Hi Michael.
At first blink I found your idea great, but after some thoughts, some thing upsets me.
You are comparing two different platforms.
System 1 Intel Core i7-6850K and system 2 AMD Ryzen 9 9950X3D.
And RTX 3080 vs 5090.
I guess ram DD4 vs DDR5.
When you change ONE parameter, you can definitely know what changes. But when here you change both GPU but also CPU, what of the two parameters is the culprit to better or slower performance? GPU? CPU? Both?
And with three parameters with Ram change, what is going on exactly?
Do you still have your system 1 available, where you could just swap GPU and going from 3080 to 5090 and see what happens?
Then going on system 2 and putting with your brand new 9950x3d the 3080, then swapping with 5090.
At least with that kind of test, we would know EXACTLY what GPU/CPU changes, in what proportion/percentage. Despite the Ram change that I think must not change a lot the results.
Because I’m sorry, going from System 1 to 2, you can just compare ONLY those 2 systems. But you can’t conclude for sure that others will get similar percentages results. You can’t get a general law I mean.
If you are on Zen 3 with a 2080, what happens when we go to 9950x3D and 5070?
You apply you percentages and that’s it? I don’t think so.
5090 has 21,760 cores.
3080 has 8960 cores for the 12Gb version.
You speak about 5090 3.5 times faster than 3080. Where did you get that? Topaz relies mainly on Cuda cores as far as I know. Not only that but mainly if you want the fastest speed, you need to get the card with the more cuda cores.
21.760 / 8960 = 2,42 times faster.
You started from 5090 is 3.5 times faster than 3080 theoretically. Ok, but where do you get this figure? I’m skeptic.
I remind one thing:
when people do reviews on CG, they use a platform/system and they test:
4050 then 4060 then 4070, then 4070ti, etc.. They ONLY swap CG. They don’t change ANYTHING EXCEPT the GPU.
Same to test CPU, they swap only CPU.
But here, you draw conclusion and percentages, but that only works comparing your 2 systems.
Now maybe we could speak of a tendency using your figures that:
If you want better results, buy a more powerful GPU except if you are mainly using hyperion and aion. But even that would need to be confirm with at least a couple of guys doing like you, testing their old system, then next day they receive their new one, testing on the new one and see if they get similar tendencies as you.
Same with Ram. Is ram affects speed? Do CL32 slower than CL28 and how much? Does ram 6800 allows faster topaz speed than Ram 6000mhz? So you have to test ONLY the ram.
What do you think?
Thanks. 
