More bit shifting issues

Boltar <boltar2003@yahoo.co.uk> writes:
> On Apr 1, 11:58 pm, "robertwess...@yahoo.com"
> <robertwess...@yahoo.com> wrote: 
>
> Well that begs the question of why the cpu integer div instruction
> doesn't simply check the divisors least significant bit to see if its
> set to zero (even) and if it is then use bit shifting to get the
> result. Same for multiply.

(Others have already mentioned that you need a power of 2, not just a
multiple of 2, but that doesn't invalidate the idea.)

A plausible explanation is that chip designers assume that most
multiplications and divisions by constant powers of 2 will be
optimized to shifts by the compiler.  Thus such a hardware check will
benefit only cases where a run-time value just happens to be a power
of 2, and will likely slow down *all* multiplications and divisions.

On the other hand, I'm no hardware expert.  Perhaps such a test can be
done in parallel without slowing anything down.  If so, perhaps it's
already being done (it would be invisible to user code, or nearly so).

--
Keith Thompson (The_Other_Keith) <kst-u@mib.org>
Nokia
"We must do something.  This is something.  Therefore, we must do this."
-- Antony Jay and Jonathan Lynn, "Yes Minister"