I bring you all the truth of what really makes new languages slow.
Just to clarify even further for you: it's not the language that's fast or slow, it's the implementation.
You want examples? OK; it's perfectly possible to write a C interpreter. And it's also perfectly possible to compile C# code down to native machine code. Which do you think is going to be faster? Interpreted C or C# compiled to native machine code?
The thing is that it isn't really interpreted. It's transformed into IL (like the other stuff from .net, thus the crazy easy redeable decompilations), and then it needs to compile into machine code before runtime (jit).
I know that the forced inlining works, that's the thing. The jit heuristic is just bad, really bad. I'm sure you could find a failed case, even if these compilers were a lot better, but i'm not telling C++ programmers what to do, and I don't really know. I mean, I think you are even ok without unrolling loops, but I would only test it if I programmed with that.
But some other HL languages do that interpreting thing. So, yeah, they are slow.
So I'm new and I'm out of date. Still, I can find some cases lol. But I could only imagine that these might need to be custom fited to a program, there's no way a perfect heursitic could know how often will something be called.