Question

Is there a way to extract a CPU word size long subsequence of bits from a bitset efficiently without iterating over each bit individually? Something like

#include <bitset>
#include <iostream>

using namespace std;

int main() {
        bitset<100> b;
        // Do something with b
        // ...

        // Now i want sizeof(long) many bits starting at position 50
        unsigned long l = (b>>50).to_ulong();
}

would do if it would truncate the bitstring instead of throwing an exception!

Answer 1

You could create a constant bitset mask that only has the bottom N bits set, eg like this:

bitset<100> const mask((unsigned long) -1);

Then you can do ((b >> 50) & mask).to_ulong() to extract the bits. If your definition of “word” isn’t the same as unsigned long, a different mask will be needed.

(I changed your left shift to a right shift, which I believe will work better.)

A sufficiently smart compiler could convert this to just a shift and reading out the result; I doubt whether any compilers are actually sufficiently smart. But I suspect the cost of the shift outweighs the cost of the and anyway.