is described as up to 22x faster, and open-source cuSpa as a 900x speedup versus CPUs at Memorial Sloan Kettering.
Unconfirmed: Speedups and cost figures are NVIDIA’s or partners’ claims, not independently verified here.
- cuPhoton correlated black-hole jet observations of J1924 and 3C279.
- The imaging pipeline runs on multi-GPU systems or an NVIDIA DGX Spark.
- NVIDIA is working with Schrödinger, AMBER, NAMD, GROMACS, LAMMPS and OMSF.
- rapids-singlecell, codeveloped with scverse, is offered in an agent-ready container.
- Basecamp Research reported HMMER searches up to 84x faster than standard HMMER.
From the original article
3 min read at blogs.nvidia.com
With expanded NVIDIA CUDA-X libraries, including cuPhoton, cuPotential and cuSpa, researchers can sift through vast amounts of data to accelerate research, from astrophysics and materials science to single-cell biology. The speedups are helping scientists analyze larger datasets and model more complex systems. The new libraries give AI agents the tools to help researchers make discoveries […]
By Kibibi Moseley · blogs.nvidia.com
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.