NVIDIA expands CUDA-X tools for scientific research
NVIDIA has expanded its CUDA-X libraries with tools designed to accelerate scientific computing and support AI-assisted research. The additions help researchers process larger datasets, model more complex systems and analyze results while experiments or simulations are running.
For astronomy, cuPhoton.xBeam reduces the time needed to correlate Event Horizon Telescope observations from over a month to eight minutes. Pacific Fusion researchers used cuPhoton.xLens to examine plasma simulations as results emerged. Coupling that workflow with a locally running NVIDIA Nemotron-powered agent can enable interactive analysis to help steer simulations and inform fusion reactor design.
In drug discovery, cuPotential accelerates existing molecular simulation software by up to 3x while improving calculation stability and accuracy. NVIDIA is working with molecular dynamics software communities to integrate it into research workflows. For materials research, Daviator is coming to NVIDIA ALCHEMI, delivering quantum-mechanical calculations up to 22x faster and at up to 75% lower cost. At Lawrence Berkeley National Laboratory, NVIDIA DAQIRI connects electron microscopy detectors to GPU processing for analysis during experiments.
In biology, the newly open source cuSpa delivers a 900x speedup over CPUs for cell polygon calculations supporting spatial models at Memorial Sloan Kettering Cancer Center. The agent-ready rapids-singlecell framework helped Therna Biosciences process data 200x faster. NVIDIA’s open source GPU-accelerated HMMER library also sped up protein-sequence annotation searches at Basecamp Research by up to 84x.