Werner changed the topic of #armbian to: armbian - Linux for ARM development boards | www.armbian.com | Github: github.com/armbian | Commits: #armbian-commits | Developer talk: #armbian-devel | Forum feed: #armbian-rss | Type 'help' for help | Logs: -> irc.armbian.com
tparys has joined #armbian
<tparys> evening all
<stipa> hi
<tparys> what's new?
mrueg has quit [Quit: http://quassel-irc.org - Chat comfortably. Anywhere.]
<tparys> always surprises me the number of people asking for Raspberry Pi support on the forums
<ArmbianTwitter> @TvPrivacy (Privacy & Tech Tips): @MisterTechBlog @codingfield @thepine64 Got boot+network ๐Ÿ˜€ Had to use slightly earlier image (had issue w/the newest image - known to Armbian forums). Upgrading now. :) Will later put up upgraded image online but you guys are welcome to try working one I downloaded if you like: (link in next reply) https://t.co/J3AEOS4HRC https://tinyurl.com/ydlanv82 (16s ago)
<ArmbianTwitter> @TvPrivacy (Privacy & Tech Tips): @MisterTechBlog @codingfield @thepine64 If interested here is the Armbian one I have running of the Sopine/LTS supported images (with Debian base): ๐Ÿ™‚ https://t.co/xkcuspGdNX https://tinyurl.com/ye5n3wo7 (20s ago)
<stipa> tparys: not much, lol really, i'm not much on the forum
mrueg has joined #armbian
<stipa> there's no sane reason why would somone search for pi support on armbian forum
<sunshavi> perhaps they want to put armbian on their rpi's :)
<tparys> wouldn't be the first to request it, but there's not a build for them
<stipa> yeah sunshavi tparys , makes sense
<stipa> i think because it's not open hardware, there are blobs
<stipa> rpi is not so "open source"
<sunshavi> actually those guys always want to know the difference between raspbian and armbian. I think for them it is something like GNU/Linux an BSD. Always there is a curiosity for the Linux guys on trying BSD
<sunshavi> or on knowing what is different.
<stipa> sure
<stipa> they both run arm so it makes sense they're asking that questions
<stipa> but it's not a pc
<tparys> and it's not immediately clear that DTB and u-boot differences can make two boards using otherwise identical hardware vastly different
<stipa> yeah
<sunshavi> yeah. But as they always iterate the same question. I think doing some simple clarification notes could help them and avoid some annoying time on armbian users
<tparys> and as much as I understand the universal [Invalid] stamp, I'm not sure it's really clear to most newbies why they got it
<sunshavi> yeah. I was very near on buying one. kotczarny on maemo point me to orange pi plus 2e.
<sunshavi> and actually nowadays 8 gb or RAM is very tempting also
<stipa> it is
<kprasadvnsi[m]> Anybody bought a 8gb rk3566 tv box?
<kprasadvnsi[m]> They are cheap on AliExpress
<stipa> are you going to hack it?
<tparys> not until I finish figuring out my I2C clocking issues on my RK3399 ...
<kprasadvnsi[m]> <tparys "not until I finish figuring out "> What is the issue?
<tparys> My NanoPi M4V2 spams the kernel log with mclk errors every time it tries to make a sound, and apparently happens on multiple RK3399's
<tparys> other forum users have reported there's also I2C clocks that report incorrect parent clocks, so that's possibly wrong as well
<tparys> I got audio via HDMI working, and sorta shelved it for now, but need to sit down with some documentation and look into it some weekend
<kprasadvnsi[m]> I have OrangePi 4. Does it too suffer the same bug?
<tparys> not sure, check dmesg for rockchip-i2s errors about mclk
<kprasadvnsi[m]> <stipa "kprasadvnsi: this? / https://www"> No, there are even cheaper ones
<kprasadvnsi[m]> <tparys "not sure, check dmesg for rockch"> Ok, I will
<stipa> kprasadvnsi[m]: this? shorturl.at/koOQ3
<[TheBug]> tparys: my experience is analog auido out is very very very low on RockPi 4c, I also have seen this i2c clock errors I believe..
<[TheBug]> [ 416.654025] rockchip-i2s ff8a0000.i2s: Fail to set mclk -22
<[TheBug]> [ 416.654031] rockchip-i2s ff8a0000.i2s: ASoC: error at snd_soc_dai_set_sysclk on ff8a0000.i2s: -22
<[TheBug]> [ 416.654456] rockchip-i2s ff8a0000.i2s: ASoC: error at snd_soc_dai_set_sysclk on ff8a0000.i2s: -22
<[TheBug]> [ 416.654452] rockchip-i2s ff8a0000.i2s: Fail to set mclk -22
<kprasadvnsi[m]> <stipa "kprasadvnsi: this? shorturl.at/k"> US $54.01 27%OFF | NEW T95PLUS Android 11 TV Box Smart Set Top Box 2.4G 5G Wifi BT4.0 RK3566 Quad Core Google Voice 1000M 8K Media Player
<[TheBug]> I remember had convo about it with Tonymac32 I think
<[TheBug]> at some point
<tparys> yeah, I brought it up to him a couple times, and he suggested the RK3399 might have multiple clocking issues based on one of it's precursors (3238?)
