Previous | Next --- Slide 33 of 44
Back to Lecture Thumbnails
jiajunbl

We mentioned in class that one approach would be to take samples of the running of the program at different points in time.

I was wondering that if the algorithm builds up a sort of data structure or cache over time, would doing something like this still be feasible (i.e. if the time slices are large enough would it not matter anymore?)

regi

@jiajunbl: If this is the case, one could run the problem for a larger period of time in each section, and ignore the performance in the first k timesteps (where k is some value that ensures the data structure or cache is filled to approximately how it would be in the actual problem).