Desmond

Desmond performs high-speed molecular dynamics (MD) simulations of biomolecular systems to enable exploration of long-timescale biological events and drug interactions.

Key Features:

• Novel Parallel Algorithms: Employs parallel decomposition methods that reduce inter-processor communication and enable scaling to large processor counts.
• Advanced Communication Techniques: Incorporates message-passing techniques and novel communication primitives to minimize communication time in large-scale simulations.
• High Accuracy with Single Precision Computation: Uses numerical techniques that maintain high accuracy while leveraging single-precision arithmetic and processor-level vector instructions.
• Performance and Scalability: Delivers high simulation throughput on commodity clusters, with reported benchmarks on an Opteron cluster (2K processors) slightly outperforming Blue Matter MD code on IBM Blue Gene/L (32K processors).

Scientific Applications:

• Long-timescale biomolecular dynamics: Enables simulation of biomolecular systems over extended timescales to observe events inaccessible to shorter simulations.
• Drug discovery and pharmaceutical development: Supports studies of drug–target interactions and processes relevant to pharmaceutical research.
• Exploration of complex biological phenomena: Facilitates investigation of complex biological behaviors that require large-scale, long-duration MD simulations.

Methodology:

Employs parallel decomposition, message-passing techniques, novel communication primitives, numerical methods for accuracy with single-precision computation, and exploitation of processor-level vector instructions.