AI supercomputing speeds up chemistry calculations 80x
Collaborators SandboxAQ and NVIDIA have reported a breakthrough that allows an 80-fold speed increase in computational chemistry calculations.
SandboxAQ leverages Large Quantitative Models (LQMs) and the NVIDIA CUDA-accelerated Density Matrix Renormalization Group (DMRG) algorithm.
Combining the CUDA-DMRG algorithm, the NVIDIA Quantum platform, and NVIDIA accelerated computing speeds up calculations more than 80x, compared with traditional 128-core CPU computations.
It also more than doubles the sizes of computable catalysts and enzyme active sites calculated by the system. SandboxAQ researchers use these computational results to train AI networks to optimise for the desired treatment or catalyst.
“AI supercomputing is helping to solve critical problems in the chemical and pharmaceutical industries,” said Tim Costa, Director of High-Performance and Quantum Computing at NVIDIA. “SandboxAQ’s use of the NVIDIA Quantum platform is facilitating simulations at an unprecedented scale, enabling scientists to rethink what’s possible in computational chemistry.”
Last year, SandboxAQ announced AI collaborations with the University of California San Francisco (UCSF), NOVONIX, and Riboscience. In early 2024, it acquired Good Chemistry, a drug discovery computational chemistry company.
“Advanced computing is opening new frontiers in scientific research. Our use of NVIDIA technology has allowed us to address some of the most challenging problems in chemistry,” said Dr Martin Ganahl, Senior Staff Scientist at SandboxAQ. “We are not only advancing our understanding of material science and chemistry, but also paving the way for the next wave of innovations in drug discovery and catalysis to tackle currently-untreatable conditions and find safer and cheaper ways to synthesize molecules and materials.”
Applications for LQMs range from biopharma to agriculture to advanced materials. In biopharma for example, the enzyme Cytochrome P450s plays a central role in human drug metabolism and is central to understanding drug toxicity. CUDA-DMRG can help solve the long-standing problem of accurately modelling cytochromes’ catalytic activity, allowing computational simulation to de-risk clinical trials before they happen.
Diana Spencer, Senior Digital Content Editor, DDW
