A three-dimensional structure model is provided to study the properties and related processes of various crystal, amorphous and polymer materials. The main methods to analyze the structure and properties of matter from the point of view of electrons, atoms and polymers are: quantum mechanics simulation based on density functional theory, molecular mechanics and molecular dynamics simulation, Monte Carlo simulation, mesoscopic dynamics (MesoDyn) and dissipative particle dynamics (DPD), statistical method QSAR and other advanced algorithms and instrumental analysis methods such as X-ray diffraction analysis. According to the type and number of atoms and molecules in the system, the related properties of materials are studied and predicted. The content of the simulation includes the main topics in the field of materials and chemistry, such as catalysts, polymers, solids and surfaces, interfaces, crystals and diffraction, chemical reactions and so on.
How to configure a graphics workstation with high performance (CPU, memory, hard disk, graphics card) needs to analyze its algorithm characteristics and computing scale.
≤ 100atomic ab initio/DFT.
100,200 atoms of semi-experience.
> 200atomic QM/MM model.
In order to find a better algorithm, through the analysis of the algorithm, give the best hardware configuration.
2.2 characteristics of MS material simulation modeling and calculation.
How to ensure not only the fluency of the modeling model, but also the reasonable configuration and the concrete analysis of the ideal calculation speed.
Modeling is to draw three-dimensional graphics, create atomic 3D models, smooth rotation, movement, etc., corresponding to hardware requirements:
CPU: higher dominant frequency.
Memory capacity: ≥ 48GB.
Graphics card: capacity ≥ 8GB, support OpenGL,nvidia graphics card.
Monitor: high resolution is the best.
3840mm 2160 (4K).
19200-4320 (splicing ultra-high resolution).
2.3 Computational characteristics of quantum mechanics algorithms.
The main algorithm is density functional theory (DFT). The calculation specification is proportional to the cubic power of the number of plane waves (N3).
Hardware configuration features: CPU limited multi-core parallelism, memory bandwidth, memory capacity is the focus.
The first category, density functional theory (DFT): large memory capacity, a large number of Icano temporary files.
The second category: plane wave pseudopotential density functional: CPU computing power, memory capacity, memory bandwidth.
If the fastest computing power is pursued, only the four-way Xeon architecture is the most suitable (high frequency + parallel core + 24 channels).
2.4 Computational characteristics of molecular mechanics and molecular dynamics algorithms.
In classical molecular dynamics simulation, the computational cost depends on the number of atoms N, molecular dynamics will not take up too much memory, mainly CPU computing power, due to the output of a large amount of trajectory data, a large capacity of hard disk is needed.
Hardware configuration features: the greater the CPU computing power, the better. GPU Mesocite and Forcite modules support a single GPU card, acceleration, and the hard disk is equipped with large capacity.