Re: highest bit, statically determined

[Brian Drummond <brian@shapes.demon.co.uk>]

On Mon, 01 Oct 2012 23:30:24 +0200, Georg Bauhaus wrote:

> On 01.10.12 10:52, Anatoly Chernyshev wrote:
>> (0|1=>0,2|3=>1,4..7=>2, 8..15=>3, 16..31=>4,32..63=>5, 64..127=>6,
>> 128..255=>7);
> 
>     68580000 2.653934000 68580000 2.021029000 68580000 27.702262000
>     68580000 1.173348000   -- first_1_bit_table
> 
> I guess the approaches can be used together for larger N-tets.

Just for completeness,

function first_bit_case(N:Octet) return natural is 
begin 
   case N is
      when 0       => return 0;
      when 1       => return 1;
      when 2..3    => return 2;
      when 4..7    => return 3;
      when 8..15   => return 4;
      when 16..31  => return 5;
      when 32..63  => return 6;
      when 64..127 => return 7;
      when others  => return 8;
   end case;
end first_bit_case;

Here (on an Atom netbook) it measures about the same as the equivalent if 
chain; faster than the recursive versions but slower than the table.

         --Z := Z + first_1_bit_ifchain(Data(i)); -- real 0m3.208s
         --Z := Z + first_bit_table(Data(i));	-- real	0m1.971s
         --Z := Z + first_1_bit(Data(i));	-- real	0m4.672s
         --Z := Z + First_1_Bit_A(Data(i));	-- real	0m3.937s
         Z := Z + first_bit_case(Data(i));	-- real	0m3.272s

The Ada-2012 case expression was no faster.

- Brian