MDBenchmark

MDBenchmark optimizes molecular dynamics (MD) simulation performance and resource utilization by benchmarking and tuning hardware-specific parameters for scalable simulations.

Key Features:

• Streamlined Benchmarking Process: Supports setup, submission, and analysis of simulation benchmarks and scaling studies and is compatible with multiple MD engines and job queuing systems.
• Optimization Across Diverse Hardware Architectures: Tunes message passing interface (MPI) ranks and open multi-processing (OpenMP) threads for CPU-only and mixed CPU-GPU nodes.
• Performance Analysis and Comparison: Measures and compares performance across different MD systems and hardware configurations, demonstrated on 23 MD simulation systems using GROMACS 2018.
• Scalability Studies: Performs scaling studies across varying node counts and evaluates the efficiency of running multiple simulations on a single node to assess time-to-solution and computational efficiency.

Scientific Applications:

• Performance and Resource Optimization: Enables optimization of MD simulation speed and resource utilization.
• Cost and Energy Efficiency Assessment: Supports reduction of monetary, energetic, and environmental costs associated with large-scale simulations.
• Biophysics: Facilitates scalable MD studies in biophysics.
• Materials Science: Facilitates scalable MD studies in materials science.
• Drug Discovery: Facilitates scalable MD studies in drug discovery.

Methodology:

Sets up, executes, and analyzes benchmarks; manages simulation parameters tailored to specific hardware configurations; tunes MPI ranks and OpenMP threads for CPU-only and mixed CPU-GPU environments; measures and compares performance across systems and hardware; and performs scaling studies across node counts and multiple simulations per node, demonstrated with 23 systems using GROMACS 2018.