Did anybody try to make a performance comparison between .net 1.1 and
2.0 in real applications ? I mean not just small tests like Generics
vs. Collections or DataSets, but the performance of the whole
application using those things ?

We experienced with the performance drop after moving to the .net 2.0.
Its a GUI application with large layer of Middleware that is using
C++.NET and C++ libraries. In general application waits for some evens
on the socket and reacts on them. We upgraded critical parts of
application to use Generics collections (like task queues, etc) and now
tests are showing increase of the reaction time with .net 2.0.

After some time of running the application I can see that .net 2.0
version consumes 10-20% more memory. From Performance counters I can
see that .net 2.0 GC collector is running 2-3 times less often than in
..net 1.0. In theory with this kind of GC behavior the application
should experience less amount of delays, but in practice it's not.

Could anybody suggest something ? Maybe I miss something ?
I would be appreciated for any help.

Please, any thoughts or assumptions on this one ?

ISerialization Frustration !
About IDisposable.Dispose()
Proper unmanaged to managed exception handling -- SEHException and
Delegate problem
Implement a Time-out in the waitQueue of the ThreadPool