Doubts about Rock 5b - before buy questions

Hello, can the USB-C port pass data or at least be used as ethernet interface Rock5 Model B 8GB RAM board? I’d like to power up the board via the USB C port of my iPad Pro M1, using the board to run CodeServer. (reference here and here

Thanks in advance for the information.

With my Rev1.3 board it doesn’t work to power it via a TB3 cable from the USB-C port of a MacBook since the board ends up in a boot loop. No idea how much amperage the MacBook delivers and about state of CC pins. In the pre USB-C days my MacBooks provided 1100mA at the USB3-A ports but with USB-C this might have changed (same might apply to iPads as well).

BS. Read any of the available reviews that report power requirements or just look here at this experiment powering Rock 5B with crappy consumer NVMe SSD in M.2 slot and powered by a dumb 10W USB-A brick.

1 Like

via the USB C port of my iPad Pro M1

Run into the boot loop on my Rock 5B (eMMC+AX200, no NVMe) with A12X iPad Pro.

1 Like

I just retried after disabling the A76 cores (adding maxcpus=4 to kernel cmdline), disconnecting NVMe SSD and network cable but still boot loop when powering via MacBook.

Disabling the A76 works:

tk@rock-5b:~$ -m
Kernel: aarch64, Userland: arm64

CPU sysfs topology (clusters, cpufreq members, clockspeeds)
                 cpufreq   min    max
 CPU    cluster  policy   speed  speed   core type
  0        0        0      408    1800   Cortex-A55 / r2p0
  1        0        0      408    1800   Cortex-A55 / r2p0
  2        0        0      408    1800   Cortex-A55 / r2p0
  3        0        0      408    1800   Cortex-A55 / r2p0
  4        0        0      408    1800       -
  5        0        0      408    1800       -
  6        0        0      408    1800       -
  7        0        0      408    1800       -

Thermal source: /sys/devices/virtual/thermal/thermal_zone0/ (soc-thermal)

Time        CPU    load %cpu %sys %usr %nice %io %irq   Temp   DC(V)
19:03:31: 1800MHz  0.57  10%   4%   5%   0%   1%   0%  25.0°C   5.40
19:03:36: 1800MHz  0.53   0%   0%   0%   0%   0%   0%  24.1°C   5.21
19:03:41:  600MHz  0.48   0%   0%   0%   0%   0%   0%  24.1°C   5.31
19:03:47:  408MHz  0.44   0%   0%   0%   0%   0%   0%  24.1°C   5.22
19:03:52:  600MHz  0.41   1%   0%   0%   0%   0%   0%  24.1°C   5.21

And when booting with an 15W RPi USB-C power brick inserted into my NetIO powermeter I can see a peak consumption of 4W while booting and after few seconds Rock 5B idling at 1W (measured at wall).

Since RPi 4 can be powered by an iPad it’s obviously not the 4W that are the issue but USB PD vs. dumb USB-C powering on the RPi.

Edit: when bringing back the A76 cores I see a 12W peak consumption directly after powering on Rock 5B.

1 Like

As the vscode part. I am running Vscode including some other applications as well.

  • VSCode server
  • Deluge torrent server
  • SabNBZD
  • PrivateBin
  • LAMP (Apache, MariaDB, PHP)
  • Nextcloud with Memcached enabled
  • Samba File server
  • Gitlab CE
  • MiniKube (all the above applications must in the end run within the pods within minikube. Easier to migrate if need to upgrade OS version)

My advice GO BIG :). Go for the 16 GB edition. I have bought 2 of them and one is currently a low power 24/7 server running the softwareapplications above.

1 Like

With all the problems around here, my suggestion is dont buy one. You are rolling the dice on getting a dud.

16g has slower ram than the 8gb. another way for Radxa to screw you.

Ooh I did not know this. But even so this compared to my raspberry pi 4 8 GB + m.2 NGFF(SATA) (Through USB 3.0) has for me personally been a much better investment. I noticed the speed immediately.

Anyways good luck searching for another SBC as all of us should have a SBC at home even if it is for the sake of blocking advertisement let alone preventing our children ending up on p*rn websites by DNS blocking.

OMG, the amount of BS spread here gets worse every day.

All three RAM variants perform exactly identical since the exact memory timings live in a boot BLOB that does CPU and DRAM initialization and uses same timings for all Rock 5B out there regardless which LPDDR chips are soldered to which.

The memory controller clocks the DRAM at either 528, 1068, 1560 or 2112 MHz regardless of what the modules are capable of as such 4GB, 8GB and 16GB perform exactly identical wrt to memory as long as same settings are used.

