Turl changed the topic of #linux-sunxi to: Allwinner/sunxi /development discussion - did you try looking at our wiki? https://linux-sunxi.org - Don't ask to ask. Just ask and wait! - https://github.com/linux-sunxi/ - Logs at http://irclog.whitequark.org/linux-sunxi
al1o has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]
<apritzel> ssvb: good job, just started a 64-bit U-Boot from SPI NOR on the Pine64
<apritzel> a bit of hackery needed as usual, but it works!
* vagrantc claps!
mosterta has quit [Ping timeout: 240 seconds]
<ssvb> apritzel: well, it's an ugly hack in v1, I only pushed it out because the U-Boot merge window was about to close in 9 minutes :-)
<ssvb> will clean it up tomorrow
<apritzel> ssvb: pah, works for me (tm) ;-)
<jrg> opi boots armbian. neat.
<jrg> let me read the armbian wiki to see how you get this thing over to the emmc
KotCzarny has joined #linux-sunxi
vagrantc has quit [Quit: leaving]
<TheLinuxBug> jrg: I have been moving it was shipped to my new place haven't checked mail yet, wonder if mine arived :D
<TheLinuxBug> Winamp Playing: One - Winamp *** 63. Caskey - I'm the One | 320 kbps - 44 khz - Stereo |
<TheLinuxBug> Winamp Playing: ne - Winamp *** 63. Caskey - I'm the One | 320 kbps - 44 khz - Stereo |
<TheLinuxBug> er oops sry
<TheLinuxBug> that wasn't intentional just dropped something on my keyboard ;(
<jrg> TheLinuxBug: well i just installed armbian on mine
<jrg> i just used the screws i had from another bananapi case to stand it up on something
<jrg> hopefully cases become a bit more available for these things
egbert has quit [Disconnected by services]
egbert has joined #linux-sunxi
<jrg> otherwise my son and i will have arts and crafts day trying to make one out of construction paper and mache
<jrg> trying to figure out how to xfer armbian to the emmc now
<jrg> nand-sata-install
<jrg> oops
<jrg> wow. it can't be this easy lol
<jrg> i wonder if the opi can use the SD still after the os is installed on the emmc... hopefully armbian sets it up that way by default
ninolein has quit [Ping timeout: 250 seconds]
ninolein has joined #linux-sunxi
<jrg> hm. so after i run nand-sata-install i can just pull the SD and reboot ?
apritzel has quit [Ping timeout: 244 seconds]
libv_ has joined #linux-sunxi
libv has quit [Ping timeout: 260 seconds]
cnxsoft has joined #linux-sunxi
libv has joined #linux-sunxi
libv_ has quit [Ping timeout: 240 seconds]
libv_ has joined #linux-sunxi
libv has quit [Ping timeout: 264 seconds]
libv_ has quit [Ping timeout: 260 seconds]
ganbold has joined #linux-sunxi
Ultrasauce_ has quit [Quit: No Ping reply in 180 seconds.]
indy has quit [Ping timeout: 250 seconds]
Ultrasauce has joined #linux-sunxi
cajg has quit [Ping timeout: 272 seconds]
indy has joined #linux-sunxi
cajg has joined #linux-sunxi
<juri_> found my problem. overheating a20. add heat sinks, folks!
<juri_> (mind you, i still have two problems in my iscsi driver.. but hey.. ;)
medvid has quit [Ping timeout: 260 seconds]
medvid has joined #linux-sunxi
orly_owl has joined #linux-sunxi
tsuggs has joined #linux-sunxi
<jrg> heh
cnxsoft has quit [Read error: Connection reset by peer]
cnxsoft1 has joined #linux-sunxi
cnxsoft1 is now known as cnxsoft
zuikis has joined #linux-sunxi
Netlynx has joined #linux-sunxi
JohnDoe_71Rus has joined #linux-sunxi
jernej has joined #linux-sunxi
fredy has quit [Excess Flood]
fredy has joined #linux-sunxi
tkaiser has joined #linux-sunxi
jernej has quit [Ping timeout: 240 seconds]
orly_owl has quit [Quit: leaving]
jernej has joined #linux-sunxi
<TheLinuxBug> jrg: My OPi 2E arrived, but the box was manhandled :(
<TheLinuxBug> gonna take a pic before I open to see if it is ok
<TheLinuxBug> phew inside box looks unharmed so far
Dodger78 has joined #linux-sunxi
<TheLinuxBug> w00t now to find a memory card
<TheLinuxBug> and see how this baby rolls
<Dodger78> hi guys, im on cubietruck, latest mainline. i want to attach a usb-sata converter, it gets recognized, the disk starts spinning up but then stops and respins up again. the adapter has a external power supply and is working on my linux notebook. did i miss something in mainline kernel config on my cubie ?
bonbons has joined #linux-sunxi
<NiteHawk> Dodger78: did you check dmesg? also: does the adapter work when used on a different linux system (PC)?
<montjoie> Dodger78: strange, I got the same problem, but at final my hdd burn...
<Dodger78> hi , i checked dmesg , and yes the adapter worked on my linux mint laptop
<Dodger78> the hdd spins up, then some access noises, then spins up again. i will grab it for a sek and try it again on my other machine
<NiteHawk> that's an ASMedia 1051 / USB 3.0 ?
<Dodger78> hi think so
<Dodger78> im just searching for the chipset
leilei has joined #linux-sunxi
<NiteHawk> don't bother - the USB VID/PID identifies it as ASM1051
<Dodger78> is it a problem ?
<NiteHawk> not sure. i'm currently reading up on a firmware update for those chips
<Dodger78> yeah found something like this also
mosterta has joined #linux-sunxi
<NiteHawk> apparently some disks are problematic when combined with that chipset: http://forum.odroid.com/viewtopic.php?f=97&t=14384
<NiteHawk> and early revisions of the chipset seemed to cause more problems than later ones: https://bugzilla.redhat.com/show_bug.cgi?id=1230336
<Dodger78> :-( oh ok then my cubie is not to blame
<NiteHawk> still it's a bit strange that the device works with mint on the laptop, but not on the cubietruck
<Dodger78> y i will try and investigate further
<Dodger78> is there a driver in the kernel needed for asm1051 ? cant find it
<NiteHawk> might also be a hardware issue - something like power supply issues / interference (e.g. ground loop) between the two?
<NiteHawk> did you enable USB mass storage support? do other drives (e.g. flash sticks) work? if so, i don't think the USB-SATA bridge should require special treatment
<Dodger78> usb mass storage is on ..
<Dodger78> will check a stick
scream has joined #linux-sunxi
<tkaiser> Dodger78: Just in case you buy enclosures again... After so much joy with ASMedia chipsets I check for JMicron JMS567/JSM568 when I buy new stuff. (Some) reasons: http://linux-sunxi.org/USB/UAS
<Dodger78> n usb stick is working
<Dodger78> thanks for the hint tkaiser
apritzel has joined #linux-sunxi
Shirasaka-Hazumi has quit [Quit: ZNC 1.6.2+deb2+b1 - http://znc.in]
Shirasaka-Hazumi has joined #linux-sunxi
ninolein has quit [Remote host closed the connection]
Dodger78 has quit [Quit: Lost terminal]
<tkaiser> ssvb: Some news regarding BPi M2+ DRAM clockspeeds. It seems not related to boot0 vs. u-boot or u-boot version: http://forum.armbian.com/index.php/topic/1322-testers-wanted-testing-dram-reliability-on-bpi-m2/?view=getlastpost
apritzel has quit [Ping timeout: 244 seconds]
<tkaiser> And IgorPec confirmed horrible throttling/thermal results on BPi M2+ (even worse than my experiences)
Dodger78 has joined #linux-sunxi
<tkaiser> Dodger78: Do you have WiFi enabled on your Cubietruck. Just remembered some 'nice' interferences with USB3 and 2.4GHz band some time ago. Though USB3 modes shouldn't be used between A20 and your ASMedia bridge.
<Dodger78> my adapter has a ASM1053 , and my laptop has got a usb3 interface it seems this make a difference
<Dodger78> tkaiser , can be i will deactivate it
<Dodger78> another , different question: is spdif audio working on cubietruck mainline now ? my optical wire is dark, should be lighted red or not ?
