Moode Forum

I love to tinker, especially with new hardware or at least with hardware that is new to me.

So I was reading yet another one of those click-bait pieces about "best alternatives to the Raspberry Pi" that continually pop up in my news feeder. As usual, the article was horribly shallow and looked mostly like a paste-up of vendor announcements. For some reason, though, its mention of the Libre Computer Le Potato (AML-S905X-CC) board selling for just 45USD caught my eye. The specs of this old-ish SBC weren't great but the price was right and I was intrigued that this is an ARM board yet its firmware uses grub and uefi to boot an OS the way x86 boards do.

Sunday morning I ordered one along with a Libre-branded power supply from Amazon and the package arrived this afternoon.

15 minutes later I had it running moOde 8.1.2 and playing through my Creative BT-W2 USB adapter to my Sennheiser Bluetooth headphones. Woof.

How was that possible, you may ask, especially since the boot process in the Le Potato firmware works so completely differently from the way the Pi loader works. 

Well, their software devs have created a script---Libre Computer Raspbian Portability, aka LRP---which one executes on a Pi running Raspbian/Raspberry Pi OS to populate new subdirectories in the running OS's /boot partition to support the Le Potato boot process. It also installs a more recent kernel.

It turns out that the script works just fine when run in moOde 8.1.2. The result is a uSD card which will boot into moOde on either a Pi or a Le Potato. In my case, the kernel is now 5.19.6.

There's also a program---Libre Computer Wiring Tool---which translates GPIOs and dtoverlays but I haven't dabbled with it yet.

Many things don't work right, at least not yet, some of them superficial (moOde can't identify the board/CPU type, for example) and some not (can't use a DAC hat, for example; for another, smb/cifs filesystems won't mount although nfs filesystems do). 

Still, it's fun to see how far I can push this. It was far easier to get to the point I can listen to music than when I tried to build moOde 4.0 on foreign boards such as a CuBox-i 4P years ago. As a friend was fond of saying, it keeps me out of bars.

Regards,
Kent

(09-06-2022, 02:11 AM)TheOldPresbyope Wrote: [ -> ]I love to tinker, especially with new hardware or at least with hardware that is new to me.

So I was reading yet another one of those click-bait pieces about "best alternatives to the Raspberry Pi" that continually pop up in my news feeder. As usual, the article was horribly shallow and looked mostly like a paste-up of vendor announcements. For some reason, though, its mention of the Libre Computer Le Potato (AML-S905X-CC) board selling for just 45USD caught my eye. The specs of this old-ish SBC weren't great but the price was right and I was intrigued that this is an ARM board yet its firmware uses grub and uefi to boot an OS the way x86 boards do.

Sunday morning I ordered one along with a Libre-branded power supply from Amazon and the package arrived this afternoon.

15 minutes later I had it running moOde 8.1.2 and playing through my Creative BT-W2 USB adapter to my Sennheiser Bluetooth headphones. Woof.

How was that possible, you may ask, especially since the boot process in the Le Potato firmware works so completely differently from the way the Pi loader works. 

Well, their software devs have created a script---Libre Computer Raspbian Portability, aka LRP---which one executes on a Pi running Raspbian/Raspberry Pi OS to populate new subdirectories in the running OS's /boot partition to support the Le Potato boot process. It also installs a more recent kernel.

It turns out that the script works just fine when run in moOde 8.1.2. The result is a uSD card which will boot into moOde on either a Pi or a Le Potato. In my case, the kernel is now 5.19.6.

There's also a program---Libre Computer Wiring Tool---which translates GPIOs and dtoverlays but I haven't dabbled with it yet.

Many things don't work right, at least not yet, some of them superficial (moOde can't identify the board/CPU type, for example) and some not (can't use a DAC hat, for example; for another, smb/cifs filesystems won't mount although nfs filesystems do). 

Still, it's fun to see how far I can push this. It was far easier to get to the point I can listen to music than when I tried to build moOde 4.0 on foreign boards such as a CuBox-i 4P years ago. As a friend was fond of saying, it keeps me out of bars.

Regards,
Kent

Thanks for this, looks like a viable alternative to the now unobtainium RPi 3B+, as they indicate the form factor matches to the extent you can use a case/enclosure meant for the 3B+ too.

So the initial build of the uSD card is done in a RPi using the LRP script, and then once completed that card boots the Le Potato board?

The Le Potato has the physical form factor of the RPi3B/3B+ but it has no onboard WiFi/BT transceiver which may or may not be a deal breaker for some.

Yes, the result of LRP is a uSD card which will boot into moOde on either a Le Potato or an RPi (I've tested that the card I built boots on Le Potato, RPi3A, RPi3B, RPi4B).

Yes, an RPi running moOde is necessary to execute the LRP script as written. It would seem to me a comparable script could be written for another host---my Linux laptop, say---but I haven't worked out just what LRP does.

In crude benchmarking using linpack.py [1] running simultaneously with moOde I found the Le Potato to be comparable to the RPi3B, a tad slower than an RPi3A, and substantially slower than an RPi4B.

It seems the reason I can't mount an SMB share is that the kernel which was installed by LRP wasn't built with cifs support. I haven't traced yet where this kernel comes from or if this kernel is necessary to support the Amlogic S905X SoC on this board.

Remember, I said I like to tinker. Translation: this is an experiment!

