Data Aware Simulation of Complex Systems on GPUs

GPUs have been demonstrated to be highly effective at improving the performance of Multi-Agent Systems (MAS). One of the major limitations of further performance improvements is in the memory bandwidth required to move agent data through the GPU’s memory hierarchy. This paper presents a formal model for data aware simulation and an empirical study into the impact of minimising data movement on performance. This study proposes a method that can be applied to the simulation of complex systems on GPUs to extract required data from agent behaviour during simulation time and how this information can be used to reduce data movement. The FLAME GPU software has been extended to demonstrate this technique. Three benchmark experiments have been applied to evaluate the overall reduction in simulation execution time under specific criteria. The results of the comparison between the current and new system show that reducing data movement within a simulation improves overall performance with up to 4.8x speedup reported.

[1]  Daniela M. Romano,et al.  Template-Driven Agent-Based Modeling and Simulation with CUDA , 2011 .

[2]  Craig W. Reynolds.  Big fast crowds on PS3 , 2006, Sandbox '06.

[3]  Vittorio Scarano,et al.  Massive Simulation using GPU of a distributed behavioral model of a flock with obstacle avoidance , 2004, VMV.

[4]  Nishchol Mishra,et al.  Load Balancing Techniques: Need, Objectives and Major Challenges in Cloud Computing- A Systematic Review , 2015 .

[5]  Bo Li.  A Comparative Analysis of Spatial Partitioning Methods for Large-scale , Real-time Crowd Simulation , 2014 .

[6]  P. Machanick.  Approaches to Addressing the Memory Wall , 2022 .

[7]  Michael Lees,et al.  PDES-MAS: Distributed Simulation of Multi-Agent Systems , 2013, ICCS.

[8]  Priya R. Deshpande,et al.  Load Balancing in Cloud Computing , 2014 .

[9]  Pierre Lemarinier,et al.  Agent Based Modelling and Simulation tools: A review of the state-of-art software , 2017, Comput. Sci. Rev..

[10]  Georgios Theodoropoulos,et al.  Synchronised Range Queries in Distributed Simulations of Multi-agent Systems , 2013, 2010 IEEE/ACM 14th International Symposium on Distributed Simulation and Real Time Applications.

[11]  Georgios K. Theodoropoulos,et al.  Distributing RePast agent‐based simulations with HLA , 2008, Concurr. Comput. Pract. Exp..

[12]  Mike Holcombe,et al.  Using X-Machines as a Formal Basis for Describing Agents in Agent-Based Modelling , 2006 .

[13]  Daniela M. Romano,et al.  High performance cellular level agent-based simulation with FLAME for the GPU , 2010, Briefings Bioinform..

[14]  Laurent Philippe,et al.  Using GPU for Multi-agent Multi-scale Simulations , 2012, DCAI.

[15]  Paul Richmond,et al.  Agent Based GPU , a Real-time 3 D Simulation and Interactive Visualisation Framework for Massive Agent Based Modelling on the GPU , 2008 .

[16]  Rajwinder Kaur,et al.  Load Balancing in Cloud Computing , 2014 .

[17]  Paulo Leitao,et al.  Simulation of multi-agent manufacturing systems using Agent-Based Modelling platforms , 2011, 2011 9th IEEE International Conference on Industrial Informatics.

[18]  Paul Richmond,et al.  FLAME GPU: Complex System Simulation Framework , 2017, 2017 International Conference on High Performance Computing & Simulation (HPCS).

[19]  Rutger F. H. Hofman,et al.  Bandwidth and Latency Sensitivity of Parallel Applications in a Wide-Area System , 1998 .

[20]  Eduard Ayguadé,et al.  Reducing Data Movement on Large Shared Memory Systems by Exploiting Computation Dependencies , 2018, ICS.

[21]  M. Pipattanasomporn,et al.  Multi-agent systems in a distributed smart grid: Design and implementation , 2009, 2009 IEEE/PES Power Systems Conference and Exposition.

[22]  Eidah Alzahrani,et al.  A Formula-Driven Scalable Benchmark Model for ABM, Applied to FLAME GPU , 2017, Euro-Par Workshops.

[23]  Osamu Tatebe,et al.  Workflow Scheduling to Minimize Data Movement Using Multi-constraint Graph Partitioning , 2012, 2012 12th IEEE/ACM International Symposium on Cluster, Cloud and Grid Computing (ccgrid 2012).

[24]  Stephen John Turner,et al.  Large Scale Distributed Simulation on the Grid , 2006, Sixth IEEE International Symposium on Cluster Computing and the Grid (CCGRID'06).

[25]  Marian Gheorghe,et al.  Communicating Stream X-Machines Systems are no more than X-Machines , 1999, J. Univers. Comput. Sci..

[26]  Poonam Goyal,et al.  A Fast, Scalable SLINK Algorithm for Commodity Cluster Computing Exploiting Spatial Locality , 2016, 2016 IEEE 18th International Conference on High Performance Computing and Communications; IEEE 14th International Conference on Smart City; IEEE 2nd International Conference on Data Science and Systems (HPCC/SmartCity/DSS).

[27]  Phil McMinn,et al.  An integrated systems biology approach to understanding the rules of keratinocyte colony formation , 2007, Journal of The Royal Society Interface.

[28]  Alcione de Paiva Oliveira,et al.  Feasibility Study of Multi-Agent Simulation at the Cellular Level with FLAME GPU , 2016, FLAIRS Conference.

[29]  Daniela M. Romano,et al.  A high performance agent based modelling framework on graphics card hardware with CUDA , 2009, AAMAS.

[30]  Prabhat Kr. Srivastava,et al.  Improving Performance in Load Balancing Problem on the Grid Computing System , 2011 .

[31]  Marian Gheorghe,et al.  Large-Scale Simulations with FLAME , 2016 .