User: Guest  Login
Original title:
A Framework to Generate High-Performance Time-stepped Agent-based Simulations on Heterogeneous Hardware
Translated title:
Ein Framework zur Generierung von zeitdiskreten und agentenbasierten Hochleistungssimulationen für heterogene Hardwareumgebungen
Author:
Xiao, Jiajian
Year:
2022
Document type:
Dissertation
Faculty/School:
TUM School of Computation, Information and Technology
Advisor:
Knoll, Alois Christian (Prof. Dr. habil.)
Referee:
Knoll, Alois Christian (Prof. Dr. habil.); Cai, Wentong (Prof., Ph.D.)
Language:
en
Subject group:
NAT Naturwissenschaften (allgemein)
Keywords:
Agent-based Simulation, High-Performance Computing, Heterogeneous hardware, GPU, FPGA, CPU, OpenCL
Translated keywords:
Agentenbasierte Simulation,High-Performance Computing, Heterogene Hardware, GPU, FPGA, CPU, OpenCL
TUM classification:
DAT 260; DAT 815
Abstract:
Agent-Based Simulation has shown ability. However, ABS suffers from long execution times. Using heterogeneous hardware is a solution. However, significant effort is required to achieve high performance. To simplify this, OpenABLext is proposed. It generates high-performance ABS programs from sequential representations. Further, OptCL is also introduced to enable co-execution. We believe the framework consisting of OpenABLext and OptCL can simplify the future development of high-performance ABSs.
Translated abstract:
Die agentenbasierte Simulation ist ein Ansatz zur Systembewertung, aber sie oft unter langen Ausführungszeiten leiden. Durch den Einsatz heterogener Hardware kann dieses Problem reduzieren. Um die Programmierung zu vereinfachen, wird in dieser Arbeit das Framework OpenABLext und eine Middleware namens OptCL entwickelt. Das vorgestellte Framework bestehend aus OpenABLext und OptCL vereinfacht die Entwicklung von hoch-performanten agentenbasierten Simulationen erheblich.
WWW:
https://mediatum.ub.tum.de/?id=1652866
Date of submission:
11.04.2022
Oral examination:
04.10.2022
File size:
2651796 bytes
Pages:
168
Urn (citeable URL):
https://nbn-resolving.de/urn/resolver.pl?urn:nbn:de:bvb:91-diss-20221004-1652866-1-8
Last change:
09.12.2022
 BibTeX