Since we are talking about a hefty $600+ US for one of these little numbers... I really think it's important you guys are not being mislead here.
Referencing the Giles 1.6lNA pump test:
http://www.vwdiesel.net/phpBB/viewtopic.php?t=68&start=15...I do believe that concerns about this test's accuracy are well-founded for the following reasons:
It was done using an older G-tech, which does not plot hp and torque versus RPM, it only gives "peak" readings. It is possible to wind the motor up and store energy in the flywheel and let the clutch out abruptly, to artifically inflate the "peak" (often not intentionally.) In the following example, look around 3 seconds to see two examples of the horsepower the G-tech sees artificially and unintentionally spiking above 50hp:
The solution is to be aware of this potential pitfall, and let the clutch out gingerly before applying full throttle. And also to use a newer, RPM sensing G-tech, and verify your HP peak occurs about where you'd expect to see it in the RPM range. A peak HP shown at an unexpectedly low RPM is a red flag it may be artificially inflated by flywheel energy.
Also, I have encountered even on the much newer and more accurate RPM sensing G-tech models, that they can sometimes report extremely optimistic results because sometimes the G-tech gets disoriented about which way is "down" (and therefore, doesn't have an accurate grasp of which direction is "forward".) The resulting skewed acceleration readings the G-tech gets can distort HP and acceleration times usually very badly, on the order of many tens of percents! The workaround for doing reliable and accurate testing is to repeat the test 2-3 times and verify that all results are grouped very close. I would put very little faith in any single G-tech test result especially a peak HP figure from the older model.
The G-tech also doesn't automatically compensate for varying weather conditions, so unless the results being compared were done very close in time when the weather conditions are the same, it is necessary to correct them for the atmospheric conditions. There are common formulas used to do this done by most chassis dynos, but the tester needs to record the atmospheric conditions and manually apply the correction formula to the G-tech results to be able to accurately compare G-tech results from different times/locations.
Finally, Ideally to get the most accurate results you'd use a perfectly flat road, but in actuality the only perfectly flat road I am aware of is a floating bridge. Comparative g-tech testing can be done on a not perfectly flat road, but it should be done on the exact same stretch of road, starting from the exact same position and heading in the same direction. Comparing G-tech results from different stretches of road, you can almost be assured that the results are going to significantly differ.
And one last thing: For horsepower testing, if the mass of the car (including fuel, driver, etc) is not correct, the G-tech hp number will be optimistic. This will not effect the 0-60 and 1/4 mi times. Which gear is used to do the hp and torque testing will also make an impact and the same gear should be used for the testing. Wide open throttle should be applied only to the selected test gear, because otherwise testing different gears will result in different levels of G-tech net horsepower.