Here is a list of tested batteries.
It's written in spanish, but it's easy to understand:
http://wiki.gp32spain.com/index.php/Pilas
I don't speak any spanish so I have no clue what the text above the table is about. 

But a quick look at the entries let open a few questions. For example
 - Did all contributers did the same things with the GP2x?
 - Did they measure the time correct?
Just look at the entries for "Energizer 2500": 3:10, 3:49, 2:05 ???
amd Ansman are well known for their accus. And then such a low time?
For me this list of no use ATM. Maybe someone can enlighten me?
salut,
 creature