9700K direct die outcomes
9700K direct die outcomes
Recently I installed a direct die on my 9700K using a Z390 Aorus Ultra and a Kraken X52 cooler. I bought the Rockicool direct die frame and the delid tool. I followed the included hardware instructions, swapping the plastic backplate for a 5.5mm nut and black washer as shown in the tutorial. I secured the standoffs for the air cooler and positioned the AIO on top of the CPU. I added four white washers to the standoffs and attached the thumbscrews. Here are my temperatures—uncertain if they’re optimal for this cooler and wattage. Core 4 and core 2 tend to run hotter, sometimes with a 15°C or more difference between them. The photo is taken at 5.2GHz, using OCCT with AVX2. 230W. Are these readings typical for the cooler and direct die setup? What about my core temperature difference? How might I lower it further?
First, I need to confirm what TIM you're using—liquid metal or regular thermal paste? If it's the latter, that’s likely the problem, as most pastes tend to match the surface they're applied to, especially compared to the original solder. Even with liquid metal, this isn’t unusual. AIOs usually have a concave base, so not all cores make consistent contact. The best fix would be to wrap the AIO.
Avoiding direct die can be beneficial. LM under the IHS tends to enhance core-to-core deltas because the IHS ensures good contact across all die sections, while the AIO maintains solid contact with most IHS areas. Temperatures may be a bit higher, but reducing hotspots could still provide extra OC headroom.
This approach seems promising in theory, but I’m uncertain if it will outperform a direct die setup. A tech demo showed direct die cooling handling 9900K temperatures with the same airflow as mine, and their peak temps dropped significantly. That suggests the core temperature difference also decreased. The cooler temperatures they achieved might surpass any IHS with LM under it. Still, I could be mistaken.
This situation highlights how the 7600X tends to outperform in the IHS benchmark compared to a direct die with an AIO from Asetek. The video runs a bit longer than it should, but the slightly curved cold plate on the AIO creates large core-to-core deltas.
I understand your point, but it might be due to the larger size of the different CCDs compared to the 9700K’s solid die. That’s something to consider. I don’t have the right tools to test it yet, so I’ll keep experimenting with the mounting pressure. It would be better to avoid constantly remounting if I only have a 1g tube.