Image | Radxa ROCK 5

[docgalaxyblock]

Formal device information

• Device name | Rock 5
• Official product URL | https://wiki.radxa.com/Rock5
• GitHub resource URL (not yet avaible) | https://github.com/radxa/
• Image download URL | https://wiki.radxa.com/Rock5/downloads (not yet avaible)

Is the SBC officially supported by the Debian installer?

• Seems like not

If not, is a reliable 3rd party Debian image available for this SBC?

• Radxa should provide one at there Wiki

My Wishes

• Hope you can develop a good stable image like your rockpi images 😉
• Hope you will get a dev sample eg from here, ask the shop and maybe you get one 🙏
  (seems like they have a preorder system like the steam deck https://shop.allnetchina.cn/products/rock5-model-b-rk3588-pre-order-promotion-voucher)

I missed the preorder start ... hope to get one to move some homelab services.

I hope the Dev team has some time 🙃

Greetings from Germany 😃

---

Vote for this image via up button at this post.

[Joulinar]

Just out of curiosity, did you tried the Zero image?

https://dietpi.com/downloads/images/DietPi_RadxaZero-ARMv8-Bullseye.7z

[MichaIng]

Jep, requires a new image, similar to Quartz64. I'll ask Radxa for a developer sample.

[isarrider]

MichaIng

did you get a reply - I am still waiting for my preorder to be able to ship...

[MichaIng]

I just sent out the request, along with one for ROCK3 (Model A) and the outstanding Radxa Zero shipment.

[docgalaxyblock]

Did you received a sample yet?

[isarrider]

I wish - but sadly no...

[docgalaxyblock]

The DEV samples are finally there ! 😍
https://forum.radxa.com/t/rock-5b-debug-party-invitation/10483
@MichaIng Ask again for every Radxa sample you did before

[MichaIng]

Converted into discussion, so here votes can be placed.

[isarrider]

it seems the boards are starting to ship,
so is there an update?

[docgalaxyblock]

AFAIK (as a user) they hope to get all preorders processed this month.
@MichaIng Did you had success regarding a dev sample? (Also the other radxa board?)

I have preordered in February and have not received a code email yet.
🤞

[isarrider]

I preordered in Jan and my board seems to ship soon...

[MichaIng]

Sadly no feedback. Armbian has unofficial community releases in the meantime, sadly for Debian Sid (unstable) only.

[docgalaxyblock]

I just got the order link emailed by allnetchina 🥳

[docgalaxyblock]

It arrived today - it took just one week to germany
https://cdn.discordapp.com/attachments/494242871583571971/1034755314207162399/IMG_20221026_110712.jpg
https://cdn.discordapp.com/attachments/494242871583571971/1034755798426996736/IMG_20221026_110952.jpg

[lukaszsobala]

There is a Rock5 Debian image available on their website (as the OP wrote)

[MichaIng]

Also Armbian legacy (vendor kernel/u-boot) is available now. But I'll wait for a mainline kernel image before I integrate it into DietPi.

[RainmakerRaw]

Watching this with interest. My Rock 5 model B 16GB arrived today. I'll flash their Debian Bullseye image for now and just strip out the X and GUI stuff, but the moment DietPi is out it'll be getting re-flashed!

[MichaIng]

PR up to add support: #5917
Test image builds are running.

There is one issue with device tree overlays (for enabling I2C/SPI/UART, 8k support for HDMI and such): The boot script uses a (file name) prefix to load (only matching) overlay files, but the ones provided by Armbian's current RK3588 kernel package have several different prefixes valid for ROCK 5B. So depending on which one is provided in dietpiEnv.txt, either only one set or another set of overlays will work, and the fixup to repair after applying an overlay failed won't work at all, as it has an individual prefix 😄. Will be fixed earlier or later, but until then, we need to use a highly customised boot.cmd/boot.scr which doesn't use prefixes at all so that you can enter it freely into the overlays setting in dietpiEnv.txt.

[lukaszsobala]

I might try them soon if I manage to boot the thing... On its way now

[RainmakerRaw]

I still hadn't gotten around to booting mine. I'll re-flash the eMMC to this test image (Bullseye for me) and let you know how it goes. Thanks for all the hard work!

