Different coolers compared – convex + concave = contact ?
So we could already show that it doesn’t seem to matter which motherboard is used, which socket brand it has installed, and whether the CPU is operated with or without e-cores, as long as the total heat density about the same. The only remaining variable is of course the cooler, which has always been the Corsair XC7 LGA1700 in our previous tests. However, for the following tests, I also adapted an Alphacool Eisblock XPX Pro waterblock and a Corsair H150i AIO each to LGA1700 and ran through the tests with them once again. With the latter, I have no way of measuring the water temperature, which is why only the absolute core temperatures are evaluated here, but over a longer test duration.
Asus Maximus Z690 Apex (Lotes), 12900K, 51/0/49, 1.35 Vcore LLC6, Alphacool Ice Block XPX Pro:
| Washer Config | P0 max Δ | P1 max Δ | P2 max Δ | P3 max Δ | P4 max Δ | P5 max Δ | P6 max Δ | P7 max Δ | Ø | relative |
| stock | 78.1 | 82.0 | 82.9 | 86.9 | 85.1 | 86.9 | 80.8 | 82.0 | 83.09 | — |
| 1.0 mm | 73.3 | 80.2 | 77.5 | 84.2 | 78.6 | 82.1 | 72.4 | 77.1 | 78.18 | -4.91 |
With the Alphacool XPX Pro, it should be noted that it was not originally designed for the LGA1700 socket and was adapted accordingly by me for higher contact pressure. But even with this block, which has a relatively flat coldplate, the before and after picture of the mod is similar to the Corsair XC7 LGA1700. By relieving the CPU pressure in the socket, a better contact to the cooler and thus a saving of around 5 °C can be achieved here as well.
Now I also have the Corsair iCUE H150i RGB PRO XT with 360 mm radiator, based on the 6th generation of Asetek all-in-one water cooling systems. Historically, Asetek’s AIOs use a relatively convex bottom plate of the water block pump unit and this is what plays a significant role for the washermod. This is because the block can already partially compensate for the concavity of the CPU without the ILM mod and achieve relatively good contact in the center of the IHS (Integrated Heat Spreader). Since, as already mentioned, it is not possible to record the delta to the water temperature here, this test refers to the absolute maximum temperature of each core. In addition, to account for the heating of the water in the AIO, the test period was extended from 5 to 30 minutes.
Asus Maximus Z690 Apex (Lotes), 12900K, Auto, Corsair H150i 360 mm AIO:
| Washer-Config | P0 max | P1 max | P2 max | P3 max | P4 max | P5 max | P6 max | P7 max | E0-3 max | E4-7 max | Ø | rel. |
| stock | 82 | 86 | 85 | 92 | 88 | 93 | 85 | 89 | 74 | 76 | 85.0 | — |
| 1.0 mm | 81 | 84 | 85 | 92 | 84 | 91 | 85 | 88 | 74 | 75 | 83.9 | -1.1 |
Here, the convex shape of the AIO base plate becomes noticeable, which can already make relatively good contact with the CPU even without a mod. Although the mod leads to an improvement, this is only just outside the measurement tolerance at 1.1 °C.
Bend out of the box or inside the socket?
In the last part, I already showed you how concave my two Alder Lake i9 CPUs are. However, it is difficult to determine retrospectively if the bending was already present out of the box or if it was only caused by hundreds of hours in the LGA1700 socket under full load and corresponding heat cycles. So I got myself another virgin Alder Lake CPU for comparison, in the form of an i3-12100F. On the one hand, I was interested in non-K overclocking and on the other hand, the CPU is of course relatively cheap at the lower end of the product portfolio.
And indeed, this CPU is not concave, but even the opposite. It is also noticeable that it has an IHS with rounded corners, which indicates a different manufacturing process to that of my i9 CPUs. Whether these rounded heatspreaders are generally convex and those with sharp edges are generally concave is unfortunately again an unknown. And also the lower heat load due to the smaller silicon die and the presumed use of thermal paste under the IHS instead of indium solder as in K-CPUs are variables that we unfortunately cannot isolate.
But anyway, I just put the CPU into a socket without a washer mod to see what happens. After the first installation, the CPU was still unchanged and even after 8h in Prime95 Small-FFT at 5.1 GHz and 1.2V, the shape of the CPU seemed unimpressed. But after 2 days and various heat cycles, something actually happens and the IHS slowly loses its convex shape. Considering that my i9 CPUs have been in use for months, I wouldn’t be surprised if the i3 also had a concave valley in the middle after a few weeks, as long as it was consistently run in a standard LGA1700 socket.
31 Antworten
Kommentar
Lade neue Kommentare
Mitglied
Urgestein
Veteran
Veteran
Mitglied
Urgestein
Mitglied
Neuling
Veteran
Mitglied
Urgestein
Veteran
Mitglied
Veteran
Mitglied
Veteran
Neuling
Veteran
Mitglied
Alle Kommentare lesen unter igor´sLAB Community →