Armbian for Amlogic S912


chocho
 Share

2 2

Recommended Posts

A question to the owners S912. As you work images Armbian in the off mode ? If you select Shutdown (poweroff), the console shuts down ? WiFi and wired network work ?

On my Beelink GT1, the ethernet was working for sure. (I think the wifi too, I need to retest the other points.) There seems to be different kind of version of the chipset (depending on the SN.) 

Link to post
Share on other sites

Armbian is a community driven open source project. Do you like to contribute your code?

I have bellink gt 1 and yes it works, ethernet working but wifi not working last time I try ( 2-3 weeks ago)

 

On my Beelink GT1, the ethernet was working for sure. (I think the wifi too, I need to retest the other points.) There seems to be different kind of version of the chipset (depending on the SN.) 

There are at least 3 different revisions of Beelink GT1, and at least 2 of them require different Wi-Fi drivers.

Link to post
Share on other sites

I am running your Armbain images on my Amlogic S905 and 905X boxes, and they work great, especially for Plex and Emby servers.  Of course, they do not have enough processing power to transcode on the fly.  My question, would this S912 have more transcoding power than my S905 boxes??

Link to post
Share on other sites

I am running your Armbain images on my Amlogic S905 and 905X boxes, and they work great, especially for Plex and Emby servers.  Of course, they do not have enough processing power to transcode on the fly.  My question, would this S912 have more transcoding power than my S905 boxes??

The ARM cores are the same Cortex A53, just S912 has more of them. So if software and codec supports parallel transcoding and the CPU doesn't overheat, it may work, but needs to be tested.

Link to post
Share on other sites

The ARM cores are the same Cortex A53, just S912 has more of them. So if software and codec supports parallel transcoding and the CPU doesn't overheat, it may work, but needs to be tested.

Thanks for your feed back.  I see that you have one of the Amlogic S912 boxes, can you test Emby server and see how it works on your box?

 

thanks

Link to post
Share on other sites

Thanks for your feed back.  I see that you have one of the Amlogic S912 boxes, can you test Emby server and see how it works on your box?

I got it only 2 days ago and didn't make any tests or images for it yet (other than trying one of @balbes150 images). I'll try to bring its support to the main Armbian build script, but currently this is at low priority, so I'll leave performance testing for the @balbes150 for now since he has a S912 box too AFAIK.

Link to post
Share on other sites

The ARM cores are the same Cortex A53, just S912 has more of them. So if software and codec supports parallel transcoding and the CPU doesn't overheat, it may work, but needs to be tested.

 

S912 is a big.little setup so unless you can confirm the kernel supports heterogeneous multi-processing, it definitely won't make any difference.

 

In other words the kernel's .config must contain:

CONFIG_SCHED_HMP=y

but even then we still have no idea what the firmware is capable of.

Link to post
Share on other sites

S912 is a big.little setup so unless you can confirm the kernel supports heterogeneous multi-processing, it definitely won't make any difference.

 

In other words the kernel's .config must contain:

CONFIG_SCHED_HMP=y

but even then we still have no idea what the firmware is capable of.

 

https://github.com/150balbes/lib/blob/master/config/kernel/linux-amlogics905x-default.config#L292

Link to post
Share on other sites

 

OK then, a quick test to see what DVFS thinks about this idea:

echo performance | sudo tee /sys/devices/system/cpu/cpu0/cpufreq/scaling_governor
wget http://www.phoronix-test-suite.com/benchmark-files/c-ray-1.1.tar.gz
tar xf c-ray-1.1.tar.gz
cd c-ray-1.1/
gcc -O3 -mcpu=cortex-a53 -o c-ray-mt c-ray-mt.c -lm -lpthread && ./c-ray-mt -t 32 -s 320x240 -r 8 -i sphfract -o output.ppm 
cd

Anything better than `7 seconds (7437 milliseconds)` will have meant more than 4 cores can be utilised at once. (gcc 5.4 required for an apples to apples comparison)

 

In case it does work, something like e.g. `-mtune=cortex-a57.cortex-a53` can be added to the gcc command line to enable big.little optimisations.

Link to post
Share on other sites

@balbes150,

 

Can I try to generate an Armbian image using your code from?

https://github.com/150balbes/lib

 

Is it the "official" porting to Armbian building system?

 

What is the stage today?

 

1. Yes, You can use this git to build (it is in the public domain).

2. No, this is not the official version. The process of inclusion in the official GIT requires considerable time.

 

 

OK then, a quick test to see what DVFS thinks about this idea:

echo performance | sudo tee /sys/devices/system/cpu/cpu0/cpufreq/scaling_governor
wget http://www.phoronix-test-suite.com/benchmark-files/c-ray-1.1.tar.gz
tar xf c-ray-1.1.tar.gz
cd c-ray-1.1/
gcc -O3 -mcpu=cortex-a53 -o c-ray-mt c-ray-mt.c -lm -lpthread && ./c-ray-mt -t 32 -s 320x240 -r 8 -i sphfract -o output.ppm 
cd

Anything better than `7 seconds (7437 milliseconds)` will have meant more than 4 cores can be utilised at once. (gcc 5.4 required for an apples to apples comparison)

 

In case it does work, something like e.g. `-mtune=cortex-a57.cortex-a53` can be added to the gcc command line to enable big.little optimisations.

 

In the next few days I'll try to run this test (now I have disabled s912 to upgrade the cooling system).

Link to post
Share on other sites

The results of running Your command sequence on s905x and s912 with a new look.

