Information Technology Reference
In-Depth Information
the time required for a single evaluation of a constraint has a huge
impact on overall runtime.
Number of constraints
Besides the time required for evaluating a single constraint, the
number of constraints is another major impact factor on overall
runtime. The overall runtime growths linearly with the number of
constraints to be evaluated.
Within benchmark suite 2, only a single constraint is considered
and the time required for a single evaluation of this constraint is
fixed. Nevertheless, the impact of the presence of constraints is
demonstrated.
9.5.4 Benchmark suite 3: Parallelization on node-level
Benchmark suite 3 is intended to demonstrate potential runtime
reduction by using a parallelized simulation engine.
Setup suite 3
All benchmarks in suite 3 operate on the same environment
env05
:
Size: 1 500
×
1 500 cells
20 000 obstacles (randomly placed walls)
40 000 randomly placed tiles
1 500 randomly placed holes
This benchmark suite uses a single-threaded event-driven simulation
engine and a multi-threaded event-driven simulation engine. Each
simulation has a duration of 100 time steps (simulation time). The
test cases are documented in Table 9.6. The simulation time and the
environment are fixed, the parameter being varied is the number of
agents as well as the simulation engine.
