That’s good to hear. Even I am trying to get it to work but the whole decoding stack have to be patched.
Good Luck.
3 Likes
If I do ‘Install SD card image to eMMC’
and later want to flash Android back to emmc, how will that work?
Rockchip AndroidTool, FactoryTool won’t detect the device anymore since there is no Android u-boot.
If pressing the MaskROM Mode buttons on the sides of the Khadas Edge-V, will it Enter MaskROM Mode and allow Android to be flashed back from a PC?
If booting Ubuntu from a micro-sd card, is there a way inside Ubuntu to
flash Android(with some script) or a different Ubuntu(perhaps with etcher) to the emmc too perhaps?
did you try erasing the eMMC ?
Yes, you can. MASKROM mode works in any situatuion to falsh the image to eMMC.
I am not sure what you want to do?
Thanks
So now I won’t be scared to flash images anymore.
For 2)
Basically I just want to avoid unnecessary steps to connect device to a PC and then also enter MaskROM Mode to change the OS on the EMMC.
Wanted to know if there is a way to do everything from Ubuntu/Linux running from a micro-sd card to access the whole emmc storage, erase it and flash Linux/Android to it.
But I think KRescue in development will solve this problem.
Something else
If I want to keep Android on emmc and only test Ubuntu/Linux OS’s like Manjaro from a micro-sd card
then the Linux OS’s can’t boot from the micro-sd card since the Android u-boot on the emmc(which is a lot older) loads first and not the one on the micro-sd card, so you just get a ‘no signal’ on your tv.
Is there some way Khadas, you or @goenjoy can
upgrade the u-boot used by Android so it can boot both Android(4.4, 4.19 kernels) from emmc and Linux(5.7 kernel) from micro-sd cards?
At the moment only a Linux OS using a 4.4 kernel can boot from a micro-sd card if Android is on emmc. If you want to eg. use Manjaro(5.7 kernel) on a micro-sd card, you need to keep the emmc erased. But a strange thing is, LibreELEC that also uses a 5.7 kernel can boot correctly from a micro-sd card if Android is on emmc but not Manjaro here - Manjaro ARM Linux - Preview1 - Edge-V Pro
Maybe it’s a Manjaro problem?
I haven’t tried this Edge Ubuntu 20.04 v0.9.1 image yet, but will later today.
Not sure if it will have the same problem but maybe since it uses 2020.04 u-boot which Android on emmc doesn’t.
just to put it here for troubleshooting, i face this error
with image
VIM3_Ubuntu-gnome-focal_Linux-4.9_arm64_SD-USB_V0.9-20200530
EDIT: solved the issue by updating Balena Etcher to latest version which is by the time of writing is v1.5.98
Hello, I have khadas vim 3, when I recorded the image on emmc and first loaded, everything was fine and I was very excited about this new firmware. But when I turned off khadas and the next time I turned it on, the screen remained black all the time, and the white led on the Board was blinking. Then I installed Android on emmc and ubuntu on cd card, and ubuntu started loading on the screen, then I waited 5 minutes and the screen went black again. Now I don’t know what to do. But still I like this new firmware, I hope that in the future you will be able to fix it, thank you for your work, it is very cool.
1 Like
@AKBAAR Balena etcher faces a lot of problems with that ability to write compressed image, better uncompress it and then flash the image, that feature was a Beta release and has been now removed,
Though it was a nice feature allowing you to save space, it was very faulty
yes krescue for edge will be ready soon Krescue for EDGE (Rockchip rk3399) - at this moment we are in testing stage
1 Like
yes solved the issue with updating Balena Etcher to latest version v1.5.98 .
1 Like
good now that it works, but i recommend win32 disk imager, it is smaller, you can take disk backup and save the image from the sd card or usb
1 Like
so now , I tried to install TensorFlow && keras
i reached the walkthrough here
i get this error , am guessing its not implemented yet in the new OS
It is not a supported package, try a different platform.
Its better to build tensorflow… Why use tensorflow it only uses CPU not even GPU… things will be too slow…
Yes, Training model on SBCs Sucks, but I don’t want to carry a GTX970 all day with me either
I hear you @AKBAAR, I totally agree, not everyone has powerful computers that can train these models on the fly,
1 Like
that’s why cloud computing exists 
But you can’t have a nice broadband connection anywhere you are either, everything has its limitations,
Only God’s computer could achieve such crazy tasks 
lol u can access cloud VM with 64Kbps connection