Orange Pi Zero went to the market

[zgoda_j]

CNX-Software just posted that H2 based Orange Pi Zero appeared on Aliexpress for $7. It has been showcased here before, but now is on sale officially, though I can't find 512MB version on the store pages.

 

I'm eager to try 256MB version as my home IoT hub (mosquitto + nginx + uwsgi + Twisted, with MySQL database running on Zyxel NAS). It started with OPi PC then I switched to OPi One but I see it still can be lower spec device.

[Igor]

"WiFi is planned, BT also."

 

BT for the Allwinner XR819 WiFi module ?

Sorry, was not thinking clearly. This chip does not have it AFAIK. Haven't checked yet.

 

@Tkaiser

http://sprunge.us/KIjc

[trhunt]

WiFi is planned, BT also.

 

Opi zero (and PC2) just arrived on my desk today and I got Armbian running on it in no time   We need to add wireless drivers and apply fine tuning ... PC2 will be tested later.

 

h2.png

That's great news! I'm excited to receive mine (tomorrow!) and contribute to testing. Also, I just noticed the Orange Pi website now has a lubuntu build OPi zero for download. I'll look around to see if there is a XR819 driver floating around in there.

[Igor]

 I'll look around to see if there is a XR819 driver floating around in there.

 

I already add a driver but it was not working out of the box. No time for debug today.

 

BTW: Opi Zero is under auto daily build and image will be ready to download in matter of hours.

[tkaiser]

Igor: Xunlong has really done some development. Please check fex contents (mmc1) here: http://pastebin.com/480BYtj3

 

Their Lubuntu image combines mainline u-boot with smelly 3.4.39 kernel.

 

Edit: boot log: http://pastebin.com/V2cPZ0Jk

 

Edit 2: module_para and wifi_para also might not match (sorry, can't check, am on a tablet right now)

 

Edit 3: and dvfs settings are BS (always on 1.3V due to wrong pmu_level1. And pmuic_type = 0 -- really!?). Other settings also seem to be made with another device (Allwinner's reference platform?)

 

Edit 4: https://github.com/igorpecovnik/lib/commit/10fd0c9eabdf69fbe2a91869c778cb12c7f6771e

[tkaiser]

Some success regarding Wi-Fi:

 

 

macbookpro-tk:~ tk$ ssh 192.168.83.153

tk@192.168.83.153's password: 

  ___                               ____  _   _____              

 / _ \ _ __ __ _ _ __   __ _  ___  |  _ \(_) |__  /___ _ __ ___  

| | | | '__/ _` | '_ \ / _` |/ _ \ | |_) | |   / // _ \ '__/ _ \ 

| |_| | | | (_| | | | | (_| |  __/ |  __/| |  / /|  __/ | | (_) |

 \___/|_|  \__,_|_| |_|\__, |\___| |_|   |_| /____\___|_|  \___/ 

                       |___/                                     

 

Welcome to ARMBIAN Ubuntu 16.04.1 LTS 3.4.113-sun8i 

System load:   0.57            Up time:       2 min

Memory usage:  13 % of 241Mb  IP:            192.168.83.153

CPU temp:      40°C           

Usage of /:    9% of 15G    

 

 

[ 24 updates to install: apt-get upgrade ]

 

Last login: Wed Nov  9 14:14:54 2016

tk@orangepizero:~$ iwconfig wlan0

wlan0     IEEE 802.11bgn  ESSID:"Snort-Honeynet"  

          Mode:Managed  Frequency:2.412 GHz  Access Point: BC:05:43:BE:C1:E7   

          Bit Rate=39 Mb/s   Tx-Power=20 dBm   

          Retry  long limit:7   RTS thr:off   Fragment thr:off

          Power Management:on

          Link Quality=66/70  Signal level=-44 dBm  

          Rx invalid nwid:0  Rx invalid crypt:0  Rx invalid frag:0

          Tx excessive retries:1  Invalid misc:0   Missed beacon:0

 