Regards,
Kent

[1] Yeah, Linpack was developed for supercomputers back in the 70s, but I was an employee at Argonne National Laboratory back then who wrote a lot of scientific Fortran code to support my experiments, and I still have a fondness for it. It's principal developer, Jack Dongarra, has been awarded pretty much every accolade known in the computer science biz including the Turing Award in 2021. See linpack for easy to follow instructions for running it in Python.

Regards,
Kent

The potato boards must be from whatever inventory they have left since the chip shortage affects everyone that needs trailing edge chips.

(09-06-2022, 04:41 PM)Tim Curtis Wrote: [ -> ]The potato boards must be from whatever inventory they have left since the chip shortage affects everyone that needs trailing edge chips.

No argument here. Just like the lack of widespread software projects based on this board, I interpret the availability of the board itself as sign that it has never been in high demand. Who knows how many are lying on distributors' shelves?

Eben Upton claims Raspberry Pi factories are cranking out approximately 500K SBC and Compute Modules a month and that this is more or less enough to satisfy demand from commercial and industrial customers. As anyone else trying to buy one knows, few units are leaking into retail channels. I'd love to see a tally of the backorders that are accumulating at their authorized resellers. It seem like it must be huge.

Meanwhile, paraphrasing Descartes, I tinker, therefore I am 

Regards,
Kent

lol, even better than the famous maxim!

From what Ive read the shortage in trailing edge chips like those used in SBC's will last into 2023...

For anyone in the U.S. that lives near enough to a Micro Center retail location, I got lucky twice in August, first with an RPi 4B (4GB), and then again towards the end of the month with the 8GB version of that board.

Micro Center will not ship these, nor even allow one to reserve it online and pickup in store, you simply have to see some available stock at a retail location within your driving distance, and then head right down to buy it in person. Sub-optimal, but it worked out twice for me in August just by checking their retail stock a few times.

https://www.microcenter.com/category/429...spberry-pi

(09-06-2022, 02:11 AM)TheOldPresbyope Wrote: [ -> ]15 minutes later I had it running moOde 8.1.2 and playing through my Creative BT-W2 USB adapter to my Sennheiser Bluetooth headphones. Woof.

I recently stumbled across the Moode project and this forum.  I bought a Le Potato because of the extreme shortage of RPis.  Very encouraging to hear that this can work for my board also.  Seems like maybe since the LRP script makes an image support both platforms, the image that gets distributed could potentially include this support?

I looked through the LRP script a bit and I agree, it seems like it could be made to be standalone.  I'm tempted to try to do that, since it would be of use to a lot of people.  Curious though, would it be difficult to put your ISO of the Le Potato version out there, for those of us who don't have an RPi to run on?

@michaelfour

Actually, after that first spurt of testing, I've been playing with the Le Potato board in other applications.

I think it very unlikely Tim would want to distribute the modified image.

• It would have been made portable only to the old Le Potato board and not any of the other, newer Libre Computer SBCs.
  
• It would not have been modified to accommodate the differences between the two environments. For example, a number of nice moOde configuration features depend on RPiOS-specific features.
  
• The whole DAC HAT and associated drivers infrastructure would have to be explored. Perhaps using their "wiring tool" is sufficient; perhaps not. Perhaps using that tool interferes with the existing RPiOS functionality; perhaps not.
  
• There likely are yet undiscovered issues similar to my finding that the S905X kernel isn't configured for CIFS.
  
• ...
  

This translates to loads of fun for a determined (or crazy) tinkerer but not for a maintainer.

As for me hosting my modified image somewhere, I'll have to think about that a minute. I did create an account on the Libre Computer equivalent to our forum so I could ask if anyone has ported the LRP to, say, x86-based Debian/Ubuntu/whatever so one wouldn't need an RPi board to use it. We'll see.

Regards,
Kent

(10-22-2022, 03:49 PM)TheOldPresbyope Wrote: [ -> ]@michaelfour

Actually, after that first spurt of testing, I've been playing with the Le Potato board in other applications.

I think it very unlikely Tim would want to distribute the modified image.

• It would have been made portable only to the old Le Potato board and not any of the other, newer Libre Computer SBCs.
  
• It would not have been modified to accommodate the differences between the two environments. For example, a number of nice moOde configuration features depend on RPiOS-specific features.
  
• The whole DAC HAT and associated drivers infrastructure would have to be explored. Perhaps using their "wiring tool" is sufficient; perhaps not. Perhaps using that tool interferes with the existing RPiOS functionality; perhaps not.
  
• There likely are yet undiscovered issues similar to my finding that the S905X kernel isn't configured for CIFS.
  
• ...
  

This translates to loads of fun for a determined (or crazy) tinkerer but not for a maintainer.

As for me hosting my modified image somewhere, I'll have to think about that a minute. I did create an account on the Libre Computer equivalent to our forum so I could ask if anyone has ported the LRP to, say, x86-based Debian/Ubuntu/whatever so one wouldn't need an RPi board to use it. We'll see.

Regards,
Kent

It's enough of a challenge to maintain support for the Raspberry Pi family of boards but I think we do a pretty good job. moOde still runs on the Pi-1B circa 2012 :-)


If you are thinking about distributing an image just make sure the license terms for whatever software you add or modify permits redistribution.