<tkaiser> Dodger78: Please read through the footnotes on the USB/UAS page I referenced before: http://www.spinics.net/lists/linux-usb/msg119432.html
<NiteHawk> Dodger78: the ASM1053 can't be detected reliably via USB 2.0, and will be treated as ASM1051 (includings quirks)
<Dodger78> Nitehawk / kaiser , i tried to remove the quirks detection via uas-detect.h for all 1051/1053
<Dodger78> problem is , currently i cant tell if it works or not, i commented out the flag settings, but from dmesg i cant tell the difference
tlwoerner has quit [Ping timeout: 244 seconds]
<NiteHawk> Dodger78: so the dmesg output / lsusb is different on the laptop (i.e. when connected via USB3)?
jernej has quit [Ping timeout: 240 seconds]
<Dodger78> from laptop without attached disk
<Dodger78> uas is found, asm1053
<Dodger78> here again cubietruck with my modified header
apritzel has joined #linux-sunxi
<NiteHawk> well... at least now you're getting a disk attached (sdb). And the previous 'Jun 4 18:01:25 cubieez mtp-probe: checking bus 2, device 3: "/sys/devices/platform/soc@01c00000/1c1c000.usb/usb2/2-1"' is now absent
<Dodger78> but the spinup of the disk doesnt work ^^ or lets say it re-spins again and again. power supply is confirmed and ok
<Dodger78> i did not attack the 1.5gb WD drive
<Dodger78> 1.5tb
apritzel has quit [Ping timeout: 244 seconds]
<montjoie> Dodger78: could you give the commercial name of your product ?
<NiteHawk> he did above (in the review link): Inateck UA0001
<NiteHawk> err... UA1001, sorry
<montjoie> thanks, I didnt see it
reinforce has joined #linux-sunxi
<tkaiser> ssvb: wens: Done with DRAM reliability testing for crapboard M2+. It seems FEL mode (or lets better say providing power also on the Micro USB port) seems to be the problem: http://forum.armbian.com/index.php/topic/1322-testers-wanted-testing-dram-reliability-on-bpi-m2/?view=getlastpost
<tkaiser> Since thermal issues are also confirmed and the board is insanely overprices wasting any more second with this broken hardware design isn't worth the time/efforts.
<ssvb> tkaiser: how does it fail in FEL mode? does the test detect corruption or just dies?
<tkaiser> ssvb: I posted the logs in Armbian forum. At 624MHz corruption gets detected, with the 672 MHz I tried out it ended up with an Oops (showing the clockspeeds used)
<tkaiser> To me it seems the whole problem is just power related so testing without something connected to the Micro USB power that might also power the board seems mandatory (but also a waste of time)
Wizzup has quit [Ping timeout: 244 seconds]
<ssvb> tkaiser: thanks for running all these tests, I think it is important to figure out what is going on
_whitelogger_ has joined #linux-sunxi
jemk_ has joined #linux-sunxi
nihcas_ has joined #linux-sunxi
<tkaiser> ssvb: Hmm... It worked fine with 720 with the u-boot+spl from your v3 package on BPi M2+ (which uses DDR3 instead of DDR3L as on the Oranges)
Rondom has quit [Disconnected by services]
ruben-ikmaak has joined #linux-sunxi
Rondom_ has joined #linux-sunxi
montjoie_ has joined #linux-sunxi
uwe__ has joined #linux-sunxi
medvid has quit [Ping timeout: 260 seconds]
nihcas has quit [Ping timeout: 260 seconds]
<ssvb> tkaiser: please just run the test with 'v4-zqdebug' and we will confirm or debunk the current speculations :-)
_whitelogger has quit [Ping timeout: 260 seconds]
jemk has quit [Ping timeout: 260 seconds]
Turl has quit [Ping timeout: 260 seconds]
ikmaak has quit [Remote host closed the connection]
sW` has quit [Ping timeout: 260 seconds]
specing has quit [Ping timeout: 260 seconds]
LostSoul has quit [Ping timeout: 260 seconds]
uwe_ has quit [Ping timeout: 260 seconds]
montjoie has quit [Ping timeout: 260 seconds]
cyrozap has quit [Ping timeout: 260 seconds]
RzR has quit [Ping timeout: 260 seconds]
agraf has quit [Ping timeout: 260 seconds]
sW`` has joined #linux-sunxi
nihcas_ is now known as nihcas
agraf has joined #linux-sunxi
RzR has joined #linux-sunxi
Wizzup has joined #linux-sunxi
<tkaiser> ssvb: Did 'dd if=orange-pi-pc-720-u-boot-sunxi-with-spl of=/dev/mmcblk0 bs=1024 seek=8 status=noxfer', now rebooting, then testing
Turl has joined #linux-sunxi
cyrozap has joined #linux-sunxi
specing has joined #linux-sunxi
LostSoul has joined #linux-sunxi
<tkaiser> ssvb: Or did you want me again to try FEL booting?
<ssvb> tkaiser: yes, FEL booting would be more interesting, because it was problematic with v3
<tkaiser> ssvb: Ok, one last try. Running at 720 MHz with the Armbian image seems stable.
libv has joined #linux-sunxi
ganbold has quit [Ping timeout: 246 seconds]
<tkaiser> With 672 the cube already was frozen when the display woke up from sleep
<ssvb> so v4-zqdebug failed at 672MHz in FEL mode?
<tkaiser> ssvb: Yes, the cube wasn't spinning while memtest ran: http://pastebin.com/haFXBXBn -- will now try 648 MHz
libv_ has joined #linux-sunxi
libv has quit [Ping timeout: 258 seconds]
<tkaiser> ssvb: Cube stopped spinning a few seconds later at 648 MHz. Memtester aborted: http://pastebin.com/A50ENmUk
<ssvb> thanks a lot for doing these tests
<ssvb> zq calibration does not seem to be related and it's good to know
<ssvb> regarding the power, how are you powering the board?
<ssvb> do you plug both the PSU and the micro usb cable?
<tkaiser> ssvb: With 624MHz cube stops maybe 10 secs later.
libv has joined #linux-sunxi
<tkaiser> ssvb: Yes, can also try it with USB only (but the host is a Pine64+ with 650mA max on his USB port)
<ssvb> hmm
libv_ has quit [Ping timeout: 258 seconds]
<tkaiser> ssvb: I don't want to waste regarding powering with this board. Since with the Oranges it's confirmed that it works. As you suggested: Wiki entry 'Strongly recommended to stay away from this board'. Period.
leilei has quit [Remote host closed the connection]
<ssvb> tkaiser: we still don't know if there might be some problem with the DRAM controller in initialization on H3
mpmc has quit [Quit: ZNC 1.6.1+deb1+jessie0 - http://znc.in]
<tkaiser> I've overwritten u-boot+spl on my Armbian test image with the 744 MHz variant of your v4. Will now test this (so powering through Micro USB doesn't play any role)
<tkaiser> With your v3 and 744 MHz I got the glowing red background pretty soon
ganbold has joined #linux-sunxi
<tkaiser> Glowing red after 5:30 min.
mpmc has joined #linux-sunxi
al1o has joined #linux-sunxi
<tkaiser> ssvb: This was with your v4: http://pastebin.com/ee3zVAQx Then I overwrote it with v3 again and re-tested. Glowing red after 6:30 min (which I would call 'identical behaviour): http://pastebin.com/ULc7JFag
JohnDoe_71Rus has quit [Quit: KVIrc KVIrc Aria 4.9.2, revision: git-6627-gf708225, sources date: 20160102, built on: 2016-05-12 17:32:15 UTC git-6627-gf708225 http://www.kvirc.net/]
<phipli> Hum
<phipli> does anyone have an Olimex A20 micro board?
<phipli> Anyone got it booting with a recent version of Armbian that they can give me a hint what options / changes I need to get it to boot properly?
