* DV decoder simplifications. Now it looks to be 6% faster. At least