TY - GEN
T1 - Dynamic load balancing in embedded systems based on Triplet-based hierarchical interconnection architecture
AU - Liu, Bin
AU - Gao, Yu Jin
PY - 2006
Y1 - 2006
N2 - Dynamic load balancing (DLB) for conventional parallel multiprocessor systems has been studied extensively; however, DLB for parallel multiprocessor embedded systems is relatively new. To efficiently utilize computing resources provided by parallel multiprocessor embedded systems, the underlying DLB algorithms must be designed deeply considering both architectural and dynamic load features of the systems. In this paper, a new type of parallel multiprocessor embedded system architectures - Triplet-based Hierarchical Interconnection Architecture (THIA) is presented along with the corresponding DLB algorithm (THIA-DLBA). THIA-DLBA is distributed controlled and sender initiated. And it makes use of the vector load index, load migration operator and message passing mechanism to help systems achieve load balance. To compare the efficiency and performance of TBHA-DLBA and other three DLB algorithms, a THIA embedded system simulator and matrix computation were considered as the simulation system. The results indicate that THIA-DLBA can help embedded systems achieve high-performance in dealing with compute-intensive tasks.
AB - Dynamic load balancing (DLB) for conventional parallel multiprocessor systems has been studied extensively; however, DLB for parallel multiprocessor embedded systems is relatively new. To efficiently utilize computing resources provided by parallel multiprocessor embedded systems, the underlying DLB algorithms must be designed deeply considering both architectural and dynamic load features of the systems. In this paper, a new type of parallel multiprocessor embedded system architectures - Triplet-based Hierarchical Interconnection Architecture (THIA) is presented along with the corresponding DLB algorithm (THIA-DLBA). THIA-DLBA is distributed controlled and sender initiated. And it makes use of the vector load index, load migration operator and message passing mechanism to help systems achieve load balance. To compare the efficiency and performance of TBHA-DLBA and other three DLB algorithms, a THIA embedded system simulator and matrix computation were considered as the simulation system. The results indicate that THIA-DLBA can help embedded systems achieve high-performance in dealing with compute-intensive tasks.
UR - http://www.scopus.com/inward/record.url?scp=44949148439&partnerID=8YFLogxK
U2 - 10.1109/MESA.2006.297010
DO - 10.1109/MESA.2006.297010
M3 - Conference contribution
AN - SCOPUS:44949148439
SN - 0780397215
SN - 9780780397217
T3 - Proceedings of the 2nd IEEE/ASME International Conference on Mechatronic and Embedded Systems and Applications, MESA 2006
BT - Proceedings of the 2nd IEEE/ASME International Conference on Mechatronic and mbedded Systems and Applications, MESA 2006
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 2nd IEEE/ASME International Conference on Mechatronic and Embedded Systems and Applications, MESA 2006
Y2 - 13 August 2006 through 16 August 2006
ER -