rjbh20,
Parasitic drag increases with rpm, period. Takes more to turn a higher rpm than at a lower one. But this is just the rpm-related component and varies with the square of the rpm (roughly) and is independent of HP.
Adding to that, side loads and thrust loads in bearings and bushings go up as the torque applied increases (torque tries to split the transmission case or rear pumpkin for example). Those increased side and thrust loads increase the friction in the bearings and are parasitic losses related to torque (HP). HP is just torque at a specific RPM, so now we have an increase in resistance to turning that is driven by HP (torque at rpm). Additionally, the faces of the gear teeth slide against each other and there is another friction component. As anyone knows, as the load normal to the friction surface increases, the friction force increases (slide an empty box across your floor and then put bricks in it and slide it again... Harder to slide with brick in it!). Since the torque acts (HP) acts directly on each toth of a gear set, as torque (HP) increases, so does the frictional loss from the sliding of the gear teeth against each other.
There are a few other parasitic sources, but these are the major ones and other than the sheer speed of turning the drivetrain addressed in the first paragraph (slinging the
oil, etc...), we have two major sources of loss that increase as HP increases. Therefore loss is proportional to HP (torque). Usually drivetrain losses are in the 15 to 23% range. IRS is usually higher - think of the u-joints as bearings with side-load (as the side load goes up the frictional loss goes up).
Hope this is enlightening.