RK3328 Kernel

[tkaiser]

2 minutes ago, mdel said:

any reason why you're not interested in their socs ?

 

Don't fit my use cases

 

Regarding Gigabit Ethernet adjusting stuff like TX/RX delay settings is mandatory if you want to get any decent performance numbers. Having these settings defined wrong simply destroys your performance for no reason, this can be considered 'fixing stuff on the hardware layer'. Then the next layer is involved and that's where we as Armbian team try to care a lot: taking care of CPU affinity and some tweaks you might double network performance for free. That's why I consider this the most important part of supporting a new board (same with thermal/throttling settings for example).

 

For example with ODROID-C2 it's two tweaks to get full performance (if you optimize for throughput and not latency!), with NanoPi M3 (should apply to M2, T2 and T3 too) it's one simple tweak.

 

Another example: in case of boards where Gigabit Ethernet throughput is limited for whatever reasons (see ODROID-C1 or all i.MX6 boards) we document this at least to give users the ability to choose an appropriate device. In my opinion this is the most important part when adding a new board to Armbian: analyzing how we can squeeze out the max, documenting pros/cons and weigh them. Fortunately this is totally unrelated to Armbian itself (or the specific distro variant running later be it Debian X or Ubuntu Y) but 'good settings' can be used on any OS image later. But funnily some 'consumers' of our build system take the output, cripple it and throw good settings away for no reason. But hey... who cares...

[manuti]

35 minutes ago, tkaiser said:

Nope, not interested in RK3288

 

 

OK thanks for answer, I'm only checking possible cheap equivalent to th Asus TinkerBoard.

I'm more or less happy with my armbian in Beelink X2 but always is nice to have 2GB of RAM and better internal memory for the OS.

There are also a lot of RK3229 Android TV boxes but without proper Linux and similar specs than Beelink X2 I think is a waste of time.

 

So just keep searching!

 

[tkaiser]

16 minutes ago, manuti said:

thanks for answer

 

BTW: my use cases don't matter for everyone else. While I'm not that excited about RK3288 devices others clearly are. @jernejfor example added mainline u-boot support for MiQi (RK3288 in general) a few weeks ago if I understood correctly: https://github.com/jernejsk/u-boot/branches

 

Then you should check bitkistl.com/@Peba for stuff like this (and keep in mind that everything that's possible with MiQi will be possible with Tinkerboard or any other RK3288 device with same amount of DRAM too). Or @wtarreau 's cluster adventures in MiQi forum. @Myy's mainline kernel branches. All great stuff... just not for me (again: that doesn't matter, now that initial RK3288 support is in Armbian and few more devs start to look into RK3288 I would believe situation will greatly improve soon and that the same happens just as with S905/S905X/S912: supporting a new RK3288 device is more or less device tree stuff)

[jernej]

3 hours ago, mdel said:

i suppose it will greatly depend on the quality of the kernel source provided by the  MiQi or Asus projects..

No need to depend on board maker support, because official Rockchip version should work pretty well and basically all drivers are mainlined anyway.

 

4 hours ago, manuti said:

Is possible? Is easy? Or interesting?

Yes, it is possible, but I'm not sure how easy or interesting it is. If it is based on reference design, then yes, it is matter of hours for some who knows DT stuff. Another issue is how hard to find or extract is manufacturer dtb file. I doubt that they provide schematic. It is also the question if it is worth the effort.

 

I guess it should be similar enough to some existing board which has support already manilined. The biggest difference between the boards I saw is which power management chip is used.

[manuti]

Thanks @jernej for the response. 

My knowledge is far away to messing with Device Tree.

[TonyMac32]

Yes, I've been messing with the device tree.  It hasn't been going well...  (Learning by doing has disadvantages)

[dememan]

Hello!

 

Thanks for your engagement to get this board up and running

I´m new at this but I´m learning fast
I´m not sure if this is right topic for my  question but I give it a shot anyway.

I download the latest image: Armbian_5.27.170330_Tinkerboard_Ubuntu_xenial_default_4.4.57_desktop

and install  it on a SD-card.

Change password and got it up and running.

I Installed Kodi with the right command in the terminal window, but when I try to open Kodi in The Armbian desktop it crash all the time.

How do I manage to fix it?

Thanks in advange.

[Igor]

14 minutes ago, dememan said:

I Installed Kodi with the right command in the terminal window, but when I try to open Kodi in The Armbian desktop it crash all the time.

How do I manage to fix it?

- those images are preview without official support,

- none of core team members has this board yet,

- video acceleration features comes at the end and if we find time to fix them,

- KODI is not trivial to install on any platform and general operating system.

 

How to fix this particular problem? Don't know. At this moment.

