Execution time variations in simulating same model repeatedely

When I simulate a model over and over again, why do I get different execution time. For example, in the following code I am grabbing the simulation time and for my model I see a wide variation of simulation executed time [ varies from 0.70 to 1.01 min]. I am trying to understand where the randomness / variations is coming from?

for i = 1:100
ts = tic;
sim(mymodel)
tf = toc(ts);
end

[Doug Hull]

What is in the model? Any random variables, network dependencies? Why does this matter?

[Abhijit]

Its a deterministic model and without any network dependencies.

This matters when you are doing millions of simulation.