Optimization help: use another algorithm?

Question

I have a program that generates and uses three dimenisional array (dimensions 256*256*256). However it takes more than 80 seconds to run just this block of code
Basically all it does is pick a pesudo-random sample using only one inital random value. kind of like in winamp when selecting 'random' play mode.
i would like to know if there is a much quicker algorithm than what i currently have.... and also if at all possible given a sample (say of size 255) get the random number that generates this sample?
here is the code:
int[] og = new int[256*256*256];
temp =randomFloatNumber*16777216+16777216;
rnd =(int) temp;
System.out.println(rnd);
for (int i=0;i<(256*256*256);i++)
og= i;
for (int i=0; i<(16777216);i++)
pos=(pos+rnd)%(16777216-i);
b1=og[pos]/65536;
b2=(og[pos]-b1*65536)/256;
b3=og[pos]-b1*65536-b2*256;
groupCount[b1][b2][b3]=i/255;
groupCount1[b1][b2][b3]=i-groupCount[b1][b2][b3]*255;
og[pos]=og[16777215-i];
if (i%167772==0)
System.out.println(i+"/16777216");

Answer 1

Your specification could be more precise. For example, what defines a "group of three bytes"? Is (data[0], data[1], data[2]) the first, (data[1], data[2], data[3]) the second, and so on? Is order within a group important? What to do if the top 254 are not precisely determined? And by "the seed value that will generate these groups", do you mean an integer j such that Random rand = new Random(j);
int[] groups = new int[254];
for (k = 0; k < groups.length; k++)
    groups[k] = rand.nextInt(1 << 24);
}will fill groups with the corresponding three-byte groups in order of occurrence in the original data set?