Accelerating Drug Discovery in AutoDock-GPU with Tensor Cores
KTH Royal Institute of Technology, Stockholm, Sweden
arXiv:2410.10447 [cs.DC], (14 Oct 2024)
@misc{schieffer2024acceleratingdrugdiscoveryautodockgpu,
title={Accelerating Drug Discovery in AutoDock-GPU with Tensor Cores},
author={Gabin Schieffer and Ivy Peng},
year={2024},
eprint={2410.10447},
archivePrefix={arXiv},
primaryClass={cs.DC},
url={https://arxiv.org/abs/2410.10447}
}
In drug discovery, molecular docking aims at characterizing the binding of a drug-like molecule to a macromolecule. AutoDock-GPU, a state-of-the-art docking software, estimates the geometrical conformation of a docked ligand-protein complex by minimizing a scoring function. Our profiling results indicate that the current reduction operation that is heavily used in the scoring function is sub-optimal. Thus, we developed a method to accelerate the sum reduction of four-element vectors using matrix operations on NVIDIA Tensor Cores. We integrated the new reduction operation into AutoDock-GPU and evaluated it on multiple chemical complexes on three GPUs. Our results show that our method for reduction operation is 4-7 times faster than the AutoDock-GPU baseline. We also evaluated the impact of our method on the overall simulation time in the real-world docking simulation and achieved a 27% improvement on the average docking time.
October 20, 2024 by hgpu
