blocks:
This feature measures the number of unique 64 byte cache blocks that are touched by a region of code. It is measured as the ratio of the number of unique cache blocks accessed to the total number of memory accesses in a code region.

Classifying blocks:
blocks can be classified into low, medium or high as follows:

Bucket	Condition
Low	The number of unique array accesses in the region of code outnumber the number of arrays declared by about 3:1. OR An array access takes place in a nested loop.
Medium	The number of unique array accesses in the region of code outnumber the number of arrays declared by about 3:2.
High	Each array in the region of code has only one unique access.

Example 1:
Consider the following code:

for (i=0; i<1000; i++)
   result[i] = sqrt(arrayOne[i]) * sqrt(arrayTwo[i]);

Example 2:
Let's look at another example:

for (i=0; i<1000; i++)
   result[i] += sqrtf(arrayOne[i]) + sqrtf(arrayTwo[i]);

Example 3:
Consider the next example:

for( j = EXCESS_PADDING_PER_DIM_BY_2(SIZE_FILTER); j < ny + EXCESS_PADDING_PER_DIM_BY_2(SIZE_FILTER); j++ )
    {
        for( i = EXCESS_PADDING_PER_DIM_BY_2(SIZE_FILTER); i < nx + EXCESS_PADDING_PER_DIM_BY_2(SIZE_FILTER); i++ )
        {
            s = i + (PADDED_IMAGE_ROWS_COLS(nx)*j);
            sum = 0.0;
            for( k = 0; k < SIZE_FILTER; k++){
                  index = INDEX2D(i,j+k,PADDED_IMAGE_ROWS_COLS(nx),SIZE_FILTER);
    for( l = 0; l < SIZE_FILTER; l++ )
                      sum += *(filter + (k*SIZE_FILTER) + l) * *(inputImage + index + l);
            }
            *(outputImage + s) = sum;
        }
}