I was wondering if there are any differences - mostly performance wise - between the two Win32 API functions EnumProcesses() and CreateToolhelp32Snapshot()
I think they are pretty much the same in terms of performance (and results) as they both call the same underlying NT API, though CreateToolhelp32Snapshot() may have a slight overhead as it creates a section object and copies all the information to it whereas EnumProcesses()/EnumProcessModules() works directly with user-supplied buffers. The difference is probably negligible in real world performance, though.
I slightly prefer EnumProcesses() as it is (IMO) a simpler API to use, but CreateToolhelp32Snapshot() returns more information if you need it. The only downside to EnumProcesses() is that you are supposed to call it in a loop as you may not have allocated a large enough buffer; CreateToolhelp32Snapshot() takes care of the buffer management for you. In practice I just allocate a buffer on the stack large enough to hold 1024 process ids or module handles; so far I have not come across a system where either of these limits was even remotely close to being reached. Of course we said the same thing about MAX_PATH not so long ago and now we are running into problems with that...