[docgalaxyblock]

The Bullseye Image boots and I did a dietpi-benchmark
~10:25 UTS+2 should be the upload time as a generic device

~10:36 i tried to install a desktop but it has some problem regarding alsa
bug report reference is fca71361-...
workaround by replacing alsactl -g store by another command and the desktop installs
Now I need to go

[MichaIng]

Many thanks! Ah, makes sense to recreate those images with dev branch, so the SBC name is correctly detected. You can switch easily:

G_DEV_BRANCH dev

Mind-blowing:

BENCH_CPU='1.87'
BENCH_RAM_WRITE='1023'
BENCH_RAM_READ='3684'
BENCH_CPU_TEMP_START='32'
BENCH_CPU_TEMP_END='36'

The fastest ever measured SBC until now was Odroid N2+ with 3.56 (average for Odroid N2 is 4.01) seconds: https://dietpi.com/survey/#benchmark
So this one has more than doubled CPU power. RAM speed however is on same level with Odroid N2, both with some distance fastest SBCs in our RAM benchmarks: https://dietpi.com/survey/#bench_ram

i tried to install a desktop but it has some problem regarding alsa

So HDMI audio doesn't seem to work OOTB, I guess. What does this show:

aplay -l

Ah, from the bug report upload:

**** List of PLAYBACK Hardware Devices ****
card 0: rockchiphdmi0 [rockchip-hdmi0], device 0: rockchip-hdmi0 i2s-hifi-0 [rockchip-hdmi0 i2s-hifi-0]
  Subdevices: 1/1
  Subdevice #0: subdevice #0
card 1: rockchiphdmi1 [rockchip-hdmi1], device 0: rockchip-hdmi1 i2s-hifi-0 [rockchip-hdmi1 i2s-hifi-0]
  Subdevices: 1/1
  Subdevice #0: subdevice #0
card 2: rockchipes8316 [rockchip-es8316], device 0: fe470000.i2s-ES8316 HiFi es8316.7-0011-0 [fe470000.i2s-ES8316 HiFi es8316.7-0011-0]
  Subdevices: 1/1
  Subdevice #0: subdevice #0

alsactl: add_tlv_comments:216: snd_tlv_convert_to_dB: Invalid argument

So somehow the default states cannot be stored:

alsactl init # should reset ALSA states to a valid default
alsactl -g store # should store those defaults to ALSA states file, but fails

Similar to this issue on RPi: #4306
We still work around this by ignoring the error in case of RPi. Not sure if still required, or whether we probably should ignore it in all cases. amixer would fail directly if we tried to add invalid arguments ourselves, and alsamixer run by user likely solves the issue, and states are stored automatically on regular shutdown, so alsactl -g store is only to force it in case or crash or such, to assure a valid default state on next boot. If alsactl init does not init a valid default state, this is redundant anyway.

[MichaIng]

It is now generally ignored if this command fails. It however means, if it keeps failing, that amixer/alsamixer changes do not survive reboots, so in this case it would need to be further investigated: e616453

[isarrider]

I am also playing around with it now - thank you...
Docker wont install - error log 2baa9a1b-e962-4f0f-8193-cb02241c112e
I am checking currently - I remember that with the Armbian Image
I needed to change sth too... (but cant remember clearly)...

[MichaIng]

Interesting, that kernel should definitely support nftables, also otherwise dietpi-software would have applied legacy alternative automatically. But probably some nftables sub feature is missing or so.

[isarrider]

shall I open an issue or only when this becomes "stable"?

[MichaIng]

Let's finish these device specific things here.

[lukaszsobala]

I tried to install ALSA but it failed - there should be a report from an hour or so ago (EDIT: I can see this was spotted earlier). Many other things installed correctly.

Another problem I saw were artifacts when playing a YouTube video. I haven't changed any graphical options or played with drivers yet, although I intend to.

Also the fan doesn't spin but I know it's plugged in correctly (but I assume I need a deb from radxa).

One more thing - I will need to investigate if WiFi indeed works, as it seems connected but may be intermittent. Ethernet works great (>900 Mbps on a Gbit connection).

[MichaIng]

As well with alsactl -g store as reported above?