tk@orangepizero:~$ nmcli dev wifi list

*  SSID                     MODE   CHAN  RATE       SIGNAL  BARS  SECURITY  

   Snort-Honeynet           Infra  1     54 Mbit/s  92      ▂▄▆█  WPA2      

   EasyBox-116D28           Infra  9     54 Mbit/s  84      ▂▄▆█  WPA1 WPA2 

   FRITZ!Box Fon WLAN 7360  Infra  1     54 Mbit/s  67      â–‚â–„â–†_  WPA1 WPA2 

   DISTORTEDPEOPLE          Infra  9     54 Mbit/s  55      â–‚â–„__  WPA1 WPA2 

   CubaLibre                Infra  1     54 Mbit/s  45      â–‚â–„__  WPA2      

   --                       Infra  6     54 Mbit/s  39      â–‚â–„__  WPA2      

   Wonder Woman             Infra  1     54 Mbit/s  34      â–‚â–„__  WPA1 WPA2 

   WLAN-341381              Infra  1     54 Mbit/s  34      â–‚â–„__  WPA2      

   FRITZ!Box Fon WLAN 7360  Infra  1     54 Mbit/s  27      â–‚___  WPA1 WPA2 

   UBNT2                    Infra  3     54 Mbit/s  25      â–‚___            

   Eichkatzenkind           Infra  5     54 Mbit/s  22      â–‚___  WPA2      

tk@orangepizero:~$ sudo armbianmonitor -u

[sudo] password for tk: 

/var/log/armhwinfo.log has been uploaded to http://sprunge.us/jNIT

Please post the URL in the Armbian forum where you've been asked for.

tk@orangepizero:~$ cat /etc/modprobe.d/dhd.conf

blacklist dhd

tk@orangepizero:~$ ls -la /system/etc/firmware/

total 140

drwxr-xr-x 2 root root   4096 Nov  9 13:57 .

drwxr-xr-x 3 root root   4096 Nov  9 13:55 ..

-rw-r--r-- 1 tk   tk     2308 Nov  9 13:57 boot_xr819.bin

-rw-r--r-- 1 tk   tk   126416 Nov  9 13:57 fw_xr819.bin

-rw-r--r-- 1 tk   tk      744 Nov  9 13:57 sdd_xr819.bin

 

 

 

So currently it's necessary to blacklist dhd module (if not things get weird when trying to load xradio_wlan) and firmware files need a proper location. Since I'm not that much interested in Wi-Fi I'll stop here

[tkaiser]

May I introduce the new low-consumption champion amongst Armbian supported devices? OPi Zero as wired data logger consumes just ~475 mW. I adjusted settings in the following way:

h3consumption -D 132 -u off

To explain these settings: -D to downclock DRAM and -u to disable USB completely. I also had to adjust fex file to disable Mali400 and CVBS since currently h3consumption is not fully ready for OPi Zero). And then I added this to /etc/rc.local to disable Wi-Fi reliably:

(sleep 60 && echo 000 >/proc/driver/wifi-pm/power) &

Will later try Wi-Fi only (disabling Ethernet completely) and also use '-c1 -m912' (only 1 CPU core active and limited to 912MHz) to limit peak consumption. Will later run both sysbench and cpuburn-a7 to get the idea how peak consumption behaves.

 

BTW: 475 mW is almost RPi Zero level but unlike RPi Zero the Orange Pi has fully working Ethernet in this mode and can immediately exceed RPi 2 performance! For comparison numbers see https://forum.armbian.com/index.php/topic/1748-sbc-consumptionperformance-comparisons/

 

Edit: With Armbian defaults for Orange Pi Zero and Wi-Fi disabled the board idles at ~900 mW, when using Xunlong's freshly released Lubuntu 14.04 image from 2016-11-8 it was almost 1000 mW more (see below) but it seems they release new images currently that might fix the broken settings)

[tkaiser]

Now also finished with Wi-Fi consumption measurement (disabled Ethernet in fex file). OPi Zero is 8m away from access point, 2 walls between, Wi-Fi powermanagement is active, board idles in an environment where 2.4 GHz band is overcrowded as hell. Idle consumption is below 550 mW which is simply great

[rajprakash]

.... Idle consumption is below 550 mW which is simply great

 

That's impressive.  That will make this board more battery pack friendly for remote applications.  Thanks for the initial numbers.

[tkaiser]

That's impressive.  That will make this board more battery pack friendly for remote applications.  Thanks for the initial numbers.

