Dissertation > Excellent graduate degree dissertation topics show

Research and Implement on Adaptive Communication Mechanism of High-performance RTI

Author: LiMeng
Tutor: YaoYiPing
School: National University of Defense Science and Technology
Course: Computer Science and Technology
Keywords: High-performance computers RTI Shared Memory Adaptive
CLC: TP391.9
Type: Master's thesis
Year: 2011
Downloads: 19
Quote: 1
Read: Download Dissertation

Abstract


Level architecture HLA-based distributed simulation technology has been widely used in various fields of national defense and economic construction. The urgent need for simulation efficiency, due to the the simulation scale simulation complexity improve application units try to transplant HLA applications to high-performance computers to improve the operating efficiency of the simulation system. High-performance computing platforms typically provide a variety of different levels of high-performance hardware communications environment (such as shared memory, Infiniband, Ethernet, etc.), however the traditional HLA the run support environmental RTI general only supports TCP / IP communication, it is difficult to take full advantage of Efficient hardware communication support for high-performance computers. Therefore, the study of RTI (HPC-RTI) for high-performance computing environment the efficient adaptive communication mechanism has important theoretical and practical significance for improving the the HLA simulation run performance and scalability. Papers in traditional RTI communication mechanisms on the basis of a comprehensive analysis of the shared memory for high-performance RTI / IB / Ethernet adaptive communication mechanism in-depth analysis and research, the main work and innovation, including: 1) At present, traditional RTI communication framework is mostly confined to support TCP / IP communications, the design of the structure is relatively simple, it is difficult to effective use of high-performance computers and efficient hierarchical hardware communication support. To this end, the paper for HPC-RTI network environment characteristics underlying communication framework for HPC environments, the communication framework oriented to high-performance computing platform-centric shared memory / IB / Ethernet communication environment lower automatically choose the best means of communication, as much as possible to minimize the communication delay; communication to achieve the upper package, which provides a framework to achieve efficient communication HPC-RTI support. 2) of the RTI platform supported single way of communication, it is difficult to effectively play the advantage of high-speed interconnect for high-performance computing nodes within the shared memory communication between nodes IB. Solve the problem, the paper proposed HPC-RTI shared memory / IB / Ethernet adaptive communication algorithms, the algorithm can determine the best means of communication communications network information, thus achieving the shared memory, high-speed IB Ethernet adaptive selection. 3) shared memory as a communication mechanism between the high-speed multi-processor (core) to run on a single high-performance node interprocess communication to provide the highest efficiency. Usually shared memory communication is often more than one process to communicate using the same memory area, the read-write conflict is likely to cause increased when increased communication process, resulting in system lock resource waits, resulting in reduced efficiency of shared memory communication. To this end, the thesis focuses on the application of distributed simulation multi-Leaguers interconnect traffic characteristics, a based the RTI the Leaguers communication algorithms on distributed shared memory, the algorithm creates a message for each federate the node receives shared memory region , while ensuring the consistency of memory access, can effectively reduce the shared memory access conflicts between the Leaguers, the test results show that, on a quad-core high-performance computing nodes, starting four Leaguers, one Leaguers to the other three The Leaguers send attribute update message distributed shared memory algorithms attribute the delay only centralized shared memory algorithms 63%, significantly improve the efficiency of shared memory communication. Based on the above, the design and implementation of a high-performance RTI adaptive communication subsystem ACS (Adaptive Communication System), after tests showed that the high-performance computing environments, compared with the RTI only supports TCP / IP in the past, based on ACS RTI node Leaguers property update delay can be reduced by 62% or more properties throughput increased more than 2.7 times.

Related Dissertations

  1. Design and Research on Stabilization Loop of Gyro Stabilized Pod Control System,V241.5
  2. Robust Adaptive Control Algorithms Research for the System of Spacecraft Attitude Dynamics,V448.22
  3. Research on Removal Algorithm of Shadows in Image Segmentation,TP391.41
  4. Study on Nonlinear Variable Structure Guidance Law,TJ765
  5. Active Power Filter and Its Application in Distribution Network,TN713.8
  6. An Adaptive Processing Node Selection Based Spatio-Temporal Query Alogrithm,TN929.5
  7. Body Sensor Networks Adaptive Communication Protocol,TP212.9
  8. Design and Implementation of PSTN multi signaling point semi-physical simulation system,TP391.9
  9. Research on the Current Situation, Methods and Improvement of Adaptive Reuse of the Old Industrial Buildings in Wuhan,TU984.114
  10. Fault Tolerance for MapReduce in the Cloud Environment,TP302.8
  11. Situaiton Assessment Using Probabilistic Graphical Models,E917
  12. SLA-based cloud service provides a framework for adaptive research and implementation,TP311.52
  13. Event Driven Monitoring of Cyber-physical Systems Based on Hybrid Automata,TP311.52
  14. The Research on Control Precision Model of SCARA Robot for Multi-point High-speed Assembly,TP242.2
  15. Study on Influencing Factors of Knowledge Sharing in Supply Chain under CAS Theory,F224
  16. Service-based workflow template adaptive modeling and dynamic execution of,TP311.52
  17. Channel wireless Adhoc network protocol design and research,TN929.5
  18. Based on the sampling timing synchronization symbol rate study,TN792
  19. Mine 35KV power grid optimization,TM727
  20. Based on permanent magnet synchronous gearless traction machine energy-saving elevator control system algorithms and software design,TM341

CLC: > Industrial Technology > Automation technology,computer technology > Computing technology,computer technology > Computer applications > Information processing (information processing) > Computer simulation
© 2012 www.DissertationTopic.Net  Mobile