This is completely incorrect, eAccelerator does nothing of the kind. It is a bytecode cache, and as such works in the same way as the bytecode cache we usually recommend: APC. It doesn't cache pages, it caches compiled source code. This is undeniably a good thing.
Ok, I need to read things more fully in future, sorry. My point still stands though - it shouldn't be taking 10 seconds to parse a small page, surely?