Armbian Buster and Bionic with 5.7.y

[igorpec]

- u-boot 2020.04
- kernel 5.7.y
- lot's of patches
- can be rebuild from sources (https://github.com/armbian/build)

Download:
https://www.armbian.com/olimex-teres-a64/

Release history:
https://docs.armbian.com/Release_Changelog/

[igorpec]

v1.1
Bump

[jflesch]

I've tried suspending the laptop from Xfce, but it hangs while suspending (Teres rev. C). I have no output on the serial line when it happens. Is there any way I can help debug it ?

Other than that, it's a nice work ! It's starting to look good.

For those wondering, from what I've seen there are still some missing parts in v1.1, for instance:

- No Bluetooth
- No sound
- No brightness control
- No battery monitoring

However I've also noticed that the eMMC has much better performance with Armbian than with the Ubuntu from Olimex (45MB/s instead of just 22MB/s according to hdparm). So IMO it's clearly promising :-)

[khumarahn]

Faster eMMC is interesting! Slow disk access is the main bottleneck in my experience. What kind of benchmark does hdparm do? Is there a way to compare some sort of random access between the old and the new kernel?

[jflesch]

hdparm -tT runs a simple sequential read. One with cache enabled (measure more the memory/cpu than the disk) and one without cache.

I can run bonnie++ instead if you want more detailed statistics.

Here are bonnie++ results with Armbian Ubuntu Bionic:

Code Select Expand


root@teres:/mnt/emmc# bonnie++ -d test -u jflesch:jflesch
Using uid:1000, gid:1000.
Writing a byte at a time...done
Writing intelligently...done
Rewriting...done
Reading a byte at a time...done
Reading intelligently...done
start 'em...done...done...done...done...done...
Create files in sequential order...done.
Stat files in sequential order...done.
Delete files in sequential order...done.
Create files in random order...done.
Stat files in random order...done.
Delete files in random order...done.
Version  1.97       ------Sequential Output------ --Sequential Input- --Random-
Concurrency   1     -Per Chr- --Block-- -Rewrite- -Per Chr- --Block-- --Seeks--
Machine        Size K/sec %CP K/sec %CP K/sec %CP K/sec %CP K/sec %CP  /sec %CP
teres            4G   141  99 21640  10 16961   9   504  99 39124   9  1661  65
Latency             61414us    4026ms    7753ms   16692us    5852us    6755us
Version  1.97       ------Sequential Create------ --------Random Create--------
teres               -Create-- --Read--- -Delete-- -Create-- --Read--- -Delete--
              files  /sec %CP  /sec %CP  /sec %CP  /sec %CP  /sec %CP  /sec %CP
                 16 12299  76 +++++ +++  8479  41 12721  76 +++++ +++  7291  36
Latency             78842us    1809us     216ms     115ms     108us     153ms
1.97,1.97,teres,1,1535564920,4G,,141,99,21640,10,16961,9,504,99,39124,9,1661,65,16,,,,,12299,76,+++++,+++,8479,41,12721,76,+++++,+++,7291,36,61414us,4026ms,7753ms,16692us,5852us,6755us,78842us,1809us,216ms,115ms,108us,153ms


[khumarahn]

Can you run the same on the old kernel? I would do it but I've sent my teres back to olimex.

[jflesch]

Here are the results with an Ubuntu Bionic + Olimex kernel 3.10:

Code Select Expand


root@teres:/# bonnie++ -d test -u jflesch:jflesch
Using uid:1000, gid:1000.
Writing a byte at a time...done
Writing intelligently...done
Rewriting...done
Reading a byte at a time...done
Reading intelligently...done
start 'em...done...done...done...done...done...
Create files in sequential order...done.
Stat files in sequential order...done.
Delete files in sequential order...done.
Create files in random order...done.
Stat files in random order...done.
Delete files in random order...done.
Version  1.97       ------Sequential Output------ --Sequential Input- --Random-
Concurrency   1     -Per Chr- --Block-- -Rewrite- -Per Chr- --Block-- --Seeks--
Machine        Size K/sec %CP K/sec %CP K/sec %CP K/sec %CP K/sec %CP  /sec %CP
teres            4G   136  97 16336   7  8579   3   814  96 26147   4  1230  38
Latency             68172us    2639ms    6300ms   23869us     176ms     889ms
Version  1.97       ------Sequential Create------ --------Random Create--------
teres               -Create-- --Read--- -Delete-- -Create-- --Read--- -Delete--
              files  /sec %CP  /sec %CP  /sec %CP  /sec %CP  /sec %CP  /sec %CP
                 16  8890  47 +++++ +++  5116  24 10781  56 +++++ +++  4070  19
Latency               418ms    2697us     347ms     244ms     133us     351ms
1.97,1.97,teres,1,2797863130,4G,,136,97,16336,7,8579,3,814,96,26147,4,1230,38,16,,,,,8890,47,+++++,+++,5116,24,10781,56,+++++,+++,4070,19,68172us,2639ms,6300ms,23869us,176ms,889ms,418ms,2697us,347ms,244ms,133us,351ms

[khumarahn]

are the filesystems the same?

[jflesch]

Both are ext4.

[igorpec]

A few numbers which are also not very representative. 4.18.y does not have DVFS yet so CPU stays clocked at 800Mhz.

Armbian: http://ix.io/1n1K
Stock: http://ix.io/1n2D

>No brightness control


At least this should be possible to implement soon, the rest I haven't researched yet.

>I've tried suspending the laptop from Xfce

Suspend works only when the lid is closed.

[igorpec]

Bump. New version.

[jflesch]

It's looking great !

I just have a minor issue with external displays: When I plug my screen in, it's correctly detected but doesn't turn on. If I reboot the laptop with the screen still plugged in, the screen turns on correctly and everything works fine.
Anyway, it's still far better than what I got with the Olimex distribution: I haven't been able to use any external screens at all with it.

As soon as brightness control is available, Armbian is going straight to the eMMC of my Teres :-)

[lambda]

As soon as brightness control is available, Armbian is going straight to the eMMC of my Teres :-)

I think for brightness control you have to write to /sys/class/backlight/*/brightness or something like that. To get brightness control via the keyboard probably some configuration for acpid or similar service is needed, but I haven't looked at the latter solution myself yet, so this is just a wild guess.

HTH,
Harald

[igorpec]

Brightness control should be operational (via next adjustment/update) via power manager, key binding is AFAIK n/a and suspend/resume solution has to be figured out (after this, beta label goes off). Bluetooth is apparently working, but is not merged yet ... There are ofc still some issues in general + involved people are simply busy to push this further.

[JohnS]

It's getting better & better

John