Under certain circumstances Virtuozzo kernels can perform better than stock Redhat kernels.
Is having a VPS going to give you the same performance as a dedicated? There are a lot of factors that go into this, the main one being how loaded your server is. If you have a good responsible service provider the support/backups/and control panels included with your VPS likely make up for any lost performance from a true dedicated.
I wouldn't call it near zero. 3-5% overhead would probably be close to the actual loss vs. a dedicated. Whether you get anything close to the maximum available is up to how many other virtual servers are running, their contention for the same burst resources, and how well configured the Virtuozzo server is.
VPS's are primarily about giving you a guarantee of minimum performance, and full isolated control of your own server that's indistinguishable from a real dedicated. Plus the added benefits Spry pointed out.
If you do a search, there was a massive thread comparing VPS's and dedicated server performance.
As mentioned it depends on alot of things. But the performance is not going to be the same no matter how low the overhead is. Considering you have multiple users fighting for the same resources such as disk I/O.
If your the only VPS on the server then its possible to have the same performance as a dedicated server.
I think the massive thread sshepherd is refering to is the VPS Benchmarks - Post Your vps performance thread. It reveals performance specs about stuff running under any one VPS, but little about a running server with many customers' VPS accounts switching in and out. That's where the overhead part comes in, and yes, with millions of unpredictable variables.
There is occasional mention to specs like time to load first page, page lag, and others, but real multi-user overhead data isn't readily available.