Since this discussion is ranging all over the place...
I've never understood how parasitic loss through the drivetrain could be a percentage. That would mean a really powerful engine would lose more power than a less powerful engine turning the same components at the same RPM.
A transmission should take a given amount of power to spin it at a given RPM. The internal friction would not change based on the engine in front of it. Same with differential and all the other sources of losses such as the axle bearings, u-joint bearings, tire rolling resistance, etc.
So let's say we have an engine that puts down 150 HP at 5000 RPM on an engine dyno. Put it in the vehicle and you're seeing 100 HP at the wheels - you have 50 HP of loss in your drivetrain at 5000 RPM. If you put an engine that is making 400 HP at the crank at 5000 RPM in the same car you should see 350 at the wheels.
Can someone explain how that line of reasoning is wrong?
There's some good technical discussions on chassis dyno vs engine dyno calculations on the internet and published in books.
But.. to simplify it.... there is a "general average" percentage of loss thrown around that is probably fairly accurate in most chassis dyno tests being done on known vehicle engine specs and the vehicle chassis they are mounted in.
Keep in mind that most dyno testing, chassis or engine, is for tuning and seeing power gains based on typical modifications or component changes. So, a base or bench mark test is typically done, then the mods done, then another set of pulls to calculate changes.
That is probably not the comment you are looking for, but if you really want to know more, do some research and I think you will find some really interesting insight into dynos, variables, and results.
Gnarls.