I have not been as convinced by AnandTech's articles since Anand Shimpi left for Apple several years ago and I have often found discrepancies like the one you mention above that deviate from the test results of other sites. As far as the others of your other observations, first look at the WCCFTech power chart (that site is generally source of unprofessional and inconsistent information) more closely since there is a clear flaw in their test results: the 8700K produces far more heat far than the 7700K yet it is drawing less power. From a physics-based standpoint, that does not follow since energy in equals energy out in any system meaning the increase of heat should be generally roughly proportional to the increase of power draw. Let me repeat: they were using premium liquid cooling (such as the Alphacool Eisbaer 420) in open air test benches for the test configurations at Tom's Hardware and at most of the other review sites.
More power draw doesn't always mean higher temperature, there is other factor to take into account. Like TIM quality and/or application. In 8700k case, it seems they improved a little bit the quality of thermal paste used as TIM (der8auer said it in the video i posted earlier). But a bad application and one of the core can reach higher temperature than other. I have a 3770k, it's first intel CPU without indium joint. When i stressed it, one of the core reach a temperature 8° higher than other.
The same sort of pattern applies to the other observations and can be summarized like this: there is no free lunch in adding 50% more cores of the same three generation-old CPU architecture (they are still Skylake-based CPU cores, after all). So it is naturally to be expected that an already hot CPU, the 7700K, would get even hotter with two additional cores in tow.
7700k and 8700k are still different (TDP,clock,process) and you cannot anticipate temperature by this statement. 7700k is using 14nm+ process, 8700k is using the last 14nm++ process wich mean less power draw at equivalent configuration. So it's not an apple to apple to comparison. 7740x is using 14nm++ process and draw less power than 7700k even with a higher TDP and base clock :
https://www.anandtech.com/show/1154...-the-new-single-thread-champion-oc-to-5ghz/17
Temperature depends on power draw, size of surfaces (IHS, Die, cooler) and conductivity of different material that are in contact (indium, thermal paste, nickel, copper, air, water...)
Power draw :
As i said earlier, it seems some motherboard apply auto-overclock and then they do not respect TDP set by intel. This is why some site have less power draw or equal to 7700k and other have very very bad result.
There is an other proof of that, the french part of tom's hardware tested the 8700 non k version. And they have almost the same power consumption :
http://www.tomshardware.fr/articles/test-coffee-lake-core-i7-8700k-8700,2-2733-15.html
If you take a look at other result, you can see that 8700 is often better than 8700k in some benchmark. Even them are surprised by this result as stated in this news (sorry it's french language)
http://www.tomshardware.fr/articles/test-coffee-lake-8700k-8700,1-65568.html
They said "Sachant que nous avons constaté les mêmes résultats sur des cartes mères de plusieurs marques différentes, nous penchons plutôt pour un microcode mal réglé pour le 8700, ou un overclocking automatique de certaines parties du processeur (interconnexion, cache ?)."
They think that these better result could come from auto-overclock or microcode error.
In this test (again in french sorry)
https://www.conseil-config.com/2017/test-intel-core-i7-8700k/2/ they found that there motherboard is setting auto-overclock by default. Resulting in more power draw, very high temperature (100° on air) and then throttling. When they uncheck this in BIOS the CPU is in TDP range and is like the other test i linked before.
One more test in french
http://www.hardware.fr/articles/970-17/attention-aux-overclockings-automatiques.html they found that there Asus motherboard is using auto-overclock when they set manualy there ram frequency. This result in higher voltage (0.1v) ,higher power draw (20w in there case), higher L3 cache frequency (4300 mhz) and better result (3.8% in application, 5.7% in games). Their gigabyte motherboard however is using TDP range set by intel and as you can see 8700k draw 100w from 12v when 7700k draw 109w
http://www.hardware.fr/articles/970-4/consommation-efficacite-energetique.html
Size of surfaces :
IHS and cooler are same size but Die is a bit bigger for 8700k so it may help a little bit to dissipate heat.
Conductivity :
8700k is using a better TIM than 7700k according to Der8auer.
To illustrate 8700K's temperature defiencies, here is a non-synthetic gaming test mapping out tempeature while using the Alphacool Eisbear 420. Per the manufacturer, "[t]he Alphacool Eisbaer 420mm is the biggest and strongest expandable CPU AIO worldwide." Naturally, therefore, most other AIOs would fare far worse in this test configuration. Higher workloads such as video encoding and synthetic testing will be naturally higher. Those higher workloads will result in even higher temperatures in cramped ventilation or with air cooling.
From Gamespot:
Temperatures
As stated earlier in the methodology, we used the NZXT Kraken X62; an all-in-one liquid CPU cooler with a dual-fan 280mm radiator. However, the 8700K still got pretty hot. At idle, the CPU sat at a mild 32 degrees Celsius and went up to 78 degrees Celsius under load during our runs of X264. With using the 5.0GHz overclock profile, it reached 86 degrees Celsius under load in X264, which is considered higher than desirable.
Let's break this down. At stock voltage and clock settings, the 8700K reached a temperature of 78 degrees Celsius. This was again with a premium AIO, NZXT Kraken X62, which is a $150+ solution with a 280mm radiator. By comparison, according to the Amazon reviews of the X62, the 7700K reaches maximum temperatures of the high 50 degrees Celsius with the same cooler. This is a twenty degree Celsius increase in temperature over the 7700K. Expect 80 to 90 temperatures with throttling with more normal liquid cooling solutions. Beware of throttling, possibly thermal shutdown and heat damage (as has already been reported with X299) if attempting air cooling.
In short, this is frankly disappointing for what is the flagship of Intel's mainstream processor line. I could understand this if this were an HEDT processor in their X299 line, but it is not. There were reports of overheating with 7700K in DIY systems and mass market builds due to users and manufacturers using standard air cooling (e.g. the Cooler Master Hyper 212). This common theme of overheating will only worsen and become more commonplace for 8700K due to its added 20 degree Celsius disadvantage over its predecessor.
As i said, i think these result are false because of auto-overclock of their motherboard and 8700k don't stay in his TDP range set by intel. Motherboard manufacturer are doing this to claim they have the best motherboard i guess.
This is why I picked up the 8700 non-k for my 3.5L build. However, I did pick up the Asus ROG Strix Z370I which arrived today. As soon as the CPU arrives, I'm tossing it into a DAN PC case until my Custom Mod arrives. I'll give a report on the temps.
If you try this with an Asus motherboard, i think you should take a look in your BIOS and disable auto-overclock. Or you will have some trouble like tom's hardware with a 8700 drawing as much power as 8700k