Although we’ve already seen plenty of leaks and benchmarks based on Intel's upcoming Kaby Lake processors, some new details on temperature behavior appear to have surfaced. Already a few days ago Thomas Soderstrom from Tom's Hardware tested the temperature behaviour of Intel’s Core i7-7700K on a Gigabyte Z170X-Ultra Gaming. In the meantime he decided to also test the same CPU on an MSI Z170A Gaming Pro Carbon, which is when he noticed interesting changes in temperatures and power consumption.
Let’s first have a look at the power consumption. Using the Gigabyte Z170X-Ultra Gaming motherboard and running the CPU at stock clocks makes their platform pull 183 Watts from the wall. Switching to the MSI Z170A Gaming Pro Carbon motherboard that wattage, also measured under load, dropped to 145 Watts. This shows us, there is a 38 Watt difference.
When it comes to temperatures - for apparent reasons - there were significant differences as well. Conducting the same tests on their test bench equipped with the Gigabyte Z170X-Ultra Gaming motherboard 91°C were reached while with the MSI motherboard temperatures hit 68°C. If you head over to their article you actually won’t find these absolute temperature numbers, what you find is a deltaT, being the difference between ambient temperature - in at the editor’s site was a surprisingly low 15°C - and absolute core temperature. At this point it’s also not specified whether the editor is talking about CPU package temperature, the highest, the lowest or the average core temperature. So in the end, from a reader’s perspective, the whole thing implies quite a lot of guess work. Nevertheless there appears to be a significant temperature difference, which indicates something. If we were to guess the reason for this we’d simply say that the MSI board is feeding the board with less voltage than the Gigabyte motherboard. It could even be, that the MSI board has a higher vdrop. This would mean, that if the editor set 1.3V in the BIOS of both boards (MSI and Gigabyte) the resulting voltage under full load in Windows is not the same and that in the end can quickly cause significant temperature differences. Apart from that such a difference in vdrop could also explain the difference in power consumption.
So far it looks like we’re having a lot of confusing testing and test results ahead of us, which will make the period until the launch of Kaby Lake and Z270 pretty interesting.