Accelerating VASP electronic structure calculations using graphic processing units.

Area of Science:

• Computational Chemistry
• Materials Science
• High-Performance Computing

Background:

• The Vienna Ab initio Simulation Package (VASP) is a widely used program for electronic structure calculations.
• Computational demands of VASP can be a bottleneck for large-scale simulations.

Purpose of the Study:

• To enhance the performance of the VASP program by leveraging graphics processing units (GPUs).
• To reduce computation time for electronic structure simulations through GPU acceleration.

Main Methods:

• Profiling VASP code to identify time-consuming sections.
• Rewriting algorithms for better compatibility with GPU accelerators (NVIDIA GPUs using CUDA).
• Optimizing memory traffic between host computers and GPU accelerators.

Main Results:

• Significant reduction in computation time by a factor of 3 to 8.
• Achieved speedup by offloading computations to GPUs on systems with up to 1100 atoms.
• No loss of accuracy was observed in the GPU-accelerated version.

Conclusions:

• GPU acceleration is an effective strategy for improving VASP performance.
• Offloading specific computational tasks to GPUs drastically reduces simulation times for electronic structure calculations.