 

Please keep in mind that both numbers were the result of having either Wi-Fi or Ethernet active. We know that disabling Ethernet on H3 boards means 200 mW less (reference) so in case Ethernet driver is built as module and Ethernet connection is only needed from time to time you might be able to limit average idle consumption to 300 mW when no networking is needed. But this has to be confirmed, OPi Zero uses different Ethernet magnetics than the other Oranges.

[trhunt]

 Idle consumption is below 550 mW which is simply great

How is the temperature situation compared to say, NanoPI NEO? I know that armbian now manages thermal output pretty well, but I am curious. Does OPi zero run generally cooler? Seems like I^2*R heating should be less if you can idle it at 550 mW. (but I know that power consumption and heating are not strictly related!).

[haiqu]

Just bought two of the 512MB cards off AliExpress a couple of days ago and currently building Xenial. I think the Orange Pi Zero has a great future.

[tkaiser]

How is the temperature situation compared to say, NanoPI NEO?

 

NEO's first PCB revision had problems here, later revisions fixed that.

 

I had idle temps around 30°C on the OPi Zero now (please keep in mind that for unknown reasons with mainline u-boot the thermal readouts are somehow off so that this might be in reality 10°C more). The following is temperature monitoring while the board ran an 'apt-get upgrade' (limited to one CPU core running at 912 MHz max -- my preferred IoT settings) and then just idling around with only Wi-Fi enabled:

 

 

root@orangepizero:~# armbianmonitor -m
Stop monitoring using [ctrl]-[c]
Time        CPU    load %cpu %sys %usr %nice %io %irq   CPU
17:30:44:  912MHz  2.49  55%  14%  14%   0%  23%   3%   35°C
17:30:50:  912MHz  2.49  56%  14%  14%   0%  23%   3%   35°C
17:30:55:  912MHz  2.53  56%  14%  14%   0%  23%   3%   34°C
17:31:00:  912MHz  2.57  56%  14%  15%   0%  23%   3%   36°C
17:31:06:  912MHz  2.45  57%  14%  15%   0%  23%   3%   37°C
17:31:11:  912MHz  2.33  57%  14%  15%   0%  23%   3%   36°C
17:31:16:  912MHz  2.30  58%  15%  15%   0%  24%   3%   34°C
17:31:21:  912MHz  2.36  58%  15%  15%   0%  24%   3%   36°C
17:31:27:  912MHz  2.25  58%  15%  15%   0%  24%   3%   36°C
17:31:32:  912MHz  2.23  59%  16%  16%   0%  23%   3%   36°C
17:31:37:  912MHz  2.21  59%  15%  16%   0%  23%   3%   35°C
17:31:42:  912MHz  2.27  60%  15%  16%   0%  24%   3%   35°C
17:31:48:  912MHz  2.33  60%  15%  16%   0%  24%   3%   35°C
17:31:55:  912MHz  2.36  60%  15%  16%   0%  25%   3%   35°C
17:32:00:  912MHz  2.17  60%  15%  16%   0%  25%   3%   35°C
17:32:05:  912MHz  2.00  60%  15%  16%   0%  25%   3%   35°C
17:32:10:  240MHz  1.84  59%  15%  16%   0%  24%   3%   35°C
17:32:15:  912MHz  1.69  59%  15%  15%   0%  24%   3%   34°C
17:32:21:  912MHz  1.56  58%  15%  15%   0%  24%   3%   33°C
17:32:26:  240MHz  1.43  58%  15%  15%   0%  24%   3%   34°C
17:32:31:  912MHz  1.32  58%  15%  15%   0%  23%   3%   34°C^C
root@orangepizero:~# cat /etc/armbianmonitor/datasources/soctemp 
35
root@orangepizero:~# while true ; do
> cat /etc/armbianmonitor/datasources/soctemp 
> sleep 10
> done
33
34
33
33
32
32
32
33
32
31
31
31
31
33
32
31
32
31
30
31
32
32
31
31
32
30
31
30
28
31
30
31
31
30
30
31
32
29
31
31
30
32
31
30
30
31
31
31
31
30
30
31
31
29
30
29
31
31
30
31
31
30
30
31
31
30
30
30
31
31
30
33
31
32
30
32
31
30
31
30
31
30
30
31
30
31
31
30
30
31
31
31
31
31
30
30
30
31
30
30
31
31
30
30
30
31
30
30
30
31
31
31
30
29
30
32
30
30
31
31
31
29
30
31
31
31
31
30
30
29
30
30
30
31
31
30
29
30
29
29
30
31
30
31
30
31
32
31
31
31
31
30
30
30
29
31
31
30
30
30
30
30
30
 30
31
29
31
30
30
31
30
31
31
30
30
29
30
29
29
31
30
30
29
31
31
30
30
29
31
31
30
30
32
30
31
32
31
30
31
30
31
32
32
31
31
31
29
30
31
30
29
30
32
31
31
30
31
31
31
30
30
30
30
31
30
31
30
30
31
29
30
30
29
30
29
29
29
31
29
31
30
^C 

