Could you better explain what you are trying to do.  Would the number
00001111 come before 00011110, or when there is the same number of ones
is there a secondary sort to bring the binary implementation into play?

i=0;
j=0;
mask=1;

for(i=0;i<32;i++)
{
        if(mask & value) j++;
        mask <<= 1;
}

Using the j parameter to determine the exchange in a quicksort.
I'm sure there are faster implementations shifting into the carry.

As far as what's the quickest method, radix will behave similarily
to quicksort with large random files.  Merge behaves similarily to
radix for random data.  I remember an old program that came with
Turbo C that evaluated which method would be quicker based on the
input data.  Perhaps you could test your data with it.

Craig


> -----Original Message-----
> From: pic microcontroller discussion list
> [mailto:PICLIST@MITVMA.MIT.EDU]On Behalf Of Samuel Winchenbach
> Sent: Wednesday, August 23, 2000 3:46 PM
> To: PICLIST@MITVMA.MIT.EDU
> Subject: [OT] FAST C/C++ sorting
>
>
> Hello all.
> I have about 8 million 32 bit integers that I need to sort (by the number
> of ones in the binary representation of the integer) I am wondering what
> the fastest sort is? Merge? Quicksort? Fast Quicksort? Radix?
> I am storing my values in my data structure contained within a vector. So
> I know I will need to manipulate the sort function around a bit. I have
> already overridden all the needed operators. If anyone can give me
> suggestions I would appreciate it.
> Writing a Quine-McCluskey program is more difficult that I would have
> ever thought.
>
> --
> http://www.piclist.com hint: The list server can filter out subtopics
> (like ads or off topics) for you. See http://www.piclist.com/#topics
>
>
>

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics