Re: Select MS Bit from an integer

[Andrew Johnson <anj@aps.anl.gov>]

On 09/04/2015 11:12 AM, Till Straumann wrote:
> You could do it with a bunch of calc (C1..C5, V) records
> 
>  C1 computes  (IN & 0xffff0000)
>  C2 computes  (c1 ? c1 : IN ) & 0xff00ff00
>  C3 computes  (c2 ? c2 : c1 ? c1 : IN) & 0xf0f0f0f0
>  C4 computes  (c3 ? c3 : c2 ? c2 : c1 ? c1 : IN) & 0xcccccccc
>  C5 computes  (c4 ? c4 : c3 ? c3 : c2 ? c2 : c1 ? c1 : IN) & 0xaaaaaaaa
> 
> V computes  (C1 ? 16 : 0) + (C2 ? 8 : 0) + (C3 ? 4 : 0) + (C4 ? 2 : 0) +
> (C5 ? 1 : 0)

If I've understood this code properly V will be the bit number of the
highest set bit, from 0 through 31. It doesn't distinguish between no
bits set and only the LSB set though, you'd have to detect that separately.

If you have the synapps calc module installed a single transform record
could do all these calculations, transform handles up to 16 expressions.

You can find lots of bit-manipulation hacks at
    https://graphics.stanford.edu/~seander/bithacks.html

- Andrew

> On 09/04/2015 07:59 AM, Dudley, David wrote:
>>
>> Quick question that I didn’t see an immediate way to perform
>>
>> How do I select the most-significant set bit in an integer?  I was
>> thinking about using a SEL, but that seems like a lot of work just to
>> do a simple thing.  Other possibility would be a SUB, but I’d like to
>> do it with base record types if I can.
>>
>> Something simple please…

-- 
Light thinks it travels faster than anything but it is wrong.
No matter how fast light travels, it finds the darkness has
always got there first, and is waiting for it.
    -- Terry Pratchett, Reaper Man