 

 

 

And since Igor had his Zero already on Tuesday I asked him how temperatures look like when running a sysbench for example: https://irclog.whitequark.org/linux-sunxi/2016-11-08#18127160;

 

All numbers without heatsink applied, IMO no heatsink is needed for normal operation since throttling is working fine (and normal workloads won't even trigger throttling). And it should be noted that Xunlong's OS images IMO ship with wrong settings (always remaining at 1.3V VDD_CPUX core voltage) which might lead to unnecessary high temperatures (to be confirmed, will test that later).

 

A final word regarding consumption: OPi Zero with only Wi-Fi enabled idled the whole night and without an active SSH session consumption fell below 450 mW (but please keep in mind that with enabled Wi-Fi consumption might be related to distance to AP and stuff like that)

[tkaiser]

Consumption update: All settings below made with 'IoT settings' (DRAM downclocked to 132 MHz, USB turned off):

 

- only Ethernet active, idle: below 500 mW (475 mW measured)

- only Wi-Fi active, idle: below 550 mW (450-520 mW measured, please keep in mind that actual consumption depends on your environment)

- both Ethernet and Wi-Fi disabled, idle: below 410 mW (400-410 mW measured)

 

That means in featureless mode OPi Zero consumes almost as less as an RPi Zero but more importantly in connected situations (be it wired or wireless) consumption is way below RPi Zero since there adding USB Wi-Fi dongles or an USB Ethernet adapter adds significantly to overall consumption).

 

This also means that difference between Ethernet active and disabled is just 70 mW (we saw 200 mW with all other H3 boards so far, the difference might be due to H2+ not featuring Gbit Ethernet capabilties any more and/or different Ethernet magnetics on this board).

 

All in all that looks really promising. A board that performs at RPi 2 level keeps real-world consumption below RPi Zero level while having almost RPi 3 feature set (Wi-Fi + Ethernet, but here Ethernet is not hanging off USB but is real, and here we have 4 real USB ports that do not have to share bandwidth). Combine that with the PoE option we can only hope that Xunlong will start to solder 1 MB SPI NOR flash as default in the future, then this board would already be perfect.

[prime_jp]

OPi Zero have i2s?

[tkaiser]

OPi Zero have i2s?

 

Nope, not available, you would have to use any other OPi (see I2S mini tutorial) or H3 based NanoPi (NEO PCB rev 1.0 also has no I2S!).

 

Linux-sunxi wiki page is already in pretty good shape regarding hardware details! http://linux-sunxi.org/Xunlong_Orange_Pi_Zero

[prime_jp]

Nope, not available, you would have to use any other OPi (see I2S mini tutorial) or H3 based NanoPi (NEO PCB rev 1.0 also has no I2S!).

 

Linux-sunxi wiki page is already in pretty good shape regarding hardware details! http://linux-sunxi.org/Xunlong_Orange_Pi_Zero

 

Thank you reply.

Oh bad. I need i2s. if it have i2s, it was the best product for me.

[haiqu]

Had some weird error messages building Xenial for the OPi Zero. After extraction and "Applying distribution specific tweaks" I got:

 

`sed can't read /home/pi/output/cache/sdcard/etc/ssh/sshd_config`

 

then a few lines later:

 

`sed can't read /home/pi/output/cache/sdcard/etc/lirc/hardware.conf` (3 times)

 

Then there was a cp error from the same .../lirc directory.

 

I ended up with an image but not sure if it will run.

[tkaiser]

Had some weird error messages building Xenial for the OPi Zero.

