I’ve trasferred the system to SSD using emmc-install command and everything looks fine, overall responsiveness is now very, very good. Unfortunately there is a problem with the video, similar to a loss of sync in the signal to the 4K display. Is it something I’ve already seen in the Manjaro I’ve tried.The problem gets worse when anything is written to SSD, much less when it is read, although the glitches still there. This problem unfortunately impairs the use of the system significantly.
From another recent thread that at a given point discussed my attempts with the Majaro ARM 20.04 distro (also with kernel 5.7 if I recall it well but not sure) I reported @Spikerguy a problem that looks like very similar, supposedly related to the device tree.
Sorry to create this confusion, no video playback was intended. I attach here below a short movie capture showing the problem. You can note that when the file copy starts the artifacts are produced on the video screen. In the test shown the file is read/written onto the SSD itself, but in normal operation when no SSD activity is involved in the system such problem never occurs. Similar artifacts are shown also with diagnostic messages when booting, in a way that recalled me the same behaviour noted with my Manjaro test (that is why I made a reference to it, perhaps concerning a totally different problem). However, the SD image without mounting the SSD is perfectly fine. System after initial SD installation has been copied to the eMMC (via emmc-install). It does not boot from SSD of course, but from eMMC however with further option of system being transferred to SSD according the relevant emmc-install option, as per its df output:
Seems like a power issue, because with the data you have given, all look like heavy loads, those SSD’s consume a lot of power, I am not sure about peak power draw, but the idle is something like 200mW, so consider your power supply
I dont think its a power issue, I also have been suffering this on my HPZ32 4k monitors… I also have an Odroid N2 and that board also suffered the issue in the early days and if memory serves me correctly it was related the the cpu governor… It was sorted mid last year in an updated image and never gives any issues now. But still stops me using my Khadas Vim 3 annoyingly.
Yes, I’m using the M2X board, I’ve tried to reseat it. Power supply is Khadas original. I attach a screenshot with some of the messages from the very first boot stage. As you can note there are some repeated messages pertaining to voltage selection but I’m not sure whether they are really relevant.
I did not get a chance to check this last night. hopefully tonight I will get the time to load that image and test.
Regarding power supply, Khadas offers two types, a 10 watt and a 24 watt PD. While the 10 watt supply is adequate, connected devices can can result in current draws that could create a power starvation issue.
Even if you are using the Khadas 24 watt supply, just to rule it out, if you have a suitable substitute power supply, give it a try.
I will let you know if I experience the issue when I run that image. It will be several hours before I get to the VIM3.
PS is 24W model (FC53). I tried a 5V/3A and the problem appears the same to say the least - indeed my hunch is that with non-original fixed 5V/3A PS the problem even gets worse. Just a little note about the screenshot I attached in my previous post. The complete boot sequence is made of two stages at different display resolution. The photo shown refers to the first stage of these two and in spite of its short duration happens to look stable. Next, the second stage changes to a higher resolution with proper messages coming from the linux kernel outright. It is at this stage the flicker begins to happen.
Also, let me add a further observation hoping to give the clearest picture I can.
The vey same equipment (i.e. VIM3/PRO with M2X board and M.2 NVMe) perfectly works with other images kernel 4.9 based and CoreELEC 9.2.2 or LibreELEC that I equally tried - not giving this specific nuisance at all. Hint is that no NVMe utilization is done by these software running.
I did this the first time, but realized I grabbed the 4.9 version. After realizing you were on 5.7, I used the SD 5.7 image.
In both cases, I installed to eMMC/NVMe. Both kernel versions installed without issue. Also in both kernel versions, I noticed no video artifacts when transferring files from USB SD card reader to NVMe or visa versa. The files I transferred were some video files and they played from either location.
Keep in mind, I do not have a 4K display, mine is 1080p.
Thank you for your replies.
The problem is still there even changing resolution to 1080p. I’ve not tried to change the display itself to a 1080p though.
I’ve also tried to exclude the M2X card connecting the M2 SSD directly to the main board but problem also still there. For further reference SSD drive is SABRENT ROCKET NVMEe PCIe M.2 2280 SSD.