Hardware
High-Throughput Structure Prediction with BioNeMo Inference Runtime
NVIDIA’s BioNeMo Inference Runtime (BioIR) accelerates biomolecular structure‑prediction models on GPUs while preserving the familiar PyTorch workflow by providing an end‑to‑end pipeline that handles
NVIDIA’s BioNeMo Inference Runtime (BioIR) accelerates biomolecular structure‑prediction models on GPUs while preserving the familiar PyTorch workflow by providing an end‑to‑end pipeline that handles parsing, tokenization, feature generation, GPU inference, and PDB/mmCIF output. On a benchmark of 1,000 human dimer targets across eight H100 GPUs, BioIR‑enhanced Boltz‑2 folds residues at 58.5 k GPU‑hours compared to 20.2 k for a torch‑compiled open‑source implementation—a nearly threefold throughput improvement—and reduces estimated energy usage from 35 MWh to 11 MWh for folding one million similar targets. The runtime achieves this via kernel selection, CUDA Graph capture, and Ray‑based pipeline scaling while allowing direct PyTorch module integration for custom code.
Related
- Build an AI Scientist for Life Science Discovery with NVIDIA BioNeMo Agent Toolkit
- How to Accelerate Protein Structure Prediction at Proteome-Scale
Source: NVIDIA Developer | 2026-09-10