home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.parallel
- Path: sparky!uunet!gatech!hubcap!fpst
- From: sun@elc04.icase.edu (Dr. Xian-He Sun)
- Subject: Re: definition of parallel efficiency
- Message-ID: <1993Jan22.161229.20180@hubcap.clemson.edu>
- Sender: fpst@hubcap.clemson.edu (Steve Stevenson)
- Nntp-Posting-Host: elc04.icase.edu
- Organization: ICASE/NASA Langley
- References: <1993Jan11.142638.7789@hubcap.clemson.edu>
- Date: 21 Jan 1993 21:21:59 GMT
- Approved: parallel@hubcap.clemson.edu
- Lines: 91
-
-
- Parallel efficiency is defined as the quotient of the speedup
- and the number of processors used, p. Since there are two
- different definitions of speedup, there are two different
- definitions of efficiency [1], [2], [3]. In one definition speedup
- is defined as
-
- (1) (seq. time of the par. algorithm) / parallel time ,
-
- which is called absolute speedup. In another speedup is defined as
-
- (2) (seq. time of the best possible sequential algorithm
- for the problem) / parallel time,
-
- which is called relative speedup. Definition (1) is a good
- measurement of parallel processing gain versus sequential
- processing. Definition (2) is a good measurement of the
- inherent parallelism of the parallel algorithm. In practice,
- (1) is oftenly used in newly designed parallel algorithms, and
- (2) is used in performance and scalability analysis [4]. The
- well known Amdahl's law and Gustafson's scaled speedup [5],[6] are
- all based on definition (2).
-
-
- Xian-He Sun
-
- ICASE
- (Institute for Computer Applications in Science and Engineering)
- Mail Stop 132C 804-864-8018 (O)
- NASA Langley Research Center 804-864-6134 (fax)
- Hampton, VA 23681-0001 sun@icase.edu
-
- ------------------------------------------------------------------
- Reference:
-
- @INPROCEEDINGS{SuNi90,
- author = "Xian-He Sun and L. Ni",
- title = "Another View on Parallel Speedup",
- booktitle = "Proc. of Supercomputing'90",
- address = "NY, NY",
- year = "1990",
- pages = "324--333",
- }
-
- @BOOKLET{SuNi92,
- author = "Xian-He Sun and L. Ni",
- title = "Scalable Problems and Memory-Bounded Speedup",
- howpublished = "ICASE Technical Report, 92-59",
- address = "ICASE, NASA Langley Research Center",
- year = "1992",
- }
-
- @ARTICLE{Hock91,
- author = "Roger W. Hockney",
- title = "Performance Parameters and Benchmarking of
- Supercomputers",
- journal = "Parallel Computing",
- Note = "This issue of Parallel Computing",
- }
-
- @BOOKLET{SuRo91,
- author = "Xian-He Sun and D.T. Rover",
- title ="Scalability of Parallel Algorithm-Machine Combinations",
- howpublished = "Technical Report, IS-5057, UC-32",
- address = "Ames Laboratory, U.S. Department of Energy",
- year = "1991",
- }
-
- @ARTICLE{GuMB88,
- author = "J.L. Gustafson and G.R. Montry and R.E. Benner",
- title = "Development of Parallel Methods for a 1024-processor
- Hypercube",
- journal = "SIAM J. on SSTC",
- Volume = 9,
- YEAR = "1988",
- month = "July",
- number = "4",
- }
-
- @ARTICLE{SuGu91,
- AUTHOR = "Xian-He Sun and J.L. Gustafson",
- TITLE = "Toward a Better Parallel Performance Metric",
- JOURNAL = "Parallel Computing",
- VOLUME = "17",
- MONTH = "Dec",
- YEAR = "1991",
- pages = "1093--1109",
- }
-
-
-
-