This is an 8GB Rock 5B:
This is a 16GB Rock 5B:

Memory performance identical since both run with performance DMC governor and the RAM is clocked at 2112 MHz all the time (see tinymembench and ramlat numbers).

If memory performance differs that’s because there’s something called DMC (Rockchip’s Dynamic Memory Interface) and with different settings different boards appear as performing different. You might want to search for ‘dmc’ here in my review.

Ah, ok, you’re just trolling intentionally.


Do NOT patronize me, jerk. You can take a leap into the lake.

I have seen this speed concern reported more than once. I have seen board images, its quite clear if you look. Its real, perhaps not the ones you have seen but i have seen it.

2nd the company IS trash. I was screwed, many others were too. if you had 1/2 a brain you would see the troubles people have have with their trash board, and that Radxa does not give a flying F about it. If they were not 1/2 way across the planet id have flown out to shove this trash down their throats. yes, its only a 150 bucks ( for me, who knows the total amount ) but its the principle of it. I am sick of companies like this, and people like you defending them.

1 Like

The guy from the lake is back !

Havent u gone away already? Last I checked you had a floating board sailing on some lake…

1 Like

i think you have me confused with someone else.

For anyone else reading this: this guy has absolutely no clue what he’s babbling about :slight_smile:

In this Android e-waste world we’re in memory timings are almost always defined in a boot BLOB and are the same for every RAM module of same type the board maker solders to whatever board. They only change when the BLOB will be exchanged and a new u-boot version gets flashed to devices/storage.

See this for example: (LPDDR4 still at 2112 MHz but LPDDR5 able to be clocked at up to 2736 MHz so once Radxa will do Rock 5B variants with 8GB LPDDR5 they are slightly faster than a 16GB LPDDR4 variant for sure).

As such it doesn’t matter if Radxa might have sourced faster specced LPDDR4 chips for the 8GB variant than for the 16GB variant since they all operate ‘below spec’ clocked at 2112 MHz max to provide some safety headroom.


Too bad it requires 15 characters.

All i have to say to you is 5 : F-off

EDIT: And im over this forum and arrogant assholes like you. My advice stands to the others: the product is a piece of shit, the company is a piece of shit, and buy at your own risk.

Dont agree with my advice, which comes from first hand experience of their piss poor product and zero response? Fuck off and die, literally…

Yeah … the kiddo from the lake is def back… if u hate this so much why do u persist? Theres other options outside radxa … wont make u smarter… u will stay the same cunt but at least with a smile on your face.
In case you havent understood this yet… you are not the majority who cant run Rock 5B. Plenty of us can are happy and keep in development… so… ask yourself why are you so helpless? Why do u keep on trashing what u dont understand instead of colaborating and learning?

F- u as well.

I do happen to know what i am doing. I got sent a screwed up board, it is simple as that. Most responsible companies would step up and take the crap back, but no, not these people. Its called basic customer service and supporting your customers, i’m sure even someone as stupid as you have heard of that ( and i wont even get into the PD issues they dont seem to want to deal with. Again, crap company, crap support. My advice to all is avoid these jokers )

Time to block notices from this PoS forum and all you ignorant fanboys.

1 Like

Simply leave the forum rather than spewing baseless claims. Insulting people and Radxa doesn’t help either. Stating that you’ve heard other people says something or you saw something doesn’t help, especially when you have not presented anything to back up those claims. You are just looking like a fool, and cluttering the forum with your nonsense.

Of course in theory its not the raw cost of the dev board that is the deciding factor. They are just that, dev boards so you can work on making a product. Not reall meant to be ‘the’ product on their own. You design your own board and buy the components you need after you do your dev work.

These do have GPIO and those x86 boxes dont, and lower power requirements being ARM. Plus the chip does have a decent NPU, while x86 does not . So while i agree with you from a raw cost for a ‘end user’ view, that really isn’t the intended market.

Lets see i have made one thread, commented on perhaps 2 others speaking from my personal experience with this crap product and company ( which is highly appropriate in this thread asking about if they should buy it. ) and that is cluttering? Freaking moron. people like you is why i’m now responding the way i am now. You deserve zero respect. So you get zero from me.

I do not need to back up any claim to you, you are insignificant. And as long as i am pissing people off for being honest perhaps i should stay. I can continue to try to steer some people away from the product, spend their money with a better company.

I have seen those, and they are not ‘new’ ( tho one of the most affordable for experimenters ), but not sure that is a true replacement for having native GPIO in a product.