Hmk.
On a completely unrelated note, do you have thoughts on this?:
Assuming eax is either 0, 1, or 2, would you think that [[buffa+eax*4]] or [buffb+eax*4b000] would be a faster operation? (they are equivalent, ie, [buffa] = buffb, [buffa+4] = buffb+4b000, [buffa+8] = buffb+96000)