Seems to me that it pretty much has to be a torque delta (not %) for any given rate of acceleration and combined moment of inertia of engine rotating mass and dyno rotating mass. Fairly easy to calculate IF you knew the rotating mass #s.... If dyno mass is significant compared to engine (??) then a little engine on a big dyno might have a significant % difference to a big engine on same dyno.
As a semi relevant aside, I did my senior engineering project testing my old GP Datsun 1600 engine on an ancient and huge DC motor type dyno. Had to use a transmission in 2nd (?) gear to avoid overspeeding the dyno, and even then the inertia of the dyno was tremendous and made testing a huge pain. Speed was varied simply by controlling the load resistance, and it took FOREVER for the RPM to stabilize at each test point. And what was interesting was watching the torque arm load drop off as speed was increased (and rotor accelerated) and then SLOWLY rise again as speed stabilized. Interesting and utterly terrifying standing 3 feet (in front of crank pulley where I was pretty sure no shrapnel would go) in front of a screaming/thrashing race engine (that I thought might blow up) as tens of seconds ticked by waiting for the torque to settle down (settle UP really)....
