I found
New Fixed Ramp Rate this about Dyno Dynamic ramp rates.
It should be remembered that that any dyno which plots torque/power against RPM has to use some ramp rate (i.e. rate of change of engine speed) in order to produce a plot, and therefore any such plot is always going to be somewhat artificial.
But having use a Dyno Dynamics rolling road for mapping myself, when the rollers are set to hold a given RPM so that fuel/ignition can be set against engine load/throttle position/boost I can tell you that the levels of torque and power achieved by an engine under that situation are always higher than in 'shoot' mode . This is simply because the engine is loaded to the point where it can't accelerate, which is where maximum output is achieved for any given engine speed.
A plot of torque/bhp against engine speed made up of all points measured in the 'rpm held' state used for mapping gives the highest output figures. It would be fair to say that the engine output shown would be accurate whilst at the same time not being 'real world' - i.e. in the real world you would expect the car to be accelerating at some rate at wide open throttle from x rpm through to y rpm. The different 'shoot modes' offered by Dyno Dynamics are analogous to measuring the engine power on the road, but by driving up hills of different gradients.
Going back to the link posted by Flea, the RS when running in 6 cylinder mode was shown to be producing more power than in 4 cylinder mode, which indicates that the loading on the rollers is higher in 6 cylinder mode giving a slower ramp rate. There is nothing wrong about the 347bhp figure - it is not the dyno adding some artificial power to what the engine is putting out. However, it may be that the engine would not deliver that in the real world when accelerating on the flat with little aero drag (i.e. in 1st, 2nd and maybe 3rd gear) but would going up a steep hill (in any gear) and at higher road speeds where aero drag is significant.