Ok. This works (replicates Nim performance) for me (on gcc-7.1.0 x86_64 Linux -
not sure what others):
float fibonacci(int x) {
return x < 2 ? (float)x : fibonacci(x - 1) + fibonacci(x - 2);
}
#include <stdio.h>
void NimMainInner() {
printf("%.0f\n", fibonacci(50));
}
int main() {
void (*volatile inner)(void) = NimMainInner;
inner();
return 0;
}
Evidently, gcc can only figure out how to unroll the top levels of the
recursion {to call just fibonacci(38..44)} only if the call site has been
isolated by being called through a volatile function pointer which Nim happens
to do.