<phipli> just tried it, and it did the same thing
ganbold_ has joined #linux-sunxi
jernej has joined #linux-sunxi
medvid has joined #linux-sunxi
<ssvb> tkaiser: thanks
<phipli> hum
<phipli> actually, looks like it might be a different issue
<phipli> stops at "<6>ADDRCONF(NETDEV_UP): eth0: link is not ready" for ages
<phipli> then actually does go to a login prompt
<phipli> but the HDMI hasn't kicked in
ganbold has quit [Ping timeout: 264 seconds]
phipli has quit [Ping timeout: 244 seconds]
jernej has quit [Ping timeout: 244 seconds]
<jrg> tkaiser: i received my orange pi plus 2e and installed a real version of armbian on it
<jrg> i'll probably work on getting samba working on it if i can (hopefully)
<tkaiser> jrg: The Armbian image is already outdated. You should use fex2bin and replace /boot/bin/orangepiplus2e.bin with the contents of this https://github.com/igorpecovnik/lib/blob/6b37f7ad48ca4e59e8026e5998eab113e32bf0f3/config/fex/orangepiplus2e.fex
<tkaiser> And then remove the line in /etc/rc.local that wants to start sun8i-corekeeper.sh on every boot. We now have an in-kernel core-keeper but in OPi Plus 2E this shouldn't be a problem anyway since the board shows nice heat spreading capabilties
<tkaiser> ssvb: I did a last test before with BPi M2+. I used your v3 settings with 672 MHz on my Armbian image, restarted with this and connected both Micro USB to Pine64 and barrel connector to PSU (to see whether the board becomes instable when 2 power sources are present). But memtester worked flawlessly. I added a note to the Wiki and will stop testing: http://linux-sunxi.org/Sinovoip_Banana_Pi_M2%2B#DRAM_clock_speed_limit
libv has quit [Ping timeout: 272 seconds]
apritzel has quit [Ping timeout: 244 seconds]
libv has joined #linux-sunxi
libv has quit [Ping timeout: 244 seconds]
libv has joined #linux-sunxi
jernej has joined #linux-sunxi
valkyr1e has quit [Quit: Bye.]
valkyr1e has joined #linux-sunxi
JohnDoe_71Rus has joined #linux-sunxi
jernej has quit [Ping timeout: 260 seconds]
<jrg> oh ok. i'll try it out
<jrg> wish you told me sooner. i already configured it to join an AD heh
<jrg> oh. you did.. i just didn't see it
<jrg> tkaiser: so after running fex2bin on the .fex (above) i just replace script.bin and reobot?
<jrg> with the bin i created with the new fex?
<jrg> only one way to find out! :)
<tkaiser> jrg: Sure. But it's not that important. Maybe 1-2 percent more performance when doing heavy workloads (different thermal definitions when throttling should start)
<jrg> ah ok. well. i replaced it with the one you pointed to.. seems to boot. also removed the line from rc.local
<tkaiser> jrg: Then you're find. From now it's just apt-get update/upgrade
<jrg> is there anything else on the image that is outdated or was that it?
<jrg> ah. opi+2e can update the kernel without major issues?
<jrg> most of it is mainlined already?
<tkaiser> jrg: Sure, same with u-boot patches. Everything delivered from an apt repository. And no, no mainline kernel images that soon since without the THS stuff working (throttling, more important for smaller H3 boards) it would be too dangerous.
<jrg> well at least it is in a better position to be used than the bpi heh
<jrg> thanks for the help. i'll work on making this what the bpi should have been. although. the bpi has been pretty solid when i throttle it to 1.2GHz.. but no easy way to update it like you said on the forums
<jrg> tkaiser: is the behavior for booting off the SD similar to the bpi? if i put the SD in will it still try to boot from it or is that something that is differnt for the opi+2e?
<tkaiser> jrg: Same boot priority. SD card wins
<jrg> damn heh. is there an easy way to fix that so emmc has priority over sd for boot?
<jrg> or does it have to be from the SD then point to the emmc?
<tkaiser> Not that I know of, so bootloader has to remain on SD card
<jrg> ah ok
libv_ has joined #linux-sunxi
<jrg> is there a wiki on how to get the bootloader on the sd or does nand-sata script do that for you when you move to the emmc?
<tkaiser> jrg: Don't know. I used dd the last time I tried it. IIRC someone posted something in Armbian forum.
libv has quit [Ping timeout: 276 seconds]
<tkaiser> And there are also known issues, see starting from #22: http://forum.armbian.com/index.php/topic/1193-orange-pi-plus-2e-now-available/
<jrg> ok. i'll read. thanks.
<tkaiser> jrg: And here you find some notes on using an SD card that will be inserted after the board booted already from eMMC: http://forum.armbian.com/index.php/topic/1187-boot-from-emmc-with-blank-sd-in-slot-bpi-m3/
<tkaiser> It's about BPi M3 but there shouldn't be a difference with H3 boards.
<jrg> yah. i was reading that earlier when trying to figure out how to boot emmc while sd was inserted
<jrg> i'm trying to just keep the sd in tho and boot from emmc while using sd for storage
<tkaiser> Should work both ways and also in a fashion that you can remove the SD card. I wanted to write a tutorial but different priorities and also my OPi Plus 2E is now at a customer's location serving files.
<jrg> already?? lol
<jrg> So we know now that both firstrun and nand-sata-install need some attention/fixes. Unfortunately my OPi Plus 2E went productive yesterday :\ <- lol.
<jrg> at least you managed to fix a lot of necessary things before you gave it up :) i'll work with just the emmc for now until i find a sane way to copy the bootloader to the SD and run it properly.. in the meantime i'll just work with the 16GB.. that should be enough for what i have to do
<jrg> i'm going to repurpose my bananapi as a webserver once i get the opi set up properly. thanks for all the help.
mosterta has quit [Ping timeout: 260 seconds]
<MoeIcenowy> oh it seems that rtl8723au's bluetooth is now supported directly in the mainline kernel by btrtl
lynxis has quit [Read error: Connection reset by peer]
<TheLinuxBug> jrg: Got my Opi Plus 2E running android so far pretty slick... going to test other oses later :D
lynxis has joined #linux-sunxi
<jrg> TheLinuxBug: doesn't it come with android pre-installed on emmc?
<jrg> i'm not a big fan of android. i didn't even attempt to run it since i'm not using the box locally at all
cnxsoft has quit [Quit: cnxsoft]
<jrg> i went straigt to armbian once i unboxed it heh
<jrg> straight
zuikis has left #linux-sunxi [#linux-sunxi]
jernej has joined #linux-sunxi
zuikis has joined #linux-sunxi
<tkaiser> jrg: BTW: Did you use Armbian's nand_sata_install script before?
jernej has quit [Ping timeout: 276 seconds]
lynxis has quit [Read error: Connection reset by peer]
lynxis has joined #linux-sunxi
<jrg> tkaiser: i did yes
<jrg> when i first installed armbian i ran it and removed the SD and it boots
<tkaiser> jrg: And what happens, when you re-insert the SD and try to boot then? Also how das /proc/partitions looks like?
<jrg> i haven't tried tbh
<jrg> if you give me a minute i can put the SD back in and try it and see what happens
<tkaiser> jrg: Would be worth a try since that's exactly how this script is supposed to work: bootloader remaining on SD card and everything else loaded from somewhere else.
<jrg> sure i'll try it out. let me walk over there
<jrg> ok it's booting now iwth the SD back in
<tkaiser> jrg: Can you please post the output from 'sudo armbianmonitor -u' (uploads a log so I can get a clue what happened)?
<jrg> hm. nope. seems as though it still booted system off the SD
<jrg> all the changes i made to the emmc aren't there
<jrg> there you go
<jrg> going to pull this SD out. but yah it doesn't seem like it made the change to the SD bootloader to use the emmc as system
<tkaiser> jrg: Thx, the problem seems to be enumeration of SD card and eMMC so the modified boot.scr points back to the SD card instead to eMMC
<tkaiser> What does 'grep root /boot/boot.cmd' outputs?
<jrg> hm. think my opi just died on me lol
<jrg> wtf?
<jrg> setenv bootargs "console=tty1 root=/dev/mmcblk0p1 rootwait rootfstype=ext4 cgroup_enable=memory swapaccount=1 sunxi_ve_mem_reserve=0 sunxi_g2d_mem_reserve=0 sunxi_no_mali_mem_reserve sunxi_fb_mem_reserve=16 panic=10 consoleblank=0 enforcing=0 loglevel=${verbosity}"
<jrg> that is with the SD in
<tkaiser> jrg: Thx
<jrg> setenv bootargs "console=tty1 root=/dev/mmcblk0p1 rootwait rootfstype=ext4 cgroup_enable=memory swapaccount=1 sunxi_ve_mem_reserve=0 sunxi_g2d_mem_reserve=0 sunxi_no_mali_mem_reserve sunxi_fb_mem_reserve=16 panic=10 consoleblank=0 enforcing=0 loglevel=${verbosity}"
<jrg> that's without
<jrg> sd takes over mmcblk0 when inserted
<tkaiser> jrg: Will look into that later, thx. It should read mmcblk1p1 instead on the SD card now.
