I bet the IC or a clamp is leaking. Boost leaks can cause big KR. If you lo timing along with KR you will see that it does no work like phantom said, although it should. The KR actually seems to be the measurement from the knock sensor NOT the amount of timing being pulled at that moment. Like i said if you log them together you can see that. I have seen up to 5.5 on my KR but the timing does not change. Then on a log with 6.0 KR you can see a big drop in timing. The way the PIDs work are a little funny with the DH sometimes.
At the risk of being Mr. Literal again... I would disagree a
boost leak would cause knock. A
boost leak should, in theory, cause the engine to run rich, which generally helps prevent knock. A leak pre-turbo while under boost would lead to potential lean conditions, but I don't see how leaking off boost would cause knock.
Also, I disagree with the statement that KR is not the timing being pulled. Take a look at this log:
While it is certainly only one log and certainly not conclusive, there appears to be a definitely relationship between the slope of the spark advance and knock retard lines. It is hard to determine at the onset of the knock retard because the engine load is rapidly changing and thus the spark advance is changing. Yet, once I settle into a cruise situation, as the knock retard falls, the spark advance increase. When the knock retard once again increase, the spark advance once again goes down in a corresponding manner. Now, admittedly, for one reason or another the spark advance does have a significant drop with no significant corresponding change in KR, engine load, RPM etc... which admittedly I can't explain. Though not shown, later in the same log from which that image came, I have a similar situation -- the spark advance and knock retard lines are closely correlated.
I would certainly be interested in seeing any logs you have with which to compare data. I also have plenty of log data where it seems like spark advance and knock retard aren't directly correlated... but the key difference is that the engine load and rpm are in a state of significant change and thus the spark advance is in rapidly changing so it doesn't seem to be a situation where one could easily compare spark advance to knock retard with any clarity; since load and RPM are changing the spark advance is changing. In my opinion, a constant load/RPM situation is the best to determine the correlation since the spark advance should be constant allowing for a direct comparison of spark advance and knock retard.