This is greatly reducing the fixed Array.Copy overhead for common cases. E.g. Copying byte array of length one:

• Array.Copy(Array, Array, int): 2.3x faster
• Array.Copy(Array, int, Array, int, int): 1.5x faster

Also, this is fixing the performance penalty of the Array.Copy(Array, Array, int) overload for common cases. This overload is faster than the more verbose Array.Copy(Array, int, Array, int, int) one for the typical cases. We can switch the places that used to explicitly pass the start index as 0 to avoid this penalty back to the less verbose overload.

cc @stephentoub

We can switch the places that used to explicitly pass the start index as 0 to avoid this penalty back to the less verbose overload.

Nice. I'm always a fan of both simpler and faster :) I can do so separately in corefx. Do you want to do it in corelib as part of this PR?

Do you want to do it in corelib as part of this PR?

My plan was to do that as follow up PR.