Poor cooling in Intel’s Alder Lake – tracking down problems with the LGA-1700 socket, including possible remedies
Badly fitting coolers, insufficient performance with actually good water blocks and warped CPUs – not only the feedback from readers is increasing now, but also very similar reports from board partners, cooler manufacturers and system integrators. Based on my own experiences, I went on a search for clues and first tried without emotion to sort out the error patterns that had occurred. Finding a cause here is certainly not easy, since the phenomena are quite complex, but certainly not impossible either.
Are the heat spreaders arched excessively or is the base unstable? Are the specified (and unfortunately also necessary) contact pressures simply too high or are the cooling systems’ fastening systems rather useless? One could also speculate about circuit boards that are too thin and flexible and still not find a clear solution, because in the end it is probably a causal chain in which several factors flow and from the result of which the problems described arise. I had already reported in very detailed articles about Intel’s new socket including the mechanical features before the launch, so I would like to save myself this excursion into the details at this point. If you want to read it again in more detail, you can do it here:
Intel socket LGA-1700 and LGA-1800 in detail – Exclusive data and drawings for the new CPUs from Alder Lake
I had already given various assessments of the heatspreader of the new Intel CPUs both in the unboxing article and at the launch and even praised it as quite flat and not particularly curved. In doing so, I relied on my own measurements of the CPUs available at the time as retail copies or engineering samples, which I had even checked in the 3D scanner. After more than 20 tested and, in some cases, painstakingly measured CPUs, I was pretty sure that the IHS would fit better this time.
If you put the test to the test, you can of course come to a different conclusion at first glance. Cooler manufacturers such as Noctua have also noticed that the support of the coldplate / heatsink and thus also the area that can be used for cooling is far from being optimal. The various cooler manufacturers are aware of this problem and it has often been reproducible (as in the picture) with a new CPU and an unused motherboard including a brand new cooler.
At this point I have to add that it is not caused by the cooler manufacturers and that Intel is still sticking to the specifications with most CPUs (but not all!). And yet it has hit us several times in practice, so that today’s article was also created in cooperation and after consultation with motherboard manufacturers and cooler manufacturers. Because at some of the companies there have already been increased customer inquiries and even RMA cases relating to the poor cooling.
In our case, there were up to 9 degrees (!) Between the individual structures, depending on the water block, CPU and motherboard, which goes far beyond possible tolerances. The picture of the thermal paste after dismantling was also very different and was similar to the result in the picture above. But where do such imponderables come from, which affect even potent water cooling and drive the CPU into thermal throttling? On the next few pages we not only see one of the possible causes that nobody had considered before the launch of Intel’s 12th CPU generation in this form.