OKdoky GoPiGokay? Nope

It’s too early to say just how much work it would take to get the really fresh Okdo ROCK 4 SBC to work for GoPiGo, and I’m not hurting for challenges at the moment, but it might be possible:

1 Like

they claim “similar form factor” to the Pi. I didn’t scrutinize the pin out to see if it’s the same.


Supposedly pin out compatible except for POE hat.

Other reported diffs:

  • the processor is on the bottom of the board
  • no numerical processing unit? (don’t know if Pi4 actually has one of these)
  • Youtube video and video playback is worthless
  • only boots from sdcard and emmc
  • supposedly runs hotter

What OS?

Will it run Raspbian?

1 Like

No only Ubuntu and Android.

They have Ubuntu 20.04 Server and Desktop images (with Xfce which is a “lightweight GUI” ) available, but supposedly “the coming soon Rock 5” version of the board will target the Pi4 Desktop capapabilities.


Then we’re not going to be able to make a “GoPiGo-3 O/S” for it.



May not be able to make a GoPiGo3 anything with it. Physically the space below when mounted in the Gopigo3 may not work for heatsinks


Putting the processor on the bottom makes it difficult.

I had problems with the heatsink on the top. I had to trim the shroud and carefully use a razor knife on the spinning fan to shave enough away to prevent interference with the GoPiGo’s board above it.

It might be doable with a longer set of mounting studs but I don’t know if I will be able to get one to experiment with.


There are some other reasons this board is not a “drop-in replacement” for a Raspberry Pi:

  1. gopigo.py expects spidev0.1, but the Rock Pi offers spidev1 so the processor with official GoPiGo3 software will not communicate with the GoPiGo3 board.
  2. GoPiGo3 depends on pigpiod to drive the GPIO - the Rock 4 SE requires different GPIO interface code
  3. It is not cheaper or faster than a Pi4 (unless you are crazy enough to do bitcoin mining on an arm64 processor)

#'s 1 and 2 are potentially solvable.

Not sure how, but if clever people can get Win-10 VM’s to run on a Mac, or RS-232/RS-485 to run via USB, this should be a cake-walk by comparison as (AFAIK) it’s an interface API issue.

I don’t know exactly how to do it, but it’s not “moon mission” complex.


Sure, with cooperation of the board manufacturer and big investment of time by people smarter than me. BTW, to the suggestion to allow headless first-boot disable of IPv6 Radxa replied:

“IPv6 should just work”What are they smoking???


Somehow I thought this situation was like the Intel-AMD PC situation of 50 years ago. AMD put out a motherboard and a set of drivers and “Voila!” DOS/Windows3.1 and all those ISA cards just worked. (If you could get the interrupts not to collide.)


That’s correct. IPv6 should “just work”.

However that attitude ignores the realities of today’s network environment.

Most of the network stacks on what are even bleeding edge systems and software are based on code written in the 70’s by Bell Labs or U of C Berkeley and many systems, even new ones, struggle with IPv6.

I don’t know where they live or what geniuses they have working for them, but asking people to build V-LANs or do other complex things when a simple “on/off” switch in their products will solve the problem is simply irresponsible product management and shows a callous disregard for the network realities of today’s users.

Feel free to write back and quote me.

I might not be sufficiently polite if I do it myself.

1 Like


Every one of them.

Every single device, every single managed switch, (that can be programmed), and every single router because all it takes is one device with a crappy IP stack to bring down your entire network.

All of this stuff may work well for you, and if it does you have my admiration - and please tell the rest of us exactly how you did it.

I have been working with advanced computer systems, hardware, and software since the time of the 8080/Z80 and 6502.

I have been a part of teams, and have worked with people who - literaly - “wrote the book” on many of the technologies you use today. These are people who have Tim O’Reilly’s private cell number and knew Bill Gates and Steve Jobs on a “come over for a beer” first-name basis, and they never arrived at a reasonably well working and generalized solution to the problem.

Others, equally talented, have also failed.

If YOU have discovered a simple, easily implemented and managed system that eliminates all of the IPv6 magic, smoke-and-mirrors, and sleight of hand, please publish it and provide the rest of the world access to your GitHub repo.

And then book tickets to Stockholm for the next Nobel Prize Awards.

1 Like