 

Please let's keep this thread focused on OPi Zero. There's absolutely no need to build an OS image yourself, simply use the one we provide now! https://www.armbian.com/orange-pi-zero/

 

This helps both you and us since we can focus on real problems.

[tkaiser]

And it should be noted that Xunlong's OS images IMO ship with wrong settings (always remaining at 1.3V VDD_CPUX core voltage) which might lead to unnecessary high temperatures (to be confirmed, will test that later).

 

OMFG, just tested with Xunlong's own Lubuntu 14.04 release relying on their default settings:

avg-cpu:  %user   %nice %system %iowait  %steal   %idle
           0.00    0.07    0.13    0.07    0.00   99.73

Device:            tps    kB_read/s    kB_wrtn/s    kB_read    kB_wrtn
mmcblk0           0.40         0.00         7.20          0         36
sda               0.00         0.00         0.00          0          0

^C
root@orangepi:~# cat /sys/devices/virtual/thermal/thermal_zone1/temp
61
root@orangepi:~# grep -c processor /proc/cpuinfo 
3

The board is absolutely idle (see %idle above: 99.73 percent), idle consumption is 1860 mW (yeah, close to 2W!!!), SoC temperature is around 60°C and one CPU core already has been killed. Reason is simple: This OS image relys on crappy Allwinner defaults, the dvfs settings are completely wrong and settings ignore everything community has done around H3 within the last 12 months.

 

The settings used with this image are broken, the CPU is always fed with 1.3V, killing CPU cores is preferred over sane throttling (that's why only 3 CPU cores are active) and even if max cpufreq with this image is limited to 1008 MHz CPU cores are killed. And I did not even do anything heavy, it was only installation of RPi Monitor that lead to 'overheating' and killing the 4th CPU core. So by using Armbian (using linux-sunxi community settings!) you get both more performance and better thermal behaviour.

 

Xunlong, please stop doing software or at least look what community did in the last 12 months! Now folks like Michael Larabel will take an OPi Zero, will take your crappy Lubuntu image, run Phoronix test suite and OPi Zero will get horribly low benchmark scores. And the whole world again laughs about Xunlong devices but in reality it's just dumb Allwinner default settings that you (again) use in your OS images.

 

Why the hell did you do that? Please stop providing OS images with your broken settings, it's easy to adopt good settings instead: https://github.com/igorpecovnik/lib/blob/master/config/fex/orangepizero.fex (cooler_table, ths and dvfs settings need to be adopted). Please immediately remove this Lubuntu crap or at least fix script.bin and include a link to Armbian download page. Or do you really want that users/testers again start to spread the word your hardware would be crap?

 

Edit: Today Xunlong released 3 new OS images for OPi Zero, I looked at something called 'Debian_server_For_OrangePiZero'. Settings still 100 percent wrong and board overheating like hell especially due to incorrect dvfs settings (it has to be 'pmuic_type = 1' and there need to be at least one dvfs operating point defined that makes use of 1.1V at a reasonable cpufreq):

 

 

 

[dvfs_table]
pmuic_type = 0
pmu_gpio0 = port:PL06<1><1><2><1>
pmu_level0 = 11300
pmu_level1 = 1100
max_freq = 1008000000
min_freq = 480000000
boot_freq = 1008000000
LV_count = 8
LV1_freq = 1200000000
LV1_volt = 1300
LV2_freq = 1008000000
LV2_volt = 1200
LV3_freq = 0
LV3_volt = 1100
LV4_freq = 0
LV4_volt = 1100
LV5_freq = 0
LV5_volt = 1100
LV6_freq = 0
LV6_volt = 1100
LV7_freq = 0
LV7_volt = 1100
LV8_freq = 0
LV8_volt = 1100 

 

 

 

Apart from that this image is still based on an unpatched 3.4.39 kernel (why on earth?) but at least a script called /etc/init.d/firstrun resized the rootfs automagically. With the Lubuntu image from two days ago not even an 'apt-get upgrade' worked sinze rootfs was filles with 95%. So there's at least one person active at Xunlong trying to re-invent the wheel providing OS images that suck so people are able to claim Xunlong boards are a fail. Great.

[ddxcb]

And I just order a neo, but this one i want for being a web cam viewer. Since it can have tv out.

