Sanity check regarding normalization with cube maps

Given that
fragment.texcoord[3] contains a vector multiplied by the inverse transpose of the modelview matrix.
( and is not already normalized )

texture[1] contains a normalizing cube map.

These two ARBfp snippets should be pretty much the same, right?

# normalize with math
DP3		normal.w, fragment.texcoord[3], fragment.texcoord[3];
RSQ		normal.w, normal.w;
MUL, fragment.texcoord[3], normal.w;
# normalize via cube maps
#TEX, fragment.texcoord[3], texture[1], 3D;
#MAD, normal, two, neg_one;  

TEX, fragment.texcoord[3], texture[1], CUBE;

– Tom

Thanks Tom. Hangs head in shame