The time that Microsoft Unity is using to resolve a big amount of transient components compared with other IoC's has a very big gap.
I can say this because I have done some tests about different IoC's (Microsoft Unity included).
Furthemore, If you compare the results of Microsoft Unity v2.1 and the new Microsoft Unity v3.0, both are very similar.
I think that you haven't improved some performance details of Microsoft Unity. It's a good opportunity to improve some things about Microsoft Unity.
Have you really thought about this to improve Microsoft Unity or isn't important for you?
Thanks in advanced.