that optimization was not for buffers.  buffers are already fast.