 

https://yadi.sk/d/K4UJK0UD3CzNbL

 

By the way, to simplify testing, I created a simple script aml_cpu.sh (of Your data), which can be downloaded here. I decided at the time to add it to the images that would be any non-trained user is able to run on any TV box that would have the information with different devices (the script is located in /root directory in the new images).

 

aml_cpu.sh

 

https://yadi.sk/d/jn_hH4GB3CzKtU

Link to post
Share on other sites

IMHO the Capacity of all processor Amlogic is much more than is possible in the manufactured home TV boxes. The reason is that for good performance you need a good (active) cooling system (for long-term work in maximum performance mode). The vast majority of TV boxes use a more stripped-down and primitive cooling system. I do not think the cooling system, those funny metal plates that are installed in sealed plastic boxes in many TV boxes. Perhaps this is due to the fact that the regular Android interface is very simplified and accordingly, do not trouble yourself a good cooling system. Perhaps for this reason, Amlogic forced to release the dtb and the source code with the restriction that-be to ensure the operability of its products in such a non-threatening environment.

Link to post
Share on other sites

Thanks @balbes150! The results (7992 vs 4781) do indicate perfect scaling. In other words, as long as there's no thermal throttling, S912 really seems to offer 66% more computing power. Not bad for a $50 box :)

 

It would be nice to be tested on different devices, users, to have more data for evaluation.

Link to post
Share on other sites

 

On my Beelink GT1 with SN starting with A912 

c-ray-mt v1.1
Rendering took: 4 seconds (4750 milliseconds)

Could you try running the renderer for a few minutes? e.g:

./c-ray-mt -t 32 -s 1900x1400 -r 8 -i sphfract -o output.ppm

It takes 254 seconds on my MiniMX III (S905) so unless S912 starts overheating it should complete the same task in about 152s.

Link to post
Share on other sites

Look at my comment #133 - the command is meant to run inside the c-ray-1.1 directory after compilation. If you ran the benchmark at least once, the c-ray-mt binary should be found there. 

Thanks, here are my results.

 To run a command as administrator (user "root"), use "sudo <command>".

See "man sudo_root" for details.
 
ron@amlogic-s905x:~$ echo performance | sudo tee /sys/devices/system/cpu/cpu0/cpufreq/scaling_governor
[sudo] password for ron: 
performance
Resolving www.phoronix-test-suite.com (www.phoronix-test-suite.com)... 23.111.154.110
Connecting to www.phoronix-test-suite.com (www.phoronix-test-suite.com)|23.111.154.110|:80... connected.
HTTP request sent, awaiting response... 200 OK
Length: 231074 (226K) [application/x-gzip]
Saving to: ‘c-ray-1.1.tar.gz’
 
c-ray-1.1.tar.gz    100%[===================>] 225.66K   302KB/s    in 0.7s    
 
2017-02-05 18:15:44 (302 KB/s) - ‘c-ray-1.1.tar.gz’ saved [231074/231074]
 
ron@amlogic-s905x:~$ tar xf c-ray-1.1.tar.gz
ron@amlogic-s905x:~$ cd c-ray-1.1/
ron@amlogic-s905x:~/c-ray-1.1$ gcc -O3 -mcpu=cortex-a53 -o c-ray-mt c-ray-mt.c -lm -lpthread && ./c-ray-mt -t 32 -s 320x240 -r 8 -i sphfract -o output.ppm
c-ray-mt v1.1
Rendering took: 8 seconds (8244 milliseconds)
ron@amlogic-s905x:~/c-ray-1.1$ cd
ron@amlogic-s905x:~$ ^C
ron@amlogic-s905x:~$ 
Link to post
Share on other sites

The results of running Your command sequence on s905x and s912 with a new look.

 

https://yadi.sk/d/K4UJK0UD3CzNbL

 

By the way, to simplify testing, I created a simple script aml_cpu.sh (of Your data), which can be downloaded here. I decided at the time to add it to the images that would be any non-trained user is able to run on any TV box that would have the information with different devices (the script is located in /root directory in the new images).

 

aml_cpu.sh

 

https://yadi.sk/d/jn_hH4GB3CzKtU

c-ray-mt v1.1
Rendering took: 7 seconds (7232 milliseconds)

on Mini MX-G (s905) running gentoo

Link to post
Share on other sites

could you please name me the files

The sd card to run armbian on s912,

everything works perfectly on S905

but not on s912 (GT1) after replacing

appropriate files. Thank advance

 

Scusate per il mio google english

 

The launch is the same as S905. To capture the image. To activate it-boot. Try to start the system without manually copying the dtb file (in the latest images this feature is fixed and should work).

Link to post
Share on other sites

Could you try running the renderer for a few minutes? e.g:

./c-ray-mt -t 32 -s 1900x1400 -r 8 -i sphfract -o output.ppm

It takes 254 seconds on my MiniMX III (S905) so unless S912 starts overheating it should complete the same task in about 152s.

It was close ! (About 40°c on start and 50 after).

vincs@amlogic-s905x:~/Downloads/c-ray-1.1$ ./c-ray-mt -t 32 -s 1900x1400 -r 8 -i sphfract -o output.ppm
c-ray-mt v1.1
Rendering took: 164 seconds (164293 milliseconds)

Link to post
Share on other sites

Rendering took: 164 seconds (164293 milliseconds)

 

Interesting! Looks not too hot @50°c, so provided you were on performance governor, this could either mean the 4-core baseline was slightly worse on that box or the benchmark doesn't scale at these settings. 

Link to post
Share on other sites

Guest
This topic is now closed to further replies.
 Share

2 2