I need to rewire my fans a bit since I've got a 40mm fan in addition to the 92mm CPU fan on the CPU header connected via a splitter (which sends through only one of the fans' RPM signals). So right now I have no feedback on the 92mm CPU fan RPM values.
So far it seems okay. It is similar to what I observed in the past, which is that using all -30 curve optimizer settings lets CB23 run at 4300mhz while without the PBO2 Tuner setting the curve optimizer settings, CB throttles at 90C with the clock speed at 4100mhz.
I'm running a Velka 7. At 6 liters this is a huge step forward compared to the 10L cases that use 240mm AIOs. I have found that although 240mm AIO is more capable than a 47mm air cooler, the volume efficiency is just not even close.. The 4 extra liters I gain is well worth losing a bit of cooling capacity.
Totally agreed with thoughts on "space efficiency" of air vs water. I feel water is best utilized when it's compensating for poor airflow in the CPU chamber and moving the heat to a place with better airflow.
Cooling the 5800X3D is a challenge. Even folks that have it under water say it has a very "spikey" heat profile. It behaves different. It goes up fast, goes down fast. It's known to have a very small cooling surface area. So, when it's hot it's not "heat soaked" per se. I wish it was, because that is a condition where a water cooler can really shine. On air, I compensate for those spikes by bumping the floor speed of the CPU fans up a little. That keeps the CPU fan from rapidly changing velocity as it annoyingly "chases" those spikes.
For your situation, it sounds like you'll want to rewire those fans. Maybe get them off of the splitter so the CPU fan header can accurately read and set the speed of the primary CPU fan? Getting accurate rpm readings and maybe boosting your fan idle speed should help those idle temps.
Last edited: