Contrary to popular belief, neither using APC or Memcache as cache/session backends will make a noticeable difference to single-user page load time - unless you had a disk I/O bottleneck to begin with (meaning that the memory backed cache will outperform the slow disk based cache).
If you don't have I/O issues, then enabling them won't make a difference to the actual page load time.
Where they'll come into play is in high concurrency situations, synthetic benchmarks like Siege/AB will demonstrate this very clearly, with an almost 50% increase in request/s. How this translates to real-world performance however, isn't so dramatic.
You'll know if it is working if the
var/cache directory is empty, and if using Memcache for sessions, if the
var/session directory is empty.
If you use file-backed
slow_backend, then you'll still see cache tag data created in
var/cache - but not the cache data itself.
Bottom line, if you are trying to improve page-load time performance, changing the cache store isn't going to make a noticeable difference unless your server's HDDs are so contrained that serving files from disk is a bottleneck.