My Ryzen 1700's are running at 3 GHz and do 2 threads Microbiome Immunity Project and 14 threads Zika at the moment. Points output is about 120k wcg points per day.

My Ryzen 2700 running at 3.2 GHz, running the same project mix, averages slightly more than 130k points per day.

They all run Linux, points on Windows would be noticably lower for Zika.

You can also find detailed specs and why I chose the components I did here:
http://dc.sheridon.de/detailed-specs-of-my-crunchers

You are right, that in theory you should get exactly the same amount of run time per day (assuming you run you pc the same amount of time per day). However, run time is not counted when you complete the work unit, but when the work unit is validated. When that is the case, depends on the project. Some projects require only one instance of every work unit to be completed. Then the unit will be valid right after you sent it back. Others require 2 instances of every work unit to be completed. Then the second instance of your work unit will be sent to another PC and only after both instances are completed, you and your wingman receive the points. And if your wingman is slow, that can take several days. So because of that, points fluctuate a bit per day. Over longer time periods they will match what you compute per day on average. Over shorter time periods this is not the case.

Or to give you another example to understand the amount of time computed per day: Assume, you have a 16-thread CPU running at 3 GHz. And assume further, every workunit you compute takes 2.4 hours to complete. Lets ignore the validation issue and assume, every workunit competed is valid right away. This means, every thread can complete 10 workunits per day, or overall you complete 16x10 = 160 workunits. 160 workunits x 2.4 hours = 384 hours or 16 days.
Now assume, you reduce the clock rate of your CPU dramatically to 300 MHz (nobody will do that, but in theory you could). Now every workunit will take 24 hours to complete. Will your daily run time be lower because of that? No. Because you will complete 16 workunits per day now. And 16 workunits x 24 hours per workunit is still 384 hours or 16 days of run time per day. Of course, you will only get 1/10th of the points, because the work you do is only 1/10th. Points are a much better guide for how much you contributed than run time (while still far from perfect).

Contrary to what the thread title says I am seriously leaning towards getting a Ryzen 9 3900 X. I have one question would you go with the stock Wraith Prism Cooler Or would you go with a aftermarket liquid cooler? TD02-RGB Bearing in mind it will be consistently under 100% load with an ambient temperature of around mid to high 20s° Celsius. I am thinking it is safer to go with the aftermarket cooler.
TIA

If you plan to run the CPU at stock frequency (what I would suggest because of the higher energy efficiency), you should be fine with the Wraith cooler, even under 24/7 load.

My undervolted Ryzen 1700s run at 3 GHz (which is the stock frequency and any Turbo Boost etc deactivated) with the Wraith cooler and never went above 60 °C even at 30 °C ambient temp this summer.

The 3900X might be a bit harder to cool, as it is 105 W TDP instead of 65 W TDP, but on the other hand the included Wraith cooler is also a better version than that of my 1700s.

If you want to go with an aftermarket cooler, I would suggest to stay with air cooling instead of water. They are more reliable and consume less energy (which adds up when running 24/7)

I don't think an aftermarket cooler will increase the amount of work done, not even by 1%. The only scenario where this could happen is, if you do an extreme overclock and your machine is throttled due to dangerously high temperatures. But for a machine running 24/7 you should avoid coming anywhere close to this scenario anyway.

On the other hand, if you want to go with a good aftermarket cooler, this is not a bad idea - the better the cooling the better. The only argument for me against this is the cost - the stock cooler is just good enough for my purpose.

Undervolting is a good idea to decrease energy consumption while not sacrificing performance. In the end, cost for electricity is a major expense, potentially adding up to more than the hardware cost after some years. A CPU needs a high enough voltage to function properly. Which exactly is dependent on the specifics of every piece of silicon and cannot be predicted. So the manufacturers want to be on the safe side and specify a voltage, that will definitely work. If you test a bit, you can lower the voltage and try at which voltage the CPU still works fine. It will then run at lower power, but with the same speed.
Undervolting can be done in the BIOS settings with a suitable mainboard (for example B350, B450, X370, X470, bot NOT A320 for the AM4 platform).

Why deactivating Turbo Boost (or running at a low clock generally)? Because the voltage a CPU needs increases with its clock speed. The power consumed increases even more than voltage alone. So for a higher clock speed the electricity consumed increases overproportionally. Therefore I run my CPUs at a low clock speed, because they are more efficient this way. I would rather invest the saved money for electricity in another cruncher, than overclocking the crunchers I have. Saving an expensive cooling solution because cooling is much easier this way, adds further to the savings.

Which clock rate is the best for you, is obviously also dependent on you electricity cost. Where I live cost is high, so I run at a very low clock rate. The optimal choice for you might be different. But any extreme overclock is not a good idea in any case.

Some numbers as an example, my Ryzen 1700 consumes the following power (complete system, no GPU installed, 80+ Gold PSU, measured at the wall):
@ 3 GHz, not undervolted: 104 W
@ 3.5 GHz, undervolted: 105 W
@ 3 GHz, undervolted: 75 W