> The cost of moving code from unmanaged to managed can be
> significant and I do not mean to minimize that fact, but did
> want to address your statement
> "...isn't possible to go completely to managed code. As
> speed is an issue..."
> There is *not* a blank rule of thumb that says you are losing
> performance for services by writing managed code.
You are right. I should not have implied otherwise.
> It really
> depends on the nature of your application on whether C++ is
> faster. More to the point, is the marshaling performance
> going to outweigh any performance benefits of keeping code in
> C++. In fact, in certain cases managed code can simply out-perform
> C++ code (memory allocation is a good example of this).
> I do not suggest that you take the advice of any (or all) of
> us on this list as the final arbiter because we cannot gleam
> the specifics of a complex system by trading e-mails in this
You are right again. But I was hoping someone could give me advice on how I could
investigate the issue myself. As I said in the beginning, I have no idea whether
marshalling costs me 10s or 2min. Here's my original question:
> How can I understand the costs involved in this transfer,
> and how can I minimize these costs?
My question, especially the first part, remains unanswered.