This Article 
   
 Share 
   
 Bibliographic References 
   
 Add to: 
 
Digg
Furl
Spurl
Blink
Simpy
Google
Del.icio.us
Y!MyWeb
 
 Search 
   
Your Favorite Parallel Algorithms Might Not Be as Fast as You Think
February 1988 (vol. 37 no. 2)
pp. 211-213
A problem that requires I inputs, K outputs and I computations is to be solved on a d-dimensional parallel processing machine (usually d>or=3). Finite transmission speed and other real-world conditions are assumed. It is proved that the time needed to solve the problem is t= Omega /sub max/ (I/sup 1/d/, K/sup 1/d/, T/sup 1/(d+1)/). This result is demonstrated for the standard algorithm for m

[1] D. C. Fisher, "Matrix computation on processors in one, two and three dimensions," TR-1556, Tech. Rep., Dep. Comput. Sci., Univ. Maryland, Aug. 1985.
[2] D. Foster, private communication, Statistics Program, Univ. Maryland, College Park, 1985.
[3] D. Heller, "A survey of parallel algorithms in numerical linear algebra,"SIAM Rev., vol. 20, pp. 740-711, 1978.
[4] W. L. Miranker and A. Winkler, "Spacetime representations of computational structures,"Computing, vol. 32, pp. 93-114, 1984.
[5] A. Schorr, "Physical parallel devices are not much faster than sequential ones,"Inform. Proc. Lett., vol. 17, pp. 103-106, 1983.
[6] J. D. Ullman,Computational Aspects of VLSI. Rockville, MD: Computer Science Press, 1984.

Index Terms:
computational complexity; finite transmission speed; matrix multiplications; parallel algorithms; real-world conditions; computational complexity; matrix algebra; parallel algorithms.
Citation:
D.C. Fisher, "Your Favorite Parallel Algorithms Might Not Be as Fast as You Think," IEEE Transactions on Computers, vol. 37, no. 2, pp. 211-213, Feb. 1988, doi:10.1109/12.2150
Usage of this product signifies your acceptance of the Terms of Use.