System-Level Design:Imprecise Computation
Imprecise Computation
The main results in imprecise computation theory are due to Liu et al. [66] who developed polynomial time algorithms for optimal scheduling of preemptive tasks on homogeneous multiprocessors without communications costs. Ho et al. [67] proposed an approach to minimize the total error, where the error of a task being imprecisely executed is proportional to the amount of time that its optional part was not allowed to execute, i.e., the time still needed for its full completion. Polynomial time-optimal algorithms were derived for some instances of the problem [66].
Tirat-Gefen et al. [68] presented in 1997 a new approach for ASHM design that allows trade-offs between cost, performance, and data-quality through incorporation of imprecise computation into the system-level design cycle.
Comments
Post a Comment