Re: Q: division vs multiplication

Mr Tomas Hulek <t.hulek@imperial.ac.uk> wrote:
)My question is: does this also apply to division by constants that are powers
)of 2? For real-world compilers (C,C++ let's say) and machines, is there any
)benefit in using
)
)r = 0.5*(a+b);
)
)rather than
)
)r = (a+b)/2.0;
)
)???
)
)I would imagine that division by 2.0 could be done very efficiently, just like
)division by 10 in our decimal system. But is it really so?


Depends on the floating point base. If the base is 2, then all one has
to do is extract the exponent, subtract one, put it back, and then
normalize the result. This is indeed much faster than dividing by 2. But
I know of no compilers which make use of this fact.


Mike
[As noted elsewhere, there are FPUs that have a floating scale operation that
does this in one instruction, but it's not always faster than a multiplication
or division. -John]
--

Return to the comp.compilers page.
Search the comp.compilers archives again.