Title :
Shared memory multistage clustering structure, an efficient structure for massively parallel processing systems
Author :
Shahhosein, H.S. ; Naderi, M. ; Buyya, R.
Author_Institution :
Dept. of Electr. Eng., Iran Univ. of Sci. & Technol., Tehran, Iran
Abstract :
A novel structure is proposed for massive parallel processing systems. This structure is expandable in a vertical and horizontal manner and covers many of the previous computer designs. Queuing theory and the Jackson queuing network are applied for constructing an analytical model for the proposed structure. This model gives a closed-form solution for the system performance metrics, such as processor waiting time, system processing power, and so on. Two novel points in development of these analytical models are: application of open queuing network rules for analyzing a closed queuing network, and calculation of the input rate of each service center as a function of the input rate for the previous center. The model can be used for evaluating the MPP system or optimizing its specification on design space.
Keywords :
formal specification; parallel processing; queueing theory; shared memory systems; Jackson queuing network; MPP; MPP system; analytical model; closed queuing network; closed-form solution; computer designs; design space; input rate; massively parallel processing systems; open queuing network rules; processor waiting time; queuing theory; service center; shared memory multistage clustering structure; system performance metrics; system processing power;
Conference_Titel :
High Performance Computing in the Asia-Pacific Region, 2000. Proceedings. The Fourth International Conference/Exhibition on
Conference_Location :
Beijing, China
Print_ISBN :
0-7695-0589-2
DOI :
10.1109/HPC.2000.846510