Can the OPI5 be Configured to Run LCNC?

When I boot the rt kernel from the sd-card it recognizes my nvme.With non rt kernel is booting from the nvme. I put the image with Kernel 6.1.75 non rt on nvme and after that I ran from it. Then I installed version 6.1.75-rt23 running from nvme and made the link. I restarted and it didn't boot anymore.I wrote several boot images from your links and it does the same thing.I have put all the previous versions on nvme so far and I have never experienced this.Please look at the following images and let me know how your system show when you do this:1. run linuxcnc2. open Htop and select linuxcnc process. Press "a" to show related cores affinities to linuxcnc.As you can see in my pics 5,7 cores show 0.00% usage.  They are not used. What it show yours?
Kernel 5.10_nvme: 
Kernel 6.1_SDcard:   

Strange that it doesn't work on yours. Maybe I oversee something, although here it works on both, the last things i can try is using the same config as the non-rt version that works for you and only change the following settings:
github.com/orangepi-xunlong/linux-orange...s/rockchip_rt.config

Already tried another power cord/adapter?

Did you try to write the bootloader to mtd (spi flash) from my previous message?
Are there other people who experience the same when booting from nvme?

Mine shows the same cpu usage by the way, even while running latency-histogram. You'll see that rtapi (the realtime component) with priority -99 runs on the last cpu, so it works how it should.

I tried the bootloader you gave me and it doesn't work.

The 5 and 7 cores are isolated and linucnc works with highest priority on 0,1,2,3,4,6 cores (Htop shows this).
So nobody uses 5, 7 cores. Why isolate them if you don't put linuxcnc on them?
I noticed that my linuxcnc works better with all cores enabled. I actually watch StarTrek when the cnc is running.

Alright, now took my last attempt of building another kernel, if this one still doesn't work it should be something else. The kernel is in the Test folder.
drive.google.com/drive/folders/1r76sUsfG...6fGcz?usp=drive_link

The realtime component is on core 7 by the way, in htop you should select the one with -99 priority before you press "a'. The gui isn't realtime so no need to put that one on a isolated core. But without isolating cores the results are still good enough indeed, so if that experience is better it might be the better choice. 
Just tested without isolating a core and the latency on the servo thread went up to 40us, while when I only isolate core 7 until 3.3us. The kernel in the Test folder is with cpu governors, so I put these via armbian-config at performance and lowest frequency at the highest.
With only core 7 isolated it has no frame drops at 1440p@30fps, unfortunately there is no full (ffmpeg) hw acceleration in chrome anymore, where the cpu usage stayed below 10% @4k.
 

With the latest deb's from the Test directory it did not boot.

I downloaded the image you put "Armbian-unofficial_24.11.0-trunk_Orangepi5_bookworm_vendor_6.1.75_cinnamon_desktop-002.img" and big surprise, it stuck when booting.

I used the image "Armbian_24.8.1_Orangepi5_bookworm_vendor_6.1.75_cinnamon-backported-mesa_desktop.img" (the one in your link on page 35), and it boots and I can apply rt kernel to it.

The answer lies in the differences between the two non real time images. One boots and the other doesn't.

Do you think that if you applied configuration for rt-kernel on the image "Armbian_24.8.1_Orangepi5_bookworm_vendor_6.1.75_cinnamon-backported-mesa_desktop.img" it would result in deb's that would work?

I don't understand exactly what you mean, here all the deb files work with boot from sata and nvme on the orange pi 5 and 5 plus, so now it also works on yours?