[dememan]

Thanks for your answer Igor!
Just wait and see than and hope for the best

I  still have my raspberry to use meanwhile

[TonyMac32]

On the global site ASUS has released a beta of "Tinker OS" 1.6  https://www.asus.com/Single-Board-Computer/Tinker-Board/HelpDesk_Download/

[mdel]

On 3/27/2017 at 1:39 PM, manuti said:

And what do you think about porting to a cheep Android TVBox like http://www.enybox.com/En/product/html/?200.html ?

 

I received my "Q8" box from gearbest, it's the exact same board without front panel board.

i need to find a way to root that android 4.4.2 so i can get access to uboot env and try sd card or tftp boot.

 

i'll have to study the miqi boot process and see if it has some similar loading of external boot script as aml_autoscript on amlogic.

 

I'm more worried not seeing a serial header on the board, i have one weird candidate (3 pins labelled "G - R").

On that matter, i wanted to ask if the otg micro usb port present is used in any way to debug those rockchip boards ? does it provide uboot serial access and so on ?

 

 

On 3/27/2017 at 2:43 PM, tkaiser said:

Regarding GbE I doubt further tests are needed (since the GbE MAC is inside the SoC and combined everywhere with an external RTL8211E PHY). If you destroy settings or at least don't care about those then you end up with something like this while everyone else reports 930-940 Mbits/sec

 

yup RTL8211E is on the bottom side surprisingly.

My initial iperf tests from android are quite promising, i got about 850-900Mbps in both directions.

 

[jeanrhum]

1 hour ago, mdel said:

I'm more worried not seeing a serial header on the board, i have one weird candidate (3 pins labelled "G - R").

I think you should have a look at this link: http://freaktab.com/forum/tv-player-support/rk3288-devices/eny-3288-devices/25044-serial-port-on-xt-q8l-v10

and this one: https://bitbucket.org/DengueTim/q8-rk3288-kernel/wiki/Home

 

I hesitate to buy this board for playing with linux and opencl since Mali4xx are not able to do that. Please, give us some feedback if you achieve to run linux 4.x on it.

[jernej]

3 hours ago, jeanrhum said:

I hesitate to buy this board for playing with linux and opencl since Mali4xx are not able to do that.

 

RK3288 has Mali-T760 MP4 which supports OpenCL...

[jeanrhum]

Yes, you're right. I answered too clickly. I was thinking of most amlogic or allwinner socs integrated in sbcs that are Mali4xx.

[Shimon]

7 hours ago, mdel said:

i'll have to study the miqi boot process

Meanwhile, you could try running a benchmark or two in termux, provided Android 4.4 is not a problem.

 

edit:

No, it's not supported. TerminalIDE or ZShaolin should be usable though.

[TonyMac32]

Getting back to the Tinker Board itself.. 

 

I swapped in the device tree from the miniarm branch at rockchip-linux, it compiles and boots properly, however I have had no luck getting the Bluetooth or wifi devices enumerated, I need to spend some more time with it.  More troubling, and the reason I have so little progress, is the adjustments made to the HDMI portions of the DT have rendered video output almost impossible to maintain (anytime the system puts the monitor output to sleep it refuses to wake up, sometimes a hotplug of the monitor will rectify it).  The mainline builds fail if I try to patch anything at the device driver level, I'll dig into that later.

[manuti]

1 minute ago, TonyMac32 said:

I need to spend some more time

Really impressive work for a "one man army". Much appreciate your effort.

[TonyMac32]

1 minute ago, manuti said:

Really impressive work for a "one man army". Much appreciate your effort.

Hold your praise until I provide something besides a functional U-boot flash.    Like I've said before, I'm an electrical engineer by profession, I work with embedded bare-metal software, this tangled web of device trees, makefiles, build instructions, and such is a little outside my core competencies...

[mdel]

36 minutes ago, TonyMac32 said:

Getting back to the Tinker Board itself.. 

 

 

yup sorry i won't hijack you thread anymore, i'll post to a new one. 

[mcerveny]

Hello, I tested armbian@tinkerboard but not happy. http://tinkerboarding.co.uk/forum/showthread.php?tid=40&pid=302#pid302

[TonyMac32]

4 minutes ago, mcerveny said:

Hello, I tested armbian@tinkerboard but not happy. http://tinkerboarding.co.uk/forum/showthread.php?tid=40&pid=302#pid302

As you'll see in the thread, this board doesn't have official support yet, the core devs don't even have a board available to them.  As far as the device tree is concerned, you are correct it leaves out quite a few things, I'm looking into that but have little experience I'm afraid.  I will look into/test out what I can of your recommendations, I was of course looking for feedback along these lines.  First and formost is to get the device tree sorted, of course I haven't yet seen two that look similar or seemingly point at all of the hardware... 

