Here's a version that at least for 64 bit might be faster: bool isPerfect(uint64_t x) { int ctz = __builtin_ctzll(x); return ((0x40051056 >>

Here's a version that at least for 64 bit might be faster: bool isPerfect(uint64_t x) { int ctz = __builtin_ctzll(x); return ((0x40051056 >> ctz) & 1) && ((x >> ctz) + 1) == (uint64_t(1) << (ctz + 1)); }