Information Technology Reference
In-Depth Information
«effector»
Move 1 step
«event»
Perceive
«event»
Move
«ev
ent»
Plan
«event»
Pick-Up
«event»
Start
«sensor»
Perceive
«effector»
Plan
«effector»
Pick-Up
«event»
Put-Down
«effector»
Put-Down
Knowledge
Figure 9.13:
Sensor-effector-chains of the Tileworld-agent.
9.5.2 Benchmark suite 0: Dependency between simulation
time and runtime
Benchmark suite 0 is intended to measure the dependency between
simulation time and runtime. As argued in chapter 8.1.2, a simulation
as defined by the GRAMS reference model only takes into account the
current model state when computing a state change. Given a model
of fixed size, runtime should scale linearly with elapsed simulation
time.
Setup suite 0
All benchmarks in this suite operate on the same environment
env03
:
Size: 1 000
×
1 000 cells
2 000 obstacles (randomly placed walls)
2 000 randomly placed tiles
100 randomly placed holes
This benchmark suite uses a single-threaded event-driven simulation
engine. The test cases are documented in Table 9.1. The size of
the model is fixed (environment, number of agents) and only the
simulation time is varied.