<jrg> sure. if i had 2 of them i'd let you just have your way with one lol. but i'm actually trying to configure this as my main shell box
Amit_t_ has joined #linux-sunxi
Amit_t_ has quit [Client Quit]
jernej has joined #linux-sunxi
<tkaiser> jrg: LOL, just looked through the code of the nand_sata_install script and it correctly detects the eMMC and then only the option to boot later from eMMC is considered. So script works as expected and this will a feature request.
jernej has quit [Ping timeout: 276 seconds]
<MoeIcenowy> is sun6i backlight pwm supported now?
<MoeIcenowy> in mainline
<wens> no
<wens> I have a sun6i pwm driver that may or may not be finished
<MoeIcenowy> wens: ah-oh
<MoeIcenowy> Thus the "Merged in 4.0" http://linux-sunxi.org/Linux_mainlining_effort shouldn't have "All SoCs: PWM driver"
jernej has joined #linux-sunxi
<tkaiser> jernej: Just a quick question: Do H3 boards with 2 GB RAM provide any benefit from OpenELEC perspective?
<jernej> tkaiser: I don't have one yet, but generally, I think not much. Except if you enable additional caching within Kodi. I got report that caching make things considerably faster.
<jernej> So, yes, if you play a bit, you can take advantage
<jernej> but I'm not sure how much
<tkaiser> jernej: Ok, is one OPi Plus 2E on its way to you already? And regarding Lite/One: Do they suffer from having only 512MiB?
<jernej> tkaiser: Not sure, I must talk to Igor. About Lite/One: They did suffer heavily in the past due to kswapd bug, but for now it's almost ok
scream has quit [Remote host closed the connection]
<wens> MoeIcenowy: the hardware is slightly different
<jernej> OpenELEC have whole FS in memory if board have 1 GB of RAM or more
<jernej> so they are a bit handicaped in this region
<jernej> and of course you can't have as much additional daemons running in background
<jernej> I think Kodi + other system takes around 350 MB of RAM
<tkaiser> jernej: Please talk immediately to Igor regarding a dev board. And good to know about the memory requirements.
<MoeIcenowy> wens: I know it, but this sentence is misleading
<jernej> tkaiser: I will. Do you want to do anything in particular with OpenELEC?
lamer14651421361 has joined #linux-sunxi
lamer14651421361 has quit [Client Quit]
tkaiser has quit [Ping timeout: 240 seconds]
tkaiser has joined #linux-sunxi
<tkaiser> jernej: Nope, just curious and trying to get 'the bigger picture' regarding the gazillions of H3 boards we now deal with
phipli has joined #linux-sunxi
phipli has quit [Remote host closed the connection]
phipli has joined #linux-sunxi
<jrg> tkaiser: is there gpu hw accel support for the H3?
<jrg> would be great to have oe for an opi+2e
<jrg> especially since on paper it would be way better than my rpi1 lol
apritzel has joined #linux-sunxi
iaglium has joined #linux-sunxi
mosterta has joined #linux-sunxi
tkaiser has quit [Ping timeout: 240 seconds]
tkaiser has joined #linux-sunxi
Zliba has joined #linux-sunxi
<tkaiser> jrg: What's this 'GPU support' everyone is talking about? If you ask in Pine64 forums for HW accelerated video decoding you get the answer this wouldn't be possible since Mali binary BLOBs would be missing.
<tkaiser> Not even normal BLOBs but binary ones ;)
naobsd has quit [Remote host closed the connection]
naobsd has joined #linux-sunxi
al1o has quit [Ping timeout: 240 seconds]
<jrg> haha
<jrg> then openelec would be moot wouldn't it?
al1o has joined #linux-sunxi
<tkaiser> Disclaimer: I'm an absolute noob regarding this. But AFAIK there's 2D, 3D and video decoding acceleration (different beasts). And _when_ you use an Armbian desktop image all 3 areas work (upgrading from a CLI image does not currently work -- since Xunlong is willing to donate some boards we try to get a few more devs on board: http://forum.armbian.com/index.php/topic/1325-claim-a-task-and-do-it/
<tkaiser> Hw accelerated video decoding works pretty well (and is unrelated to 'GPU' -- that's Mali and responsible for 3D acceleration): http://linux-sunxi.org/Cedrus#Supported_codec_matrix
<tkaiser> You have to say thank you to the great linux-sunxi devs here! Then we ship with a current fbturbo for 2D acceleration and users reported 3D acceleration would also work on H3 (gaming, we got some Quake 3 scores reported and I've absolutely no idea whether they're good or not)
apritzel has quit [Ping timeout: 244 seconds]
Mr__Anderson has joined #linux-sunxi
jernej has quit [Ping timeout: 276 seconds]
fredy has quit [Excess Flood]
fredy has joined #linux-sunxi
jernej has joined #linux-sunxi
<jrg> OK... i think i actually may have samba/cifs mouns working
<jrg> although it required the use of the nounix flag :/
<jrg> going to test it out with a few files to see if they rm correctly
<KotCzarny> its just debian jessie
<KotCzarny> how do you like opi+2e?
<jrg> not sure yet tbh. i'm still setting it up
<jrg> the banana pi isn't too bad either once you accept 1. it is not a true octocore 2GHz board 2. what you get is what you get with images.. pray pkgs will be updated forever
<jrg> lol
<jrg> once i sort out this opi i'm going to move the bpi over as a web server and see how it handles that
<jrg> i'm still kind of concerned about having the opi out in the open like it is tho... no cases for it yet. at all.
<willmore> Is there a .STL file so one could be 3d printed? :)
<jrg> not sure. doesn't matter tho. i don't have a 3D printer lol
* willmore needs to learn to design in some CAD program this summer.
<willmore> jrg, what continent are you on?
<jrg> North American
<willmore> I would gladly print you one.
<willmore> The Pine64 case printed quite nicely.
<willmore> I just need a ton of screws. Hopefully those will arrive from China shortly.
<willmore> Is the +2E the same size as any other Opi board?
<willmore> One could always dremel out the extra connectors....
<KotCzarny> jrg: make a case for it yourslef then?
<KotCzarny> i made me a case for opi+2e + opipc + 50W psu
<KotCzarny> all it takes it a drill and few screws+screw mounts
<KotCzarny> and ANY plastic/metal box
<KotCzarny> and 1-2h
<jrg> 50W psu?? LOL
<KotCzarny> willmore, its bigger, but comparable to opi2 i think
<jrg> i was hoping the opi had the holes in the same place as the bpi
<KotCzarny> jrg: opi+2e has ~3A max power and opipc has ~2A, which you can reach without much troubles
<jrg> since i have half of a bpi case lol
<jrg> KotCzarny: just thought 50W was a bit overkill
<KotCzarny> nope
<jrg> KotCzarny: i considered drilling holes in my nas plate and seeing if i can find something run run the connections to open slots
<jrg> ie: a computer in a computer
<KotCzarny> because add some usb devices or ssd and you can fill it in a blink
<KotCzarny> and because its better to calc the power usage with some margin, than to fry something
<jrg> KotCzarny: figure all i need to do is get a few extensions to USB, lan, and barrel connector
<jrg> and drill a couple holes.. then i can have everything in 1 case lol
<KotCzarny> barrel is quite common, simple 4.0/1.7
<jrg> KotCzarny: well all i'd need are extension cables that can reach an empty card slot
<jrg> i wonder how many i can fit in the case. lol
<KotCzarny> :)
<jrg> my newer nas is in that old thermaltake case i had in my parents' attic for like 3 years lol
<KotCzarny> here you have comparison of the sizes
<jrg> i bet i can drill a few holes and fit like 10 "pi" based boards in it
<KotCzarny> plenty of space
<jrg> would be interesting to see if i can put like 5-10 opis in it with extension cables going to card slots in the back
<jrg> since i'm only using 1 slot for the M1015
mosterta has quit [Ping timeout: 246 seconds]
<jrg> then i can just have a nas with a bunch of network cables going to the slots... i'll probably look into that later.
<KotCzarny> i will take photos of my dual opi, 8 core monster ;)
<KotCzarny> which has space probably for another 2 boards
<KotCzarny> (i though about leaving space for the disk in case i wanted to add it later)
<willmore> KotCzarny, the mounting holes and the peripherial layout are completely different. ;( Cases will not interoperate.
<KotCzarny> willmore: just redrill the holes?
<KotCzarny> i started with generic abs box
<willmore> And that reminds me, now I remember what I forgot to put in my last Ali order. :( Barrel connectors for Opi1 and Odroid-C2
<KotCzarny> :)
<willmore> KotCzarny, they're mounts from the bottom case. I'm thinking in the 3D printing case, not the generic box one.
apritzel has joined #linux-sunxi
<KotCzarny> edit out the holes and leave them for the user?
