Результаты (
английский) 2:
[копия]Скопировано!
The methods
mentioned above can partition the network simulation task for homogeneous computing environment reasonably as load balancing, communication overhead, and
lookahead are all taken into consideration, yet these methods can hardly be used
for partitioning the simulation task in heterogeneous computing environment, as
they are lack of consideration about computation ability of different machines. The
work in Ref.7proposes moving entities between machines during the running time
to reduce the number of remote queues, which is a major source of slowing down in
parallel event simulations, yet in this paper, we focus on both the load balancing
and the communication overhead.
переводится, пожалуйста, подождите..