Majdoleen,
I dont know if any specific performance software, but you could look at the timing that shows in the browsers developer tools (usually accessed using F12 with misread browsers). There is usually a time stamp. For basemap/rasters I think you will notice a difference the 2nd time a user visits the same area but it will be slower the first time.
For vector data, with the newer versions (10.2/10.3) I typically do not cache those anymore, unless it uses a complex line symbol that doesn't need to change in the service. In that case, it may save a second for rendering, but in general, those versions deliver the vector/feature data just as fast as cached data.
Since I think your original question in the thread was answered previously, you should probably mark all those that were helpful, and one of the responses as the answer. Since this question about testing might produce a variety of answers, you might was to fork it to a new question. That will help others find the Q&A if they have the same question in the future.