<KotCzarny> maybe make it modular, so some break-out fields for connectors
<willmore> Well, the offer stands. If you're in NA and can find a case design on some place like Thingiverse, let me know.
<willmore> breakouts don't print well.
<willmore> Generally easier to just have different designs. It's not like you have tooling costs. ;)
<tkaiser> I build enclosures from wood only ;) http://kaiser-edv.de/tmp/fkgs4d/ (Lamobo R1 with 2 PSUs for LCD display and a couple of PoE powered Raspberry Pi)
<KotCzarny> well, you can make simple 'case' with just 2 plates and 8 mount holes
<KotCzarny> i must make a photo of my bpi-r1 too
<KotCzarny> (mounted inside desktop pc case)
iamfrankenstein1 has joined #linux-sunxi
<KotCzarny> with psu, switch and liion battery ;)
<KotCzarny> and 2 drives
<KotCzarny> in general, pc cases are nice to repurpose with some tinkering, as they are usually made from metal and can be ventilated well
<willmore> I'm going to repurpose an old PC case as an external optical drive chasis.
apritzel has quit [Ping timeout: 244 seconds]
<KotCzarny> willmore, is there some scripting tool to create 3d printing layouts? ie. create x*y*z plate, add rectangular x*y hole at x*y
<KotCzarny> should work pretty well for designing new cases
IgorPec has joined #linux-sunxi
lvmc has joined #linux-sunxi
<willmore> KotCzarny, there is OpenSCAD which is a CAD tool with a kind of scripting/programming language that can be used to create designs.
<lvmc> tkaiser: hello
<lvmc> IgorPec: hello
<lvmc> How are you guys?
<IgorPec> hey
<willmore> LVMC? Is that proper Roman numerals?
<lvmc> No. It is from my name initial letters.
<willmore> ;)
<lvmc> Luiz Vitor Martinez Cardoso
<lvmc> LVMC ;)
<tkaiser> lvmc: Hola
<willmore> Need more middle names. ;)
iaglium has quit [Quit: ZNC 1.7.x-nightly-20160225-9b31a077 - http://znc.in]
<KotCzarny> and profession names
<IgorPec> mr igor von p :)
<lvmc> tkaiser: I was having a great convesation with IgorPec now about M2P.
<lvmc> He sent me a link with some information about serious issues on thermal.
<lvmc> I'm running a very important project based on M2P and want to support you do solve the issues.
<lvmc> Or at least to understand what is going on.
<tkaiser> lvmc: 'M2P' as in BPi M2+?
<lvmc> Yes.
<willmore> Not the consulting company out of Frankfurt?
<tkaiser> lvmc: Forget about, threw the board out of the window. A cat ate it already
<lvmc> I'm also looking for a solution for OV5640.
<KotCzarny> or dunk in oil case
mosterta has joined #linux-sunxi
<longsleep> use it as paper weight
<lvmc> tkaiser: I have a commercial project running on it, it was a terrible surprise to me.
<willmore> KotCzarny, big vat of Flourinert?
<tkaiser> lvmc: LOL, already imagined something like that ;)
<KotCzarny> willmore: worked out car gunk should work too
<lvmc> What board do you recommend to replace M2P?
<lvmc> I was looking at M3 from FriendlyARM.
<KotCzarny> lvmc, forget bananas
<tkaiser> lvmc: Please, why does anyone choose such a thing like BPi M2+? What's the reason even thinking about it?
al1o has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]
<KotCzarny> tkaiser: out of not knowing about the alternatives?
<lvmc> tkaiser: size and because it has NAND.
<lvmc> What do you think about NanoPi_M3 board?
<tkaiser> lvmc: Can you read on CNX already ;)
<KotCzarny> i think nanopi is the pne with crippled regulator
<tkaiser> KotCzarny: There exist a couple of different NanoPi. M1 is a clone of OPi PC, M3 something different
al1o has joined #linux-sunxi
<tkaiser> lvmc: Have you seen the HUGE heatsink for the bigger NanoDings with the same SoC? I wouldn't think about the M3 since there you can't attach a heatsink easily
<KotCzarny> hmm, no nano pis on sunxi wiki?
<lvmc> So, what is the option with ETHERNET + WIFI upto 65x65 size?
<tkaiser> KotCzarny: Sure, I add only boards I have here
<KotCzarny> lvmc, you've said about commercial application, is stability the thing you seek?
<tkaiser> KotCzarny: And there seem to be no other linux-sunxi people interested in this board so far (since the only intersting thing would be the different camera module)
<lvmc> Define stability.
iamfrankenstein1 has quit [Quit: iamfrankenstein1]
<tkaiser> lvmc: Not really kidding but since you seem to need a working camera solution maybe the new RPi Zero version (with camera connector but without everything else ;) )?
<KotCzarny> lvmc, using board without the crashes during high load (if your project requires much of the cpu)
<KotCzarny> (crashes due to insufficient power or overheating)
<lvmc> That is a long history... but I need a quad-core CPU.
<lvmc> It runs computer vision algorithms.
<KotCzarny> i think you need a bigger board or bigger heatsink then
<KotCzarny> and a fan maybe
<lvmc> My restriction is 65x65.
<lvmc> I've been running M2P for 2 months with 100% without any heating problem.
<lvmc> What is really going under the scenes?
<tkaiser> lvmc: At 240 MHz?
<KotCzarny> lvmc, it probably dropped freq and cores
<KotCzarny> or you just didnt use that much cpu
<lvmc> I don't know, I've just generated the .raw image from Armbian lib repo 2 weeks ago.
<tkaiser> lvmc: Huh? 2 weeks or 2 months?
<KotCzarny> tkaiser, does armbian monitor show freq/cores ?
<lvmc> I've been running M2P for 2 months, but Armbian is running for 2 or 4 weeks.
<lvmc> Don't remember exactly.
<KotCzarny> lvmc, is it still running? did you check the dmesg or logs?
<tkaiser> lvmc: Did you run 'sudo armbianmonitor -u'? Installs RPi-Monitor and then you get funky graphs showing you what has been killed already. If you used SinoVoip images before you used a single core system. They used Allwinner's defaults which lead to CPU cores being killed almost immediately when you start to think about using the board
<tkaiser> lvmc: Since the only other confirmation of horrible thermal behaviour is Igor please do the above mentioned step and start monitoring.
IgorPec has quit [Ping timeout: 264 seconds]
<lvmc> root@bananapim2plus:~# armbianmonitor -m Stop monitoring using [ctrl]-[c] Time CPU load %cpu %sys %usr %nice %io %irq CPU 19:18:22: 1008MHz 0.04 0% 0% 0% 0% 0% 0% 44°C 19:18:27: 240MHz 0.03 0% 0% 0% 0% 0% 0% 44°C 19:18:32: 240MHz 0.03 0% 0% 0% 0% 0% 0% 44°C 19:18:37: 240MHz 0.03 0% 0% 0% 0% 0% 0% 44°C 19:18:43: 240MHz 0.03 0% 0% 0% 0% 0% 0
<lvmc> sorry!!!!!
<KotCzarny> now make it do some work and wait few minutes?
<tkaiser> stress -c 3 -m 3
zuikis has left #linux-sunxi [#linux-sunxi]
<KotCzarny> tkaiser, thing is it should do the work that it will be doing normally
<KotCzarny> maybe workload is light
<lvmc> cat loop.c int main() { for (;;) {}}
<lvmc> running this
<lvmc> It is running
<lvmc> 1.2Ghz
<KotCzarny> temp?
<lvmc> 48
<lvmc> 19:23:01: 1200MHz 0.86 1% 0% 0% 0% 0% 0% 48°C
<KotCzarny> one core or 4 in top?
<tkaiser> lvmc: Pretty lightweight. But again: With SinoVoip settings you had a single core system when doing anything a bit challenging
<KotCzarny> you can press 1 in top to show cores load
<tkaiser> And please forget about armbianmonitor and install RPi-Monitor, takes 5 minutes and you get a lot more information. And it's that easy using armbianmonitor -r
<lvmc> installing
<tkaiser> lvmc: So your workloads are single-threaded and you don't need a quad core system ;)
<lvmc> no
<lvmc> I'm just running a simple loop
<KotCzarny> then load it as you would in the target case
<lvmc> My application is multthread.
