Development of an equation-based parallelization method for multiphase particle-in-cell simulations

نویسندگان

چکیده

Manufacturers have been developing new graphics processing unit (GPU) nodes with large capacity, high bandwidth memory and very intra-node interconnects. This enables moving amounts of data between GPUs on the same node at low cost. However, small packet bandwidths latencies not decreased, which makes global dot products expensive. These characteristics favor a kind problem decomposition called “equation decomposition” rather than traditional domain decomposition. In this approach, each GPU is assigned one equation set to solve in parallel so that frequent expensive product synchronization points distributed linear solvers are eliminated. exchange, method involves infrequent movement state variables over bandwidth, To test theory, our flagship code Multiphase Flow Interphase eXchanges (MFiX) was ported TensorFlow. known as MFiX-AI can produce near identical results original version MFiX significant acceleration multiphase particle-in-cell (MP-PIC) simulations. The performance single 4 NVIDIA A100s connected NVLINK 2.0 shown be competitive 1000 CPU cores (25 nodes) JOULE supercomputer, leading an energy savings up 90%. substantial benefit for small- intermediate-sized problems. expected grow become more powerful. Further, poised accept native artificial intelligence/machine learning models further development.

برای دانلود باید عضویت طلایی داشته باشید

برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

منابع مشابه

development and implementation of an optimized control strategy for induction machine in an electric vehicle

in the area of automotive engineering there is a tendency to more electrification of power train. in this work control of an induction machine for the application of electric vehicle is investigated. through the changing operating point of the machine, adapting the rotor magnetization current seems to be useful to increase the machines efficiency. in the literature there are many approaches wh...

15 صفحه اول

The midpoint method for parallelization of particle simulations.

The evaluation of interactions between nearby particles constitutes the majority of the computational workload involved in classical molecular dynamics (MD) simulations. In this paper, we introduce a new method for the parallelization of range-limited particle interactions that proves particularly suitable to MD applications. Because it applies not only to pairwise interactions but also to inte...

متن کامل

POSIX threads parallelization for example of Particle - In - Cell density calculations in plasma computer simulations Anna

The TRQR program [1–4] simulates trajectories of charged particles (electrons or ions) in the electromagnetic field. TRQR is based on the Particle-In-Cell method whose basic guideline is the use of computational particles (called macro particles) that represent a large number of real particles of the same kind moving in the same direction. The program calculates particles charge density distrib...

متن کامل

POSIX threads parallelization for example of Particle-In-Cell density calculations in plasma computer simulations

The TRQR program [1–4] simulates trajectories of charged particles (electrons or ions) in the electromagnetic field. TRQR is based on the Particle-In-Cell method whose basic guideline is the use of computational particles (called macro particles) that represent a large number of real particles of the same kind moving in the same direction. The program calculates particles charge density distrib...

متن کامل

ذخیره در منابع من


  با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید

ژورنال

عنوان ژورنال: Engineering With Computers

سال: 2022

ISSN: ['0177-0667', '1435-5663']

DOI: https://doi.org/10.1007/s00366-022-01768-6