Topic: models of parallel computation
| |
Reference
Grama, A.Y., Gupta, A., Kumar, V.,
"Isoefficiency: measuring the scalability of parallel algorithms and architectures",
IEEE Parallel & Distributed Technology, August 1993, pp. 12-21.
Google
Quotations
13 ;;Quote: a scalable parallel system preserves efficiency if both processors and problem size increase together
| 14 ;;Quote: a parallel system consists of a parallel algorithm and a parallel architecture; they cannot be separated
| 14+;;Quote: the total overhead of a parallel system is a function of problem size and number of processors; overhead due to idling, communication, contention, etc.
| 14 ;;Quote: the isoefficiency function balances problem size with number of processors; a highly scalable system allows small increments in problem size to use increasing numbers of processors efficiently
| 15 ;;Quote: a parallel system is cost-optimal if the total parallel time is proportional to the best serial time
| 16 ;;Quote: use isoefficiency to compare two parallel algorithms; low overhead is not OK if it limits concurrency
|
Related Topics
Topic: models of parallel computation (33 items)
|