On my machine
$ nim c -d:danger -d:lto --listcmd reverse.nim
......................................................................
$ time ./reverse
real 0m31.493s
user 0m31.491s
sys 0m0.001s
$ nim c -d:danger -d:lto --listcmd -d:faster reverse.nim
......................................................................
$ time ./reverse
real 0m3.487s
user 0m3.485s
sys 0m0.002s
Run
`-d:lto` flag makes code much faster.
If you really want to know why, you need to read Nim generated C code and
assembly code generated by backend C compiler. Nim generates assembly code with
`--asm` option. But if you use `-d:lto` flag, see:
<https://internet-of-tomohiro.netlify.app/nim/faq.en.html#nim-compiler-how-to-produce-assembler-codeqmark>
You can get simpler assembly code with `-d:danger` as it remove runtime checks.
<https://godbolt.org> shows assembly code but it is less readable because it
doesn't tell function name `call` instruction calls and hard to see where jmp
instructions jump to.
