With regards to the OP's issue, some insight in simple terms here - Out of the box you've essentially got your stock system which has its own files and then you install applications. Each application installation will have its own minimum files required to run. Once those applications are run they'll create temporary storage files (cache) as well as data files (with your pertinent information such as account info and custom settings within the app). The cache files do not store important data, they're merely there to speed things up. Say for example the application aggregates data from various sources on the internet, each time you open the app it will verify the data it already knows and skip downloading it again if it already has it in cache.
Clearing cache frequently is wasteful since it only forces the applications to get the data that was in there again, resulting in slower performance. Sometimes the need arises though when an application gets updated and the cache files that were done by a previous version of the application are no longer compatible. The newly updated version of the app may have trouble with the cache files that were done by the prior version. This is why it is common that people will ask if a user has restarted and cleared cache if they're having software issues.
As far as clearing cache and the impact to the benchmarks that jcords is describing I would postulate that possibly caching is going on during the second benchmark test and impacting performance during that time. It may not be the benchmark test itself that relies on its own cache files that were cleared, it may be other background processes which now are rebuilding cache in the background during testing. Hard to say .. . .
Facebook
Twitter
Instagram