Ideally every part of the game should be loosely coupled and easily benchmarked on its own.
In these cases where a small component needs to be tested separately, we should simply use jsperf. For instance if we want to check if a linked list is faster than an array for a certain type of job.
But sometimes this is not possible, therefore it could be useful to have a system that allows us to setup one or more different settings, for instance is it best to use array or linked list for the particle system, and then run benchmarking on each of these configurations. The output could be highest, lowest and average fps and the average execution time of the part of the loop in question.
Ideally every part of the game should be loosely coupled and easily benchmarked on its own. In these cases where a small component needs to be tested separately, we should simply use jsperf. For instance if we want to check if a linked list is faster than an array for a certain type of job.
But sometimes this is not possible, therefore it could be useful to have a system that allows us to setup one or more different settings, for instance is it best to use array or linked list for the particle system, and then run benchmarking on each of these configurations. The output could be highest, lowest and average fps and the average execution time of the part of the loop in question.