[hojnikb]

Xunlong should just stick to making hardware and force allwinner to release appropriate sdks, documentation and source code, so community can do better.

[rajprakash]

... Wi-Fi ... 

 

 

Does the OPi Zero onboard wifi support AP mode?  Hit us with the output of 'iw list' please!

[tkaiser]

Does the OPi Zero onboard wifi support AP mode?  Hit us with the output of 'iw list' please!

 

AP mode is supported, see http://filez.zoobab.com/allwinner/h2/201609022/lichee/linux-3.4/drivers/net/wireless/xradio/

 

And here 'iw list':

  ___                               ____  _   _____              

 / _ \ _ __ __ _ _ __   __ _  ___  |  _ \(_) |__  /___ _ __ ___  

| | | | '__/ _` | '_ \ / _` |/ _ \ | |_) | |   / // _ \ '__/ _ \ 

| |_| | | | (_| | | | | (_| |  __/ |  __/| |  / /|  __/ | | (_) |

 \___/|_|  \__,_|_| |_|\__, |\___| |_|   |_| /____\___|_|  \___/ 

                       |___/                                     

 

Welcome to ARMBIAN Ubuntu 16.04.1 LTS 4.9.0-sun8i 

System load:   0.05            Up time:       4 min

Memory usage:  10 % of 241Mb  IP:            192.168.83.162

Usage of /:    16% of 4.0G   

 

Last login: Fri Nov 11 13:08:38 2016

tk@orangepizero:~$ sudo -s

[sudo] password for tk: 

root@orangepizero:~# iw list

root@orangepizero:~# armbianmonitor -u

/var/log/armhwinfo.log has been uploaded to http://sprunge.us/jNJW

Please post the URL in the Armbian forum where you've been asked for.

 

 

Sorry, am already on latest and greatest mainline kernel and there Wi-Fi driver is missing (needs to be ported).

[tkaiser]

And now I'm pretty confused. When I did some performance/consumption tests a while back I tested with various Raspberries and got sysbench results as follows: RPi Zero: 915s, RPi B+: 1311s, RPi 2: 192s and RPi 3: 120 seconds. The latter result is confirmed eg by this RPi 2 with RPi 3 comparison.

 

So how on earth is it possible that $7 Orange Pi Zero outperforms even RPi 3 and needs only 117.5 seconds?!

 

macbookpro-tk:~ tk$ ssh orangepizero.local

tk@orangepizero.local's password: 

  ___                               ____  _   _____              

 / _ \ _ __ __ _ _ __   __ _  ___  |  _ \(_) |__  /___ _ __ ___  

| | | | '__/ _` | '_ \ / _` |/ _ \ | |_) | |   / // _ \ '__/ _ \ 

| |_| | | | (_| | | | | (_| |  __/ |  __/| |  / /|  __/ | | (_) |

 \___/|_|  \__,_|_| |_|\__, |\___| |_|   |_| /____\___|_|  \___/ 

                       |___/                                     

 

Welcome to ARMBIAN Ubuntu 16.04.1 LTS 4.9.0-sun8i 

System load:   0.22            Up time:       29 min

Memory usage:  11 % of 241Mb  IP:            192.168.83.162

Usage of /:    16% of 4.0G   

 

Last login: Fri Nov 11 14:58:56 2016 from 192.168.83.91

tk@orangepizero:~$ uname -a

Linux orangepizero 4.9.0-sun8i #4 SMP Fri Nov 11 13:03:52 CET 2016 armv7l armv7l armv7l GNU/Linux

tk@orangepizero:~$ lsb_release -c

Codename: xenial

tk@orangepizero:~$ cat /etc/armbian-release 

# PLEASE DO NOT EDIT THIS FILE

BOARD=orangepizero

BOARD_NAME="Orange Pi Zero"

VERSION=5.24

LINUXFAMILY=sun8i

BRANCH=dev

ARCH=arm

tk@orangepizero:~$ sysbench --test=cpu --cpu-max-prime=20000 --num-threads=4 run

sysbench 0.4.12:  multi-threaded system evaluation benchmark

 

Running the test with following options:

Number of threads: 4

 

Doing CPU performance benchmark

 

Threads started!

Done.

