NVIDIA expands CUDA-X libraries for scientific computing and AI-driven research

NVIDIA expands CUDA-X libraries for scientific computing and AI-driven research
NVIDIA powers scientific AI

Researchers are using expanded NVIDIA CUDA-X libraries to speed up data-intensive work in fields including astrophysics, fusion energy, drug discovery and single-cell biology. The additions, including cuPhoton, cuPotential and open-source cuSpa, are aimed at helping scientists process larger datasets and run more complex models with AI agents integrated into their workflows.

Highlights

  • NVIDIA expanded its CUDA-X stack with cuPhoton, cuPotential, and cuSpa libraries, targeting faster simulations and agent-assisted research in imaging, simulation, and biology.
  • cuPhoton.xBeam cut Event Horizon Telescope correlation time from over a month to eight minutes, with workflows running on NVIDIA DGX Spark and distributed multi-GPU systems.
  • Daviator accelerates ab initio calculations in Vienna Ab initio Simulation Package up to 22x faster with 75% lower costs, while cuSpa delivered a 900x speedup for cell analysis at Memorial Sloan Kettering Cancer Center.

New tools target faster simulation and analysis

NVIDIA Newsroom said the company is expanding its CUDA-X software stack with libraries designed to accelerate scientific workloads across imaging, simulation and biological data analysis. The package includes cuPhoton for astronomy and plasma research, cuPotential for molecular simulation, and cuSpa for spatial biology, with the company positioning the tools as building blocks for agent-assisted research.

In astrophysics, NVIDIA said cuPhoton.xBeam cuts the time needed to correlate Event Horizon Telescope observations from more than a month to eight minutes. The company said the toolkit has already been used to correlate observations of black hole jets J1924 and 3C279, and that the full workflow can run on distributed multi-GPU systems or an NVIDIA DGX Spark system.

For fusion research, NVIDIA said Pacific Fusion researchers are using the cuPhoton.xLens module to extract insights while plasma simulations are still running. The company said the workflow can be paired with a locally running NVIDIA Nemotron-powered agent, allowing interactive analysis that helps guide simulations in real time for reactor target and chamber design.

Adoption spans drug discovery, materials science and biology

In molecular modeling, NVIDIA said cuPotential, based on work by researchers at the Simons Foundation's Flatiron Institute and D. E. Shaw Research, can accelerate existing simulation software by up to 3x while improving stability and accuracy. The company said it is working with Schrödinger and the AMBER, NAMD, GROMACS, LAMMPS and Open Molecular Software Foundation communities, while Cadence, Fathom Therapeutics, Topos Bio, Genesis Molecular AI and Terray Therapeutics are adopting the technology for drug discovery.

NVIDIA also said Daviator, a collection of batched GPU kernels for the Vienna Ab initio Simulation Package, is coming to NVIDIA ALCHEMI with quantum-mechanical calculations up to 22x faster and at up to 75% lower cost. At Lawrence Berkeley National Laboratory, the company said its DAQIRI networking library moves electron microscopy analysis into the experiment itself, allowing researchers to identify useful signals and adjust measurements as data is collected.

In life sciences, NVIDIA said the newly open-source cuSpa delivers a 900x speedup over CPUs for cell polygon calculations at Memorial Sloan Kettering Cancer Center. The company added that rapids-singlecell, codeveloped with scverse, helped Therna Biosciences process data 200x faster, while 10x Genomics has integrated the framework into its Sentira platform; NVIDIA also said its GPU-accelerated HMMER library enabled Basecamp Research to run benchmark searches up to 84x faster than standard HMMER for protein sequence annotation.

As noted in our earlier article, the Saturn Funding Trust note offering highlighted asset-backed financing for SpaceX’s leased NVIDIA GPU systems.

This material may contain third-party opinions, none of the data and information on this webpage constitutes investment advice according to our Disclaimer. While we adhere to strict Editorial Integrity, this post may contain references to products from our partners.