<KotCzarny> if you want simple cpu stress, then use stress command
<tkaiser> lvmc: Remember: In the middle BPi M2+ with heatsink: http://kaiser-edv.de/tmp/Gn4wTB/Bildschirmfoto%202016-05-25%20um%2018.14.54.png
<lvmc> I'm using a small aluminun 14x14 heatsink
<tkaiser> lvmc: And since Igor confirmed today the shitty thermal behaviour time to not think about this board any longer :)
<lvmc> how can I use stress to test the 4 cores?
<tkaiser> lvmc: Sure, but even a smaller OPi Lite (on the right in the screenshot and without heatsink) performs WAY better
<tkaiser> stress -c 4
<tkaiser> But just use 'stress -c 3 -m 3' to see BPi M2+ vanishing away :)
<lvmc> rpi installed
<lvmc> you can access it from this url
<jrg> heh
<KotCzarny> cpu freq 240mhz already
<KotCzarny> hehe
<jrg> my bpi m3 with a heatsink even throttled to 1.2GHz always uses 4 cores
<jrg> it's gotten to the point where it runs hot doing nothing
<tkaiser> KotCzarny: These are idle settings, on BPi M2+ we allow downclocking to 240 MHz
<KotCzarny> tkaiser, it showed load: 3 in that time
<lvmc> back to 1008
<lvmc> back to 240
<lvmc> 240
<lvmc> 408
<tkaiser> When do people start to understand average load in Linux? Most probably NEVER
<KotCzarny> tkaiser, and cpu usage was also high
<KotCzarny> anyway, fire up the stress
<jrg> 14:32:00: 1200MHz 3.39 58% 10% 31% 6% 4% 5% 70°C
<lvmc> running it again
<tkaiser> KotCzarny: 'Cooling State' is what you should be interested in since everything else is pretty irrelevant
<KotCzarny> it started to throttle
<tkaiser> lvmc: Do you run stress or the lightweight loop that does exactly nothing?
<KotCzarny> ;)
apritzel has joined #linux-sunxi
<lvmc> no
<lvmc> stress -c 3 -m 3
<KotCzarny> go for -c 4
<lvmc> c4 m3?
<lvmc> c4 m4
<lvmc> doing
<jrg> damn. i totally forgot bitcoind is running on the bpi as a daemon heh
<lvmc> running
<KotCzarny> -c, --cpu N
<KotCzarny> spawn N workers spinning on sqrt()
<KotCzarny> -m, --vm N
<KotCzarny> spawn N workers spinning on malloc()/free()
<jrg> at least i finally found a pam_mount setting for mount.cifs that might actually function properly with samba+debian
<jrg> albeit a bit awkward of a method
<lvmc> stress reports
<lvmc> FAIL failed run completed in 19s
<jrg> lvmc: what aare you running the stress test on?
<tkaiser> lvmc: Yes, that's why stress is not the tool of choice when CPU cores get killed (that was lesson No 1 ;) )
<tkaiser> Better try 'sysbench --test=cpu --cpu-max-prime=200000 run --num-threads=4'
<lvmc> HOW to test?
<lvmc> running
<lvmc> are you checking the RPIm?
<KotCzarny> but, if you saw core killed, then imagine what would happen with real application
<lvmc> 912Mhz.
<lvmc> 100% loaded cpus
<tkaiser> H3 needs a few minutes to heat up
<lvmc> 62
<lvmc> oC
IgorPec has joined #linux-sunxi
<lvmc> Is it the expected behaviour?
<lvmc> It is running...
<lvmc> 64ºC
<lvmc> 66
<lvmc> what is the maximum?
<lvmc> 70?
<tkaiser> lvmc: This looks better than on my BPi M2+. But you're using pretty conservative settings (not your fault, just the result of us trying to improve this stuff currently)
<lvmc> let's do the worse test
<lvmc> just tell me how
<tkaiser> lvmc: git clone https://github.com/ssvb/cpuburn-arm
<tkaiser> cd cpuburn-arm
<tkaiser> gcc cpuburn-a7.S
<tkaiser> ./a.out
<tkaiser> Enjoy
<lvmc> running
<lvmc> are you checking the rpim?
<lvmc> 74!
<lvmc> 76
<lvmc> what is the maximum to not damage?
<tkaiser> lvmc: 100*C are pretty safe
<lvmc> cpu freq is scalling
<tkaiser> But the THS settings you're using are conservative
<lvmc> I'm using the THS from Armbian
<tkaiser> lvmc: Yeah, from two weeks ago ;)
<lvmc> guys
<lvmc> What the best we can do for this board?
iamfrankenstein has quit [Quit: iamfrankenstein]
<lvmc> What is the best we can do?
<tkaiser> lvmc: Ignoring it?
<lvmc> If you already had spent thousand dollars it is not an option.
<lvmc> I can change the board to M3, for example or other that is near 65x65.
<lvmc> But how can I help you
<lvmc> to let armbian support M2P the best it can?
<tkaiser> lvmc: No idea. But if you could grab https://raw.githubusercontent.com/igorpecovnik/lib/master/config/fex/bananapim2plus.fex and then use fex2bin to create a script.bin to replace /boot/script.bin, then reboot and let cpuburn-a7 run for at least 15 minutes we could an idea whether both Igor's and my BPi M2+ are somewhat broken (regarding thermal issues) or not
<lvmc> ok
<lvmc> wait
<lvmc> root@bananapim2plus:~/sunxi-tools# ./bin2fex m2p/bananapim2plus.fex E: fexc-bin: Malformed data: version 1886154596.762210664.1646276976.
<NiteHawk> fex2bin, not bin2fex ;)
<lvmc> sorry
<jrg> blah. waiting for something to download before i can move everything off the bpi to the opi
<jrg> kind of excited
<lvmc> can I just copy new script to /boot?
<lvmc> give 777?
tkaiser has quit [Remote host closed the connection]
tkaiser has joined #linux-sunxi
<lvmc> tkaiser: done
<lvmc> it is running
<tkaiser> lvmc: Needs at least 10 min, better 15
<lvmc> ok
<lvmc> are you seeing the results on http://cbed6fc1.ngrok.io ?
<lvmc> 79ºc
<tkaiser> lvmc: Yes, looks already terrible compared to any OPi but not that bad compared to my M2+
<tkaiser> Already ar 648 MHz, so maybe I was wrong and it's as terrible as with Igor's and my BPi M2+
<jrg> what cpu does an m2+ use?
<lvmc> h3
<jrg> oh
<IgorPec> looks better than mine, yes
<tkaiser> jrg: Two differences: Different voltage regulator and therefore always at 1.3V (1.2V according to wens/schematic) and obviously the PCB makes a HUUUUGE difference
JohnDoe5 has joined #linux-sunxi
<lvmc> what is the conclusion?
<lvmc> Is it sooo terrible?
<tkaiser> lvmc: If the name contains Banana then check whether the SoC is A20 (good) and otherwise run away as fast as you can?
<lvmc> Why are you putting this board on armbian website?
<tkaiser> lvmc: How does your heatsink looks like?
<lvmc> Armbian is a respectable project.
<lvmc> By serious guys.
<lvmc> If you support this board... I can't agree that it is a shit
<tkaiser> lvmc: Since we wasted an insane amount of time already with this board and since users buy it anyway since they have no clue and then at least we could try to provide software that sucks less [TM]
<jrg> there is an official armbian image for an M2+ ? the only image i was able to find for my M3 was the pseudo one that tkaiser made
JohnDoe_71Rus has quit [Ping timeout: 276 seconds]
<jrg> and even then it had the kernel caveats
<tkaiser> jrg: M2+ is a H3 board trying to copy as much as possible from OPi PC/Plus. It was easy to add support just to realize now how poor the design is
Netlynx has quit [Quit: Leaving]
<lvmc> It is a 14x14 aluminum
<lvmc> I will provide you a photo
<tkaiser> lvmc: Running cpuburn-a7 your BPi M2+ seems to throttle down to just 648 MHz (which is quite good and maybe related to the heatsink).
<lvmc> I'm using this one
<lvmc> running for ~10min
<tkaiser> lvmc: Yeah, that's enough. And you heatsink _looks_ not that efficient. If you want help us you could grab a 4GB SD card and do this test as last resort: http://forum.armbian.com/index.php/topic/1322-testers-wanted-testing-dram-reliability-on-bpi-m2/
mpmc has quit [Ping timeout: 276 seconds]
<ssvb> jrg: there are also other sets, which include power supplies, cables, etc.