[nightseas]

On 2017/3/16 at 2:01 AM, tkaiser said:

And another RK3288 device, maybe doing a few things better (eg. barrel plug for DC-IN and not crappy Micro USB, no internal USB hub, mounting holes for large heatsink):

 

 

 

 

http://linksprite.com/wiki/index.php5?title=PcDuino9#Hardware_specifications 

 

Well, my last attempt to deal with Linksprite's software was shocking

And it has 4GB dual-channel LPDDR3 by default, a small cortex-M4 MCU for board management or real time processing.

The real purpose of this board is not for SBC but micro blade server cluster.

that's why all connectors are placed to one side, as front panel in the future.

I was also planning a real open cluster project to DIY blade system with managed switch blade ( the SOM on clearfog could be a good start ), backplane power delivery and cooling control.

Just need more TIME. Ouch...

[TonyMac32]

Small update:  I simply can't get any device tree to compile other than the one already patched in.  The dts fails to compile, then the kernel compile fails.  I'm going to start looking at the bindings and see if everything is in there that's used on the newer device trees, although to be honest I don't see anything that should be causing the pain (spoken like a true amateur, I'm sure someone will find in it 10 seconds once they start looking.)

[Myy]

9 hours ago, TonyMac32 said:

The dts fails to compile, then the kernel compile fails

Could you paste the content of the failing compilation lines ?

Also, you might want to create a quick git repo containing your DTS files.

 

While I could try compiling a DTB file for Tinkerboards during the next update, I'll not be able to test such DTB as I do not own any Tinkerboard

[Nick Allen]

Hi, 

Just a quick vote of thanks for the people involved in creating the OS for this board. Since buying the board a week or so I have tried several different OSś, but this is definitely the best available. Iḿ using it as a desktop with most of my favourite apps working.

I appreciate there is still a long way to go, but the work is appreciated.

 

Thanks again

 

Nick

[TonyMac32]

Well, I decided to quit being lazy and starting hacking away, I have a better understanding of the structure now and have begun updating the device tree, testing it out now.  I found the issue I was having before, my brain was stuck in C territory and I was assuming declarations flowed from the top down, and didn't stop to think that the various subsections (like, oh , pinctrl) would be referenced elsewhere within the same file.  I transplanted the LED information from a Feb patch on Kernel.org, it had aliases for the pin numbers that were breaking everything.   I'm testing and tweaking, forgot to add the appropriate modules to the board config after I got the dtb :-/

 

[Edit] is there any reason the board is equipped with  the 8723bs but every DTS has ap6212  listed?

 

 

[TonyMac32]

OK, so I have the LED's, 4k, and an assortment of other things working properly, I have, in theory, everything in the DTS to make it work.  Of course at this point I truly am lost, the wlan type in the DTS's that other working distros use says AP6212, but the board has 8723bs.  I also am not enough of a Linux guru to patch driver things.

 

Some notes:   I am trying to track down the most recent rk3288.dtsi, I had to add A LOT to it to cover the additions for hevc and gpu handling.  (have not tested any of that yet).  I plan on starting from the latest "official" one to make sure there's as little hackery as possible.  Unless patching in the one from rockchip-linux is acceptable, then I'll save myself the time.

 

Also, preliminarily working on getting the DTS to compile with the dev kernel.  will keep you posted.

[tkaiser]

On 13.4.2017 at 5:07 AM, TonyMac32 said:

is there any reason the board is equipped with  the 8723bs but every DTS has ap6212  listed?

The RTL8723 uses the 'Ampak footprint' but costs less. I think back at the time the 'ASUS Tinker board' was the 'Rockchip Miniarm' Rockchip engineers decided to go the Ampak route since there are modules available even with 802.11ac (dual-band, dual-antenna so not that AP6212 crap they used on first PCB revisions). Then ASUS decided to replace AP6212 with even cheaper crap to make a little bit more money and now we have RTL8723 here (Wi-Fi 'performance' comparable to AP6212 so it really doesn't matter that much for end users)

[lafalken]

Hi!

I have been following this tree for some time now. Bougt this board some weeks ago on discount here in Sweden but today I got some time to test this device. I took the nighly build 20170414 and copied it  to a 32 Gb card, using a Samsung mobile charger and network cable to log on to it by ssh. It worked out well without any problems but I would prefer to have an image without desktop, because I want to use it for a joomla server. 

 

My coding experience is of no use here but will be happy to test those images that comes up here. I have not tried any from Asus site. Armbian is my first choice if possible

 

Cheers 

 

[TonyMac32]

Push request in to update rk3288.dtsi and rk3288-miniarm.dts.  There are no kernel driver updates in that patch, just the device tree updates.  All info was pulled from Rockchip-linux, which I sincerely hope is the leading source of info for the chip (maintained by Rockchip).

 

The rockchip-linux kernel has a huge patch involving the wlan module, I'm wondering what on earth they did, or maybe it was just a driver update?

 

Obviously testing of the Miqi might need done, but I doubt there will be any issues.  A question I do have, does the miqi not have hardware decoding/vpu functionality enabled?