Why do people still assume C is always fastest? You can write C++ and D like it's C and it'll produce identical machine code. Which means that after profiling an idiomatic C++/D program(and assuming for the sake of argument that C is faster, which isn't even close to necessarily true) you could rewrite the hotspots in C-style and achieve the same performance.

I don't understand why this myth persists, despite the fact that Fortran has been beating C at several benchmarks for as long as C has existed.