<tkaiser> (Again installing RPi-Monitor before to get a clue about the thermal behaviour when running lima-memtester)
<lvmc> downloading
<jrg> ssvb: oh. i didn't see that.
<jrg> maybe because it's in spanish? heh
<tkaiser> jrg: That's not spanish ;)
<jrg> that's not spanish...
kelvan has quit [Remote host closed the connection]
<jrg> at first glance it looked spanish until i saw Laranja
kelvan has joined #linux-sunxi
<ssvb> jrg: aliexpress sometimes switches languages pretty randomly, I have no idea why it happens but you can switch it back to English
<jrg> yah i just did. ugh.. if i order this tho i'll have to wait weeks for it again :)
<ssvb> jrg: regarding the cases for opi+2e, they have only added them recently to their shop
<jrg> yeah
<jrg> they weren't there when i bought the 2e .. i was looking to order with it
<lvmc> tkaiser: Armbian_5.14_Bananapim2plus_Ubuntu_xenial_3.4.112_desktop.7z
<lvmc> is it right?
<tkaiser> lvmc: Yes
<lvmc> downloading
<lvmc> tkaiser: can I stop the previous test?
<tkaiser> lvmc: Then just execute lima-memtester as outlined in the thread. Will run with 648 MHz DRAM clock but that's not important since the throttling behaviour is more interesting. Igor's M2+ clocked down to 240 MHz with just 2 CPU cores remaining and mine killed just one CPU core (but it takes 10-15 minutes)
<tkaiser> lvmc: Sure
<jrg> ssvb: well i guess i'll grab a couple of them
<jrg> in case i break the first one like i did with the bpi case heh
<lvmc> Ok, can I run this new image from sdcard or need to flash to nand?
<tkaiser> lvmc: SD card
<lvmc> dding to sd
<jrg> Order Processing Time: 7 Days
<jrg> :/
<tkaiser> jrg: When I tried to drill in a hole fore different DC-IN connector in an RPi enclosure to be used with an 'UP board' (Intel Atom in RPi size) it took me just 2 seconds to destroy the acrylic enclosure :)
<jrg> tkaiser: i snapped it trying to shove in a wifi antenna.. which in hindsight was a bit pointless
tkaiser has quit [Read error: Connection reset by peer]
iamfrankenstein has joined #linux-sunxi
<jrg> ssvb: is that antenna part supposed to justlay on the board like that?
<ssvb> jrg: I don't know, I'm only using ethernet on my boards :-)
<jrg> i am too but i'd rather mount it properly if possible. never know when it may be repurposed
mpmc has joined #linux-sunxi
<lvmc> tkaiser: booting
JohnDoe5 has quit [Quit: KVIrc 4.9.2 Aria http://www.kvirc.net/]
tkaiser has joined #linux-sunxi
<lvmc> tkaiser:
TheSeven has quit [Disconnected by services]
<lvmc> as soon I login it reboots
[7] has joined #linux-sunxi
<lvmc> it is working now
vagrantc has joined #linux-sunxi
<lvmc> lima test is running
<jrg> wow pretty neat
<lvmc> tkaiser: I'm seeing a cube
<lvmc> no visual artifacts
IgorPec has quit [Ping timeout: 240 seconds]
<jrg> it's down to 640MHz?
<jrg> lol
<lvmc> 648
<lvmc> jrg: are you checking it too?
<ssvb> tkaiser: btw, about the lima-memtester tests, did you see tearing when running it on your armbian kernel?
<jrg> sorry. wasn't trying to cheat you out of that 8MHz :)
<lvmc> 20:46:40: 648MHz 1.81 44% 11% 29% 0% 0% 3% 82°C
<jrg> how are you doing that monitoring thing?
<jrg> that's pretty neat
<tkaiser> ssvb: Tearing? I would say yes.
<jrg> oh my. cooling state 4
<ssvb> tkaiser: there is one tricky thing, if the vsync feature is enabled, then the frame rate is limited to 60Hz and you get tearing free updates
<tkaiser> lvmc: After 10 minutes still at 648 MHz seems pretty good for BPi M2+
<ssvb> tkaiser: but it is not intensive enough as a memory tester
<lvmc> ;)
<lvmc> so what is the trick?
<lvmc> the heatsink?
<tkaiser> ssvb: Hmm... well, I don't really understand this graphics stuff at all (being a server guy running all devices headless and not even knowing how they look like ;)
mpmc has quit [Ping timeout: 260 seconds]
<ssvb> tkaiser: I disabled the automatic vsync feature via this patch - https://github.com/ssvb/linux-sunxi/commit/d4662ee456eef8873c23f1400c60b93c0ca835eb
<tkaiser> jrg: sudo armbianmonitor -r (prefix it with 'time ' and please report back)
<KotCzarny> memory bandwidth is important for firefox for example, or media en/decoding
<lvmc> it is running perfectly
<lvmc> 312Mhz now
<lvmc> 85ºC
<lvmc> 648
<tkaiser> ssvb: So would it be a requirement to try to apply this patch to the new BSP kernel sources?
<ssvb> tkaiser: vsync makes sense for games, but it is bad for tests and benchmarks
<tkaiser> ssvb: Please understand that I've no idea what vsync is :)
<ssvb> tkaiser: also you said that you tasted the standalone lima-memtester on the opi-pc board, was it showing the same results as the FEL based test?
<KotCzarny> vsync == eye pleasure, but needs more respurces
<tkaiser> ssvb: On OPi PC and PC Plus absolutely identical results
<ssvb> tkaiser: the trick is that you monitor refreshes the picture 60 times per second, so if your accelerated application renders 300 frames per second, a lot of this work is a pure waste
<lvmc> tkaiser: do I need to use the deb packages and try other frequencies different from 648??
<ssvb> tkaiser: that's why graphics drivers normally synchronize the frame rendering with screen updates, and as a bonus this also eliminates the tearing effect (when a part of the screen is showing the current frame and a part of the screen is showing the next one)
<tkaiser> lvmc: For me (and the thermal stuff) it's already enough. Your BPi M2+ performs not that shitty as Igor's and mine. If you want to further waste time with this board you could try increase DRAM clockspeed and test again
<lvmc> Ok
<tkaiser> ssvb: Partially understood (please be aware that I'm such an ignorant regarding everything that is drawn on a display :) )
<lvmc> But what is the Armbian final decision?
<ssvb> tkaiser: anyway, since you said that you get no difference between FEL results and normal runs of lima-memtester from the OS on your opi-pc, then this was probably a wrong lead again :)
<tkaiser> lvmc: None
<lvmc> Can I use the last commit and it will be stable as it has been showing?
<tkaiser> lvmc: Sorry, which commit?
<lvmc> master
<ssvb> tkaiser: google says that this might be a relevant article on the vsync subject :) http://www.anandtech.com/show/2794/2
<lvmc> I mean, build M2P images from master.
phipli has quit [Ping timeout: 264 seconds]
<lvmc> It is the best we can have with M2P, right?
<tkaiser> ssvb: But that's GUI stuff! Who needs this?! ;) Really I don't use Linux with GUI. Only by accident or to test out stuff I don't understand or care of
<tkaiser> lvmc: We need still some more results regarding the thermal behaviour since I still can't believe that an OPi Lite running the same task with identical settings is able to perform at +1000 MHz while a BPi M2+ clocks down to 312 MHz
<lvmc> How can I help you find that answer?
<tkaiser> lvmc: Get a few more boards and/or testers and let them run the tests we walked through this evening?
<ssvb> tkaiser: anyway, the idea was that, let's say, 500 fps spinning textured cube is likely much more stressful for the DRAM, when compared to the 60 fps spinning textured cube :-)
<lvmc> but my point is
<lvmc> it downs to 312Mhz
<tkaiser> ssvb: Ahhh, now I get the point. Then it might be important to apply the patch but since my other tests with OPis showed no differing behaviour I just want to go to bed now ;)
<lvmc> but it back to ~1.2Ghz
<lvmc> as soon the temperature downs, right?
<ssvb> tkaiser: probably we just need a fps counter in lima-memtester :-)
<tkaiser> lvmc: With Armbian kernel you even get killed CPU cores back. Mikhail patched the budget cooling stuff and now we have a better behaviour even in the kernel code
<tkaiser> lvmc: Before when overheating occured the kernel killed CPU cores and they never came back until a reboot occured
<lvmc> Ok.
