Final Report on Data Locality Enhancement of Dynamic Simulations for Exascale Computing

Teaching feedback control theory is challenging because it is important to cover theoretical material intended for fundamental understanding as well as material directly related to industrial practice. One approach to reach this dual objective and prevent control theory from becoming abstract to students is to assign a design project that requires integration of all main concepts taught in class. This approach has been successfully used in eight offerings of the course ChE 3601 Process Dynamics and Control in the Chemical Engineering program at the University of New Brunswick. The one-semester course is an introduction to the dynamic behavior of chemical processes and feedback control loops. The project is assigned at the beginning of the course and involves the design of a feedback control system for a realistic chemical process. The design project is divided into five milestones with deliverables due every two weeks. The final report due at the end of the course must include a description of the proposed system using a P&I diagram, specifications for all control equipment, a dynamic model for all components of the feedback loop, settings for the tuning parameters of the PID controller, and dynamic simulations using Polymath to validate the proposed solution. The course is organized around the project in a manner similar to that used in problem-based learning. The active learning approach used in ChE 3601 provides a deeper understanding of control theory and its application.

Download Full-text

Final Report on XStack: Software Synthesis for High Productivity ExaScale Computing

10.2172/1261628 ◽

2016 ◽

Author(s):

Armando Solar-Lezama ◽

Keyword(s):

Final Report ◽

Software Synthesis ◽

High Productivity ◽

Exascale Computing

Download Full-text

Reduced order modelling for dynamic simulations: LDRD feasibility study final report

10.2172/1059464 ◽

2012 ◽

Author(s):

K Chand ◽

W Henshaw ◽

T Vassilevska

Keyword(s):

Feasibility Study ◽

Final Report ◽

Dynamic Simulations ◽

Reduced Order ◽

Reduced Order Modelling

Download Full-text

Locality properties of 3D data orderings with application to parallel molecular dynamics simulations

The International Journal of High Performance Computing Applications ◽

10.1177/1094342019846282 ◽

2019 ◽

Vol 33 (5) ◽

pp. 998-1018

Author(s):

Ibrahim Al-Kharusi ◽

David W Walker

Keyword(s):

Molecular Dynamics ◽

Molecular Dynamics Simulations ◽

Spatial Data ◽

Data Locality ◽

Data Reuse ◽

Dynamic Simulations ◽

Data Movement ◽

Execution Speed ◽

Dynamics Simulations ◽

The Impact

Application performance on graphical processing units (GPUs), in terms of execution speed and memory usage, depends on the efficient use of hierarchical memory. It is expected that enhancing data locality in molecular dynamic simulations will lower the cost of data movement across the GPU memory hierarchy. The work presented in this article analyses the spatial data locality and data reuse characteristics for row-major, Hilbert and Morton orderings and the impact these have on the performance of molecular dynamics simulations. A simple cache model is presented, and this is found to give results that are consistent with the timing results for the particle force computation obtained on NVidia GeForce GTX960 and Tesla P100 GPUs. Further analysis of the observed memory use, in terms of cache hits and the number of memory transactions, provides a more detailed explanation of execution behaviour for the different orderings. To the best of our knowledge, this is the first study to investigate memory analysis and data locality issues for molecular dynamics simulations of Lennard-Jones fluids on NVidia’s Maxwell and Tesla architectures.

Download Full-text