Der8auer demonstrates the bungled welding of Intel Core i9 9900K processors

The soldering of the Intel Core i9 9900K is a real bungling (yet another) as well demonstrated by the expert overclocker Der8auer.

A few days ago Gigabyte published an overclocking guide for the new Intel Core i9 9900K processors in which a possible excess temperature was warned. These new processors are welded, so there should be a big difference in heat transfer compared to processors using dubious quality heat conductive compound. Apparently the temperature is not being expected, being higher than predictable and logical.

The Intel Core i9 9900K have temperature problems despite being soldered.

This new processor has eight cores and nine processing threads and to control the temperature the DIE and IHS are soldered, eliminating the low-quality thermal paste used in the latest families of processors and that has given so many headaches.

Der8auer, the reputed overclocker, has shown that soldering these processors is a real disaster. To demonstrate this, he has made the processor delid, removing the thermoconductive welding compound and has put a new one, specifically liquid metal, allowing the temperature to improve a lot, something that should not be the case.

For the tests, the processor was used at a frequency of 4.8GHz in a stress test of 10 with the Prime95 12K, the Core i9 9900K reaching 93ºC. After the delid, cleaning and applying the Thermal Grizzly Conductonaut compound and recomposing the IHS, the temperature has dropped by almost 10ºC, too big a difference (at most it should be 3ºC in the best case)

Also of interest is the die thickness, which has increased from 0.42mm in the Core i7 8700K (six cores and twelve threads) to 0.87mm in the Core i9 9900K. This issue affects the temperature, causing it to rise and worsening heat dissipation. It's worth noting that Der8auer has sanded down the die, something we strongly advise against for anyone without extensive experience, as it carries a serious risk of damaging the processor.

If we take a look at the CPU from the perspective that it is a system, then we have the silicon at the bottom, we have the thermal interface material, and then we have the heat sink ... If we make the chip much thinner, the Thermal conductivity should be much better, because essentially the circuits are at the bottom of the chip, they are not at the top. Therefore, the heat has to go through the whole chip, it has to go through the 0.87 mm of the silicon. If this was thinner, the thermal conductivity would be better

After this, he overclocked the processor to 5.0GHz and ran Primer95 12K again, bringing the processor to 96.5ºC before changing the thermal compound. After changing the thermal compound, it dropped to 88.5ºC, and by reducing the die thickness by 0.15mm, it fell to 84.66ºC ( NEVER DO THIS OR YOU MAY DESTROY THE PROCESSOR ). He then removed another 0.05mm of the die, and the temperature dropped to 83ºC. This is a significant temperature reduction, but unless you are an expert, this is not recommended for any processor. Lapping the IHS is easy, but it should never be done to the die, as it will usually break the silicon and ruin the processor.

It is clear that Intel has not excelled precisely in this case, although it is true that using that conductive paste that it uses in Coffee Lake, for example, things would be much worse. This to see the difference is very good and see the problem of Intel, but we want to reiterate that you do not do it in this processor because you will break it, surely and with any other processor on the market, because you will destroy it and the warranty will not cover you.