Z9 video: surprised how good UHD/60 is vs NRAW 4.1K

I have tried over and over again to prove to myself that the 4.1K/60 NRAW 12b NLOG is superior to the H265 UHD/60 10b NLOG, but I haven't found more than a single example to substantiate any benefit, and in all other examples the 4.1/RAW appears the same or even worse than the much lower data rate UHD/60/H265.

For context, the UHD/60 maxes out at 350Mbps (although in these tests my static clips averaged only 100Mbps, these being tripod shots of static objects so compression was high) , the 4.1K/60/Normal is about 880Mbps and 4.1K/60/High is about 1.8Gbps - so massively higher data rates.

I took a variety of test clips, duplicated in

1) UHD/60/10b NLOG H.265 Extended Oversampling On

2) RAW 4.1K/60/12b NLOG Normal

3) RAW 4.1K/60/12b NLOG High

4) RAW 8.3K/60/12b NLOG Normal

Effectively using the UHD max data rate as 100% relatively speaking the data rates for these modes are 1x, 3x, 5x, 10x (i.e. 8K/RAW/Normal being 10x larger file sizes than the UHD/60/H.265)

However, I can't in any of the 4.1K clips see any detail or compression benefit from the massively higher data rate, or 12b data sets. In only one clip, out of the many that I shot, with some deep shadow did I see that the 4.1K/high held a noticeable advantage on the noise of the clip versus the UHD/60 (H265). But this is one part of one clip that showed some advantage, rather than a general advantage.

By comparison, when using 8K/60/Normal which is 10x the data rate versus UHD/60 I very clearly, always, see significant advantages.

I've been previewing the content via Davinici Resolve Studio 18, with a Decklink card and 4K HDR grading monitor obviously in full screen, with a UHD project resolution. It's true that I haven't graded the colors, so given the monitor simulates 10b (8+2FRC) I might not be able to see any differences between 10b vs 12b anyway until perhaps some significant color or exposure adjustments, however I'm looking for detail as the starting test.

At this point, my initial results and conclusions point to either using UHD/60/10b or 8K/60/Normal and skipping using the 4.1K/RAW - however I'm reluctant to believe that 5x as much data wouldn't be expected to be showing better results.

Has anyone else tried to compare the UHD vs NRAW 4.1K quality? Or does anyone have some suggestions on how to better test this to really expose the difference, incase I'm missing something important here.