Fast memmove for x86 and +1 shift (for move-to-front conversion)

Question

For fast MTF ( http://en.wikipedia.org/wiki/Move-to-front_transform ) I need a faster version of moving char from inside the array to the front

char mtfSymbol[256], front;
char i;

for(;;) { \\ a very big loop 
    ... 
    i=get_i(); \\ i is in 0..256 but more likely to be smaller.

    front=mtfSymbol[i];
    memmove(mtfSymbol+1, mtfSymbol, i);
    mtfSymbol[0]=front;
}

cachegrind shows that there are many dummy predictions for memmove.

For a different version of the code (not memmove in the first example, except this one)

do
{
   mtfSymbol[i] = mtfSymbol[i-1];
} while ( --i );

there are many read / write bytes, conditional branches and incorrect branch predictions

i is not very large, since MTF is used for "good" input - a text file after BWT (Burrows-Wheeler conversion)

The compiler is GCC.

Answer 1

, , - ( , ), ( ), , .

, , , , , , , .

Answer 2

, , . , .

. http://code.google.com/p/kanzi/source/browse/java/src/kanzi/transform/MTFT.java

, , .