Exploiting OpenMP and OpenACC to accelerate a geometric approach to molecular docking in heterogeneous HPC nodes

Abstract

In drug discovery, molecular docking is the task in charge of estimating the position of a molecule when interacting with the docking site. This task is usually used to perform screening of a large library of molecules, in the early phase of the process. Given the amount of candidate molecules and the complexity of the application, this task is usually performed using High-Performance Computing (HPC) platforms. In modern HPC systems, heterogeneous platforms provide a better throughput with respect to homogeneous platforms.In this work, we ported and optimized a molecular docking application to a heterogeneous system, with one or more GPU accelerators, leveraging a hybrid OpenMP and OpenACC approach. We prove that our approach has a better exploitation of the node compared to pure CPU/GPU data splitting approaches, reaching a throughput improvement up to 36% while considering the same computing node.

An extreme-scale virtual screening platform for drug discovery

“…To enable the deployment on any available type of architecture, we are currently moving the accelerated part using SYCL. Already in the past, we faced this problem using OpenACC [4]. However, we are finding the new SYCL standard a more suitable opportunity for our goals.…”

EXSCALATE: An extreme-scale in-silico virtual screening platform to evaluate 1 trillion compounds in 60 hours on 81 PFLOPS supercomputers

“…The score of the ligand is the score of the best pose that we found. We implemented the algorithm in C++17 to target CPUs, while we implmented an OpenACC and CUDA implementation to target accelerators [25]. The CUDA implementation is the fastest when we target NVIDIA GPUs.…”

Understanding the I/O Impact on the Performance of High-Throughput Molecular Docking

“…These tools use OpenMP for multithreaded molecular docking on a single node. Autodock has also port to use accelerator with OpenCL [15], Nvidia CUDA [3] and OpenACC [16], [17]. VinaMPI is an MPI wrapper of Autodock Vina to use MPI on extreme scale supercomputers [18].…”