If this is really important to your application - ie you really need the utmost performance - you will have to do it in assembler.

There is no guarantee that any particular 'C' source line will (or will not) use any particular machine instruction(s). You might find a way that happens to give what you want with your current version, but there's no guarantee that it might not change with a different set of options, or for a different version, or...

That's the whole point of a high-level language like 'C' - you forgo the fine-detail control of Assembler in favour of the programmer's convenience.

Remember, the '>>' operator in 'C' has to work with any combination of signed/unsigned, char/int/long/float, etc, etc operand and result types - therefore the Compiler has to generate code that'll work in any general case.
Again; programmer convenience at the cost of Assembler control.