 

Maximum prime number checked in CPU test: 20000

 

 

Test execution summary:

    total time:                          117.5139s

    total number of events:              10000

    total time taken by event execution: 469.8815

    per-request statistics:

         min:                                 46.72ms

         avg:                                 46.99ms

         max:                                288.34ms

         approx.  95 percentile:              46.77ms

 

Threads fairness:

    events (avg/stddev):           2500.0000/0.71

    execution time (avg/stddev):   117.4704/0.04

[hojnikb]

Thats either slower ddr2 memory on the rpi, slower l2 cache or throttling. Could be a combination of all those things.

[tkaiser]

Thats either slower ddr2 memory on the rpi, slower l2 cache or throttling. Could be a combination of all those things.

 

Nope, sysbench does not depend on memory bandwidth at all. It's an example for a close to moronic benchmark that shouldn't be used. But it demonstrates perfectly how marketing works since RPi foundation constantly uses these BS numbers to demonstrate how much faster newer Raspberries are compared to their predecessors.

 

OrangePi Zero is 'faster' than Raspberry Pi 3 since sysbench is not testing hardware but compiler switches and the availability of specific CPU instructions. It's a joke when used as 'general purpose benchmark'.

 

OPi Zero had multiple advantages:

• I ran Ubuntu Xenial on OPi Zero while Pi users for whatever reasons all the time use Debian Jessie (even if there's an Ubuntu Mate version available for RPi 2/3 that is a lot faster). The sysbench binary is 30% faster in Xenial compared to Jessie
• H2+ as used on OPi Zero is ARMv7 (Cortex-A7) while BCM2837 on RPi 3 is ARMv8 (Cortex-A53). H2+ uses software made for ARMv7 while RPi 3 has to use software made for ARMv6 (since RPi foundation prefers to just maintain one single set of software for all the Raspberries). So those bad benchmark scores of both RPi 2 and 3 are just the result of running inefficient software for an outdated CPU instruction set.
• If RPi 3 would've been allowed to run ARMv8 code then it would not take 120 seconds to finish but less than 10 seconds. Important: Using Raspbian on anything that is not a 1st gen Raspberry is always wrong since performance sucks automatically.

If it's about CPU power the poor RPi 3 suffers from RPi foundation not providing appropriate software. But if it's about stuff like network or IO bandwidth then the $7 device (OPi Zero) really outperforms any RPi since all those Raspberries have just one single USB 2.0 OTG port to the outside.

 

The combined overall bandwidth of the $7 device is magnitudes better compared to an overpriced RPi: 4 real USB 2.0 ports and not just 1, one real Ethernet port and not just one hanging off the USB bus (and having to share bandwidth with the other 4 USB ports which makes every RPi out there such a horribly slow NAS).

 

BTW: Since I'm on kernel 4.9 now all the time another consumption update: OPi Zero idles with 4.9 with schedutil cpufreq governor (jumping between 240 MHz and 1200 MHz), USB and only Ethernet active at 620 mW. DRAM clockspeed is 408 MHz like with legacy defaults. I believe my kernel also supports HDMI output now, so time to search for OPi One and check whether this is true or not. The lower default consumption now with vanilla kernel compared to legacy might change over time. More features --> higher consumption.

[frosticecold]

Ordered the 512 variant board like 2 days ago, waiting for it so i could help test stuff out.

 

I think it will really shine when it gets the real treatment from you guys.

 

tkaiser, how hard is it to port the driver to the mainline kernel?

[tkaiser]

tkaiser, how hard is it to port the driver to the mainline kernel?

 

If we're lucky MoeIcenowy finishes porting tomorrow  She pushed already a bunch of H2+ patches today to mainline.

[Igor]

My 2cents.

 

Wireless driver works now out of the box. Almost Firmware is at standard place, dhd driver was patched to prevent false recognition, but driver needs to be loaded twice to work   I am done for the week, so if one have time to proceed from here.

Probably we should need to power WiFi in u-boot ... I try with: CONFIG_MMC_SUNXI_SLOT_EXTRA=2 which I thought it could do the work, but no luck. 

[hojnikb]

Nice to hear things are progressing on the H2+ front. Hopefully H5 and PC2 get the same love as well

Question: Is armbian gonna switch to 64 bit images for H5 boards ? Would that even make sense on a 1GB board ?