Interaction sorting method for molecular dynamics on multi-core SIMD CPU architecture

Area of Science:

• Computational Biology
• Bioinformatics
• Molecular Modeling

Background:

• Molecular dynamics (MD) simulations are crucial for understanding molecular interactions and biological processes.
• Current MD algorithms face computational expense challenges, driving the need for more efficient methods.
• The interaction sorting (IS) algorithm, developed in 2007, offered a significant performance improvement over existing MD techniques.

Purpose of the Study:

• To introduce a novel modification of the interaction sorting (IS) algorithm.
• To enhance the computational efficiency of molecular dynamics simulations.
• To integrate Single Instruction, Multiple Data (SIMD) processor instructions into the IS algorithm.

Main Methods:

• Development of a modified interaction sorting (IS) algorithm.
• Implementation of SIMD processor instructions within the modified IS algorithm.
• Performance benchmarking of the new IS modification against the original IS method.

Main Results:

• The modified IS algorithm successfully utilizes SIMD processor instructions.
• Performance gains ranging from 9% to 45% were observed compared to the original IS method.
• The enhanced algorithm demonstrates increased efficiency for molecular dynamics simulations.

Conclusions:

• The proposed IS modification offers a substantial performance improvement for molecular dynamics.
• Leveraging SIMD instructions is an effective strategy for accelerating computational biology algorithms.
• This advancement contributes to more efficient and accessible molecular dynamics simulations.