I've just been looking over the Walkers 346XP dyno run and there looks to be correlatable truth to it. I have no reason to doubt Walkers evidence and as I said before, these dynos are really accurate. The problem with reading the graph is that RPM is plotted as a dependent variable (along with HP and torque), versus the independant variable, dyno pull time on the X axis. This makes quick comparisons of data cumbersome. HP and torque should be plotted versus RPM. Also notice the "corrected" flag in the upper left corner of the graph. This indicates that temperature, barometric pressure and humidity are used to normalize the torque numbers by the dyno software if these values were put in. What this means is that you could make a run at another time with different atmospheric conditions and the corrected values would be directly comparable.
Nonetheless, the stock 346XP makes it's peak HP of about 2.5 at 10,000 RPM. It's peak torque of about 1.7 ft. lb. occurs at about 7,000 RPM. What I don't understand is why the HP and torque curves don't intersect at 5252 RPM. They should always be equal at this RPM, given the HP=T X RPM/5252. This could be because the throttle was closed near the end of the run, so it's no biggie. The dashed, Walkerized curves show a peak HP of about 3.8 at about 11,000 RPM which seems to make sense, in that the peak power has moved up by 1000 RPM. The peak torque has also moved up to about 2.3 ft. lb. at an also higher RPM of about 8,000. What Walkers has done to make a 50% increase in HP is to build in a lot more torque in the higher RPM range. I find it amazing that they can get this kind of power increase with such a small increase in RPM over stock, but what do I know? Note that the modified torque curve is also much more peaky than the stock one, although this is somewhat deceiving due to the use of two separate RPM curves. A quick check of a couple of points says that the dyno is correct mathematically in deriving HP from torque and RPM.
In regard toward the X Axis, which is in seconds: The sequence of events that happened here is that the dyno is initially set to a minimum load condition at time=0 sec. The saw throttle is then opened fully, then the dyno load (restriction valve for the water) is increased (valve closed) gradually and the computer takes RPM and torque readings over a period of 12 or 18 seconds here. Note that the load was backed off momentarily in the stock test at about t=14 seconds causing the HP and RPM to rise and the torque to fall as they should under this condition.
If I get a chance tomorrow, I'll plot the stock and modified HP and torque curves versus a single RPM scale as they should be and post it up here. The results are at least in the ballpark and seem to indicate that, at least in this case, Husky's 3.4 HP factory claim is very inflated.
Sorry for the verbosity, but I'm sober tonight.