<lvmc> tkaiser: What can I do to help you do the best fro armbian to support this board?
ricardocrudo has joined #linux-sunxi
<lvmc> Guys has anyone been trying to support OV5640 on M2P?
<tkaiser> lvmc: There's nothing from an 'Armbian perspective'. We tried to get a clue how to solve thermal issues and that's it. As soon as mainline kernel OS images will be available the whole BPi M2+ 'problem' will be resolved automagically since running mainline kernel each and every BPi M2+ will kill itself.
<jrg> how is armbian controlling swap?
<jrg> was curious if i could increase it a bit
<lvmc> Is it planned for 5.17?
<tkaiser> Megi's THS stuff might work but since no one else is looking into it or pushing it upstream running mainline kernel on BPi M2+ will definitely kill H3 :)
megi has joined #linux-sunxi
<megi> I'm trying to update that branch right now :) If anyone is trying to test that, you'll also need u-boot patch https://github.com/megous/u-boot/commit/49c833f38e71fa21e6870a8057214a476a8ec9ca
<megi> Otherwise it will lock up on boot for some reason
<lvmc> both for oranges and m2p we still have ov5640 not working
<lvmc> that is the error
phipli has joined #linux-sunxi
<willmore> lvmc, is that that camera that the orange pi people sell?
<lvmc> yes
<willmore> I've pondered getting one. I have no need for it, though.
<tkaiser> megi: When I played with this stuff a few weeks ago I failed. Patches applied cleanly but then OPi PC (IIRC) didn't boot any more. And I'm too dumb to get a clue why (at least I was somewhat lost between u-boot and kernel)
<megi> tkaiser: I remember that you wrote something about that on github.
<tkaiser> megi: BTW: IIRC your OPi One died? Do you have an interest in getting a Lite instead?
iamfrankenstein has quit [Quit: iamfrankenstein]
<agraf> ssvb: if you're bored, it'd be nice to have something like this for sunxi devices too: https://patchwork.ozlabs.org/patch/630589/
<megi> tkaiser: I'm not sure what actually happend, but it came back after a while
<megi> tkaiser: Do you know if lite has the same regulator setup as OPi One? Meaning 1.1/1.3V switching using gpio pin?
<tkaiser> megi: Yes, it's the same. And PCB is a bit better/thicker (maybe containing a copper layer) so that heat dissipation _might_ be improved. But since my OPi One definitely has a broken thermal sensor no real comparisons are possible
<megi> I wonder how many variants of OPi will be out there by the end of the year. I guess somewhere between 30-500
<megi> Hmm, isn't the thermal sensor part of H3... the same chip used on OpiPC? It doesn't make much sense why it would work on opi pc and not on opi one
<tkaiser> megi: NanoPi M1 (close clone of Orange Pi PC/One) also uses the same regulator. But until now no one confirmed the settings we use (not even FriendlyARM themselves: https://github.com/friendlyarm/h3_lichee/issues/2#issuecomment-218707471 )
<tkaiser> megi: I tried to fry my Orange Pi One several times until I realized that the GPIO header is rotated by 180 degree. Maybe that is the culprit? Since I tried to provide power through (wrong) GPIO pins
<tkaiser> megi: Anyway: Try to push your patches upstream. Otherwise many BPi M2+ will die ;)
<megi> I'll try to save them! :D
<jrg> hm. that's odd
<jrg> i can't edit files with nano getting permission denied
<jrg> but i can edit them with vi heh
<jrg> samba ftw
<lvmc> tkaiser: sorry, I usually don't develop OS level things.
<lvmc> Are some patches from megi for M2P?
<tkaiser> lvmc: With 'legacy' kernel we have thermal protection (pretty shitty with Allwinner's defaults but at least it works). With mainline kernel w/o megi's patches there's nothing.
bonbons has quit [Ping timeout: 250 seconds]
<lvmc> What kernel is used by Armbian by default? legacy or mainline?
mpmc has joined #linux-sunxi
<tkaiser> lvmc: When I got my BPi M2+ a few weeks ago I thought it would be ok to limit CPU clockspeed to 816 MHz when we have no THS stuff working. But it seems that's way too much.
<tkaiser> lvmc: uname will tell ;) We don't provide mainline OS images at the moment until THS is working.
<jrg> nice. got the hard part of this orange pi working :)
<megi> when i was testing opi one with fixed 1.3v and variable frequency, the only safe frequency was somewhare around 320mhz
<tkaiser> But in case you want to fry your M2+: http://linux-sunxi.org/Banana_Pi_M2%2B#OS_images (everything's working except of throttling)
<megi> with that i could max out all cores and still keep temperature without cooling around 85°C
<lvmc> megi what your patches do?
<megi> several things:
<megi> - add thermal sensor driver, to be able to read core temperature
<megi> - add dts cofiguration for operating points (dynamic voltage/frequency scaling)
<ssvb> agraf: I'll have a look at it, thanks
<megi> - link the two together with thermal_zone setup in dts
<megi> - add driver for voltage regulator used on orange pi pc
<megi> that's about it
<vagrantc> i've been running an orange pi plus2 off of linux v4.7-rc1 ... but it seems a little less stable than the 4.4.x kernel + some patches from ssvb's branch ... is that likeyl due to overheating issues?
<megi> - also dts setup for gpio based voltage switching used on opi one
libv_ has quit [Ping timeout: 276 seconds]
libv has joined #linux-sunxi
<tkaiser> megi: Please push these patches ASAP since without them H3 with mainline kernel simply is asking for troubles
<lvmc> tkaiser: who is the best guy on cameras on armbian community?
<megi> tkaiser: I'm working on it, I'll hopefully send them out to ML after testing with the new kernel
<lvmc> I suspect we have an error on .fex for OV5640
<tkaiser> Especially when we're talking about BPi M2+ -- OPi Plus 2E instead shows impressive thermal behaviour: https://github.com/igorpecovnik/lib/issues/298#issuecomment-223659625
<tkaiser> lvmc: No idea. I'm pretty clueless as usual and just interested in getting camera stuff working.
<lvmc> Ok
<lvmc> Your testing is running here.
<lvmc> No artefacts
jrg has quit [Quit: ZNC - http://znc.in]
libv_ has joined #linux-sunxi
reinforce has quit [Quit: Leaving.]
libv has quit [Ping timeout: 240 seconds]
<tkaiser> vagrantc: Without megi's patches we'll never know.
jrg has joined #linux-sunxi
<jrg> almost done heh
<jrg> just need to build bitcoind on the orange pi :)
<tkaiser> Seriously: H3 is prone to overheating so running a combination of u-boot/kernel that doesn't prevent that is weird.
<tkaiser> So either limit cpufreq to something pretty low from within u-boot or expect strange things
jrg has quit [Client Quit]
libv has joined #linux-sunxi
libv_ has quit [Ping timeout: 264 seconds]
tkaiser has quit [Quit: jIRCii - http://www.oldschoolirc.com]
megi has quit [Quit: megi]
phipli has quit [Ping timeout: 258 seconds]
libv_ has joined #linux-sunxi
lvmc has quit [Quit: Page closed]
libv has quit [Ping timeout: 244 seconds]
jernej has quit [Ping timeout: 244 seconds]
libv_ has quit [Ping timeout: 240 seconds]
jrg has joined #linux-sunxi
mosterta has quit [Ping timeout: 276 seconds]
<vagrantc> hrm... [ 4.011975] cpufreq-dt: probe of cpufreq-dt failed with error -2
<vagrantc> course, it's the same on the other boards running 4.4.x + patches
ricardocrudo has quit [Remote host closed the connection]
libv has joined #linux-sunxi
gzamboni has quit [Read error: No route to host]
al1o has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]
<jrg> ugh
<jrg> ran into this samba issue because i had to get rid of the nounix flag
<jrg> no idea what is up with freenas smb4 but something is broken when using mount.cifs.. i need to make a vm and see if it's an arm debian thing
jrg has quit [Quit: ZNC - http://znc.in]
jrg has joined #linux-sunxi
Mr__Anderson has quit [Remote host closed the connection]
libv_ has joined #linux-sunxi
libv has quit [Ping timeout: 264 seconds]
libv has joined #linux-sunxi
libv_ has quit [Ping timeout: 240 seconds]
libv has quit [Ping timeout: 244 seconds]
tsuggs has quit [Quit: No Ping reply in 180 seconds.]
tsuggs has joined #linux-sunxi
tsuggs has quit [Quit: No Ping reply in 180 seconds.]
libv has joined #linux-sunxi
tsuggs has joined #linux-sunxi