In this entry – example with actual data of B58 engine ignition.
As we can understand from the names of the graphs and the story in the internet forum, the correctly performing B58 stock engine (M240i) has been logged. Gear 4; in both examples, we see the accelerator pedal pressed completely (for 100%). The temperature of the inlet air is different: 29oC and 47oC.
Why do I use the logging data of another engine, not mine? The answer is simple: to exclude any suspicion that my car’s engine is damaged and the data are unreliable. But, if the parameters of several engines match, we assume that these engines are working correctly.

Light blue: position of the accelerator pedal;
bluish-green: short-term correction of the 2nd cylinder ignition;
black: short-term correction of the 4th cylinder ignition;
reddish/purple: overall (adaptive) correction of all cylinders.

Light blue: position of the accelerator pedal;
red: short-term correction of the 1st cylinder ignition;
light green: short-term correction of the 5th cylinder ignition;
orange: short-term correction of the 6th cylinder;
reddish/purple: overall (adaptive) correction of all cylinders.

What do we see in these images?
In the first image: in the 2nd and 4th cylinders, knocking starts; in the second image: in cylinders, No.1, No.5 and No.6, knocking is observed.
In this case, it is actually a “classical” knocking because, in the case of super knocking, the current cylinder is switched off for some time (3 .. 6 cycles). The reason for such a reaction of DME is simple – later ignition does not help in case of super knocking.
A second nuance we see is – the “recovery” speed of the ignition is around 3 degrees in 5 seconds (0.6 o/sec).
A third nuance – DME does not use any fixed “step” in case of detecting knocking but evaluates the required intervention depending on the amount of knocking (as we see, the impact depth is different: 2 .. 4 degrees).
A fourth nuance – in different conditions, the knocking is observed in different cylinders with varying RPM ranges. There are no one/several “culprits”.
A fifth nuance – the depth of the operative/short-term intervention is not relevant/significant – as we see, it is enough with a 2 .. 4 degrees later ignition to solve this problem.

In the following two images – the total ignition curve:

Red curve -correction of the ignition angle;
Bluish-green – total ignition target curve.

What do we see in these images?

Correction (adaptation) curves differ in both graphs; the correction depth reaches 5 degrees. The correction is not “simple”; it can be clearly seen that (at least, depending on RPM) correction is done with a fine “step”. The target curve is significantly different in both graphs; the target curve is not “simple” either – the ignition advance curve changes with a high resolution (depending on RPM).

I think both these graphs allow us to understand that ignition management is complicated enough. Yes, in these images, we don’t see the cylinder’s individual corrections/differences in their adaptation maps. The cylinder’s dependant/individual adaptation maps can be created only in stable driving conditions (RPM and required load for 10+ seconds and more). It’s necessary for DME to have enough time to measure necessary parameters (for example, the acceleration changes of the crankshaft in different sectors; to adjust knocking amount based on knock sensors data, etc.).

In conclusion: a very simplified story of how DME creates the adaptation maps. We will take the impact of the knock sensors as an example.
If DME detects knocking, it immediately creates a later ignition for the exact cylinder in the exact driving segment (RPM and required torque; it can be seen in the first images).

At the same time, DME records 10 .. 30% of the amount of short-term regulation in the adaptation map. So, if the “amount” of the problem was 3 degrees (applying 3 degrees later ignition, knocking disappeared), in the exact segment of the adaptation map, the ignition is modified for 0.3 .. 0.9 degrees later. Accordingly – if the knocking is a one-time “accident”, it does not significantly impact in the long term. Instead, during the next spurt in the same cylinder, knocking appears again:
a. the knocking amount will be smaller because the ignition is already slightly later;
b. DME will slow down the ignition for one more step for the same cylinder in the exact conditions.
In this way (step by step, cycle after cycle) DME achieves that it does not “step on the same rake” repeatedly – the engine knocking again and again, but – the knocking is avoided, and at the same time – the ignitions of all cylinders are very close to the threshold of the knocking.
DME observes the knocking data to avoid the possibility that, once used, poor-quality fuel will “break” the life for many years. If these data indicate too soft fuel combustion, SME slowly creates earlier ignition.
Of course, the actual algorithms are much more complicated. DME analyses each cylinder’s knocking and performance data and makes conclusions regarding common nuances of combustion in all cylinders.
Ignition management (short-time correction and adaptation maps) takes into account many nuances: inlet air temperature, pressure, and humidity; coolant/oil/engine temperature; octane number and “quality” of the fuel, and, of course, the basic parameters (RPM and required torque).

These graphs confirm my observations for 100% – B58 works VERY CLOSE to the knocking threshold. Creating an earlier ignition (and for at least SIX degrees, as several shippers are ready to promise) can be claimed only by a complete know-nothing. In the graphs, we can see that with the stock ignition advance angles, 5 from 6 cylinders start to knock, and their ignition advance angles are made later.
Second nuance, which has to be kept in mind – you can, of course, before “chipping”, fill the fuel with octane number 100, “turn” earlier ignition, do power measurement, and be happy with gained “power”. Or do more wisely – fill the fuel, drive for some time, and allow DME itself to create earlier ignition. In the second case, the result will be more stable, without knocking, but – with maximum (safe) gain.

Taking into account that the chippers community were not aware that the ignition system is adaptive, the following fraud scheme may be performed:
a. initial measurements on the power stand are done using AI 92 .. 95 fuel;
b. AI 100 fuel is filled; “tuning” (earlier ignition is created) is performed;
c. The measurement, which confirms the power increase, is performed.
From my point of view – it is a cheap fraud, even if the organizer of it does it because of insufficient knowledge.

Why the individual adaptation of each cylinder ignition is necessary?
Different cylinders have different burnouts on the walls of the ignition chambers and pistons. Different cylinders have different (and proportionally unchanged) amounts of air;
a. inlet manifold is not “the same” for all cylinders; the air vortexes are created in it;
b. the exhaust manifold (till the turbo aggregate) is not identical; the TwinScroll turbo aggregate creates a different load for each group of cylinders (1st to 3rd and 4th to 6th cylinder);
c. Technological differences of the Valvetronic hub create the different lift of the valves (amount of the sucked air differs).
In addition to the different amounts of air, the fuel parameters are also different. Even if the amount of the injected fuel is perfectly equal (DME adapted the injectors perfectly), the nuances of fuel atomization and its distribution (in the beam) are different.
Due to these reasons, cylinder-selective ignition and injection timing control and adaptation maps are created.