Biagini Tommaso, Petrizzelli Francesco, Truglio Mauro, Cespa Roberto, Barbieri Alessandro, Capocefalo Daniele, Castellana Stefano, Tevy Maria Florencia, Carella Massimo, Mazza Tommaso
Bioinformatics Unit, IRCCS Casa Sollievo della Sofferenza, Roma, Italy.
ICT, Innovation and Research Unit, IRCCS Casa Sollievo della Sofferenza, Roma, Italy.
Evol Bioinform Online. 2019 May 22;15:1176934319850144. doi: 10.1177/1176934319850144. eCollection 2019.
In several fields of research, molecular dynamics simulation techniques are exploited to evaluate the temporal motion of particles constituting water, ions, small molecules, macromolecules, or more complex systems over time. These techniques are considered difficult to setup, computationally demanding and require high specialization and scientific skills. Moreover, they need specialized computing infrastructures to run faster and make the simulation of big systems feasible. Here, we have simulated 3 systems of increasing sizes on scientific- and gaming-enabled graphic processing unit (GPU) cards with Amber, GROMACS, and NAMD and measured their performance accounting also for the market prices of the GPU cards where they were run on.
在多个研究领域中,分子动力学模拟技术被用于评估构成水、离子、小分子、大分子或更复杂系统的粒子随时间的动态运动。这些技术被认为难以设置,计算要求高,并且需要高度专业化的科学技能。此外,它们需要专门的计算基础设施才能运行得更快,并使大型系统的模拟变得可行。在这里,我们使用Amber、GROMACS和NAMD在支持科学计算和游戏的图形处理单元(GPU)卡上模拟了3个规模不断增大的系统,并在考虑运行这些系统所使用的GPU卡市场价格的情况下测量了它们的性能。