As you surely know, the new Intel Alder Lake CPUs have not exactly hit the market without problems. We are not talking about the stock of its models or the DDR5 itself, but about temperature problems, which we have already dealt with in previous articles and a partial solution was seen. Now a new “trick” or mod (whatever you want to call it) corrects even more the defect that these CPUs integrate. How do you do it? Does it fix bending problems on Intel Core 12 CPUs?
It all has to do with the pressure exerted by the now called ILM or Independent Loading Mechanism, which is the part where the docking mechanism makes vertical pressure down on the processor so that it touches the pins. The IHS are logically prepared for such contact, but since the CPUs are now longer (rectangular), the problems generated have changed and Intel was not counting on certain imbalances in said pressure.
CPU bending returns, total disaster on Core 12?
Although many do not believe it and having a server more than 10 years of experience with delid and PCB corrections what is happening with the Intel Alder Lake CPUs is not surprising. The pressure points are only two, longitudinal and far from the points, since as we have commented and you can see with the naked eye, the CPU is rectangular and not square.
The problem, or rather, the problems, are two that interchange and complement each other:
- The solder of the die and the IHS concave the center of the CPU.
- The anchor bends the PCB and increases the cavity effect.
Therefore and although we already recommend that in the case of heatsinks or blocks with a backplate, this should be installed first and then the CPU to try to avoid this effect as much as possible, the reality is that after spending a certain time and with the thermal deviations of cooling and heating of the materials the stratum ends up yielding as well as the pins.
The Alder Lake Bending Solution
The solution is simple, inexpensive, and easy to apply: spacer washers on the CPU retainer socket or clamping mechanism. All you have to do is remove the four anchor bolts and place the washers (preferably nylon) and then reassemble the system. The most obvious question would be how thick? The comparison is not wasteful, since four different measures have been tested: 0.5mm, 0.8mm, 1mm and 1.3mm. The result? Come and see:
The pressure exerted from the retention system to the CPU is so high that with only 0.5 mm we already gain almost three degrees, with 0.8 mm there is hardly any improvement, but with 1 mm we will win 5.76 degrees, something really remarkable. Already with 1.3 mm the data worsens achieving 5.04 degrees improvement.
This evidences what we already knew: Intel CPUs do not achieve good contact with cold plates or cooling blocks, something that did not happen before but simply was not an optimal contact because it was precisely the center that ended up touching leaving the periphery “in the air”.
In short, in Alder Lake it is more optimal to place the Nylon entry washers before placing the heatsink to improve contact by reducing the pressure of the anchor. It is also up in the air if the bending of the Core 12 to the anchor manufacturer (Foxconn, Lots etc.) or if it is an Intel design problem. Didn’t they take into account the thermal cycles and their influence on the substrate? Did you not realize that the pressure exerted was greater than necessary to ensure contact between pins and PCB as well as between cold plate and IHS? We will surely never know …