<tparys> he mentioned he'd look into it, but suspected he got working on other things and never got back to it
<[TheBug]> yeah he has been busy recently it seems
<lanefu> Yeah i have that message on a lot of my rk3399 devices
<tparys> and I've been sorta poking at other things too, getting a Zynq up and running with Armbian in case anyone wants to spin their own virtual hardware
<tparys> I'll get some documentation up on github for the adventurous at some point
<lanefu> Hmmm
<lanefu> Actually i dont see that on my rpi4
<lanefu> s/rpi4/ORANGEPI 4/
<ArmbianHelper> lanefu meant to say: Actually i dont see that on my ORANGEPI 4
<tparys> good. just wanted to make sure I wasn't hallucinating things
<tparys> Raspberries = Broadcom, right?
<stipa> he got me there for a moment
<lanefu> Nor my rockpi 4b
<lanefu> lane@rockpi4b:~$ uname -a
<lanefu> Linux rockpi4b 5.11.12-rockchip64 #shedday SMP PREEMPT Thu Apr 8 21:17:39
<lanefu> EDT 2021 aarch64 aarch64 aarch64 GNU/Linux
<[TheBug]> lanefu: gasp, sacrilege, that will be three lashes!
<[TheBug]> lol
<[TheBug]> rpi4 thjough is a bit ambiguous now
<tparys> one difference I remember seeing was there was an "mclk" pinctl handle in the FriendlyElec DTB
<lanefu> lane@orangepi4:~$ uname -a
<lanefu> Linux orangepi4 5.10.25-rockchip64 #trunk SMP PREEMPT Sun Mar 21 10:00:58
<lanefu> EDT 2021 aarch64 GNU/Linux
<tparys> but I'm not sure I totally follow the rockchip-i2s driver, atm
<lanefu> You know those 2 boards are headless
<lanefu> So no hdmi or i2s signal in hdmi
<lanefu> [TheBug]: to you get that error on headless boards?
<tparys> anyway, I gotta get some sleep, later folks
<sunshavi> good night
<lanefu> Night
tparys has quit [Quit: Leaving]
<stipa> kprasadvnsi[m]: i guess we'll start to see 8GB SBCs soon with that RK3566 chip
Redentor has joined #armbian
<lanefu> Yep pine quartz64 is in manufacturing for dev samples
<stipa> fancy
<sunshavi> that would be excellent. I am curious. How much time is going to happen before another SOC get the kind of support that enjoys H3 nowadays?.
<kprasadvnsi[m]> <stipa "kprasadvnsi: i guess we'll start"> Yah, I believe Quartz64 will come with 8gb but it's gonna be pricey.
<stipa> cool
<stipa> if tv box is 50$
<sunshavi> everything is pricey. because chips and semiconductors are scarcey cos of the global virus
<kprasadvnsi[m]> SBC vendors should go for DDR3 instead of LPDDR4
<kprasadvnsi[m]> <stipa "if tv box is 50$"> It's DDR 3
<lanefu> Man ddr4 is whay makes a55 cores a good time
<kprasadvnsi[m]> <lanefu "Man ddr4 is whay makes a55 cores"> There is a difference between LPDDR4 and DDR4
archetech has quit [Quit: Leaving]
<kprasadvnsi[m]> There are very few application where memory speed become a major bottleneck
<lanefu> kprasadvnsi[m]: not to me ๐Ÿ˜
<stipa> sunshavi: yeah, price of hardware could be a problem
<kprasadvnsi[m]> DDR3 significant cheaper and have comparable performance to LPDDR4
<lanefu> Oh interesting
<lanefu> What's the difference in lpddr
<stipa> hmmm, why would they expensier lpddr4 on quartz64 when ddr is cheaper and same?
<stipa> they put*
<stipa> ddr3*
<stipa> tv box with same chip is running fine with ddr3
<stipa> maybe it wastes less power?
<stipa> or the ram manufacturers want to leave ddr3 behind
<stipa> and lpddr4 will be available on demand for pine
<stipa> less hassle maybe
<kprasadvnsi[m]> <lanefu "What's the difference in lpddr"> It's the low power version. Only single channel and limited in many ways to conserve power uses
Redentor has quit [Quit: Leaving]
<Tonymac32> it's quote season in the automotive world, and I am a lead for such activities in my group. So I've been stupidly busy.
<Tonymac32> as for the RAM discussion, 4K media in general needs as much bandwidth as you can throw at it memory wise, remember these SoC's don't have separate RAM for the graphics/vdec/etc
<Tonymac32> my email was exploding with zzz* patches, extlinux.conf, and compilation complaints. wtf is happening over here
<Tonymac32> :D
<kprasadvnsi[m]> <Tonymac32 "as for the RAM discussion, 4K me"> As far as memory goes LPDDR doesn't offer much, they are designed to prioritise power saving over performance.
<kprasadvnsi[m]> <Tonymac32 "as for the RAM discussion, 4K me"> As far as memory bandwidth goes, LPDDR doesn't offer much. They are designed to prioritise power saving over performance
<Tonymac32> correct, but there are a few boards using DDR4, not lpddr4
<kprasadvnsi[m]> Original discussion was about pricing. 8GB SBC will be expensive if they go for LPDDR4. It could be made cheaper if they used DDR3 as most TV Box vendors do.
<kprasadvnsi[m]> There isn't much value in using LPDDR4 for non battery powered devices
<Tonymac32> XD
<Tonymac32> I've got symlinks for patch folders pointing to a bucket full of folders
<Tonymac32> ugh so much for my WIP branches, git branch -d lol
<stipa> wb
bantu has quit [Ping timeout: 252 seconds]
bantu has joined #armbian
<Tonymac32> lol merge trunk into branch, ugly git history engage
pinkieval has quit [Excess Flood]
pinkieval has joined #armbian
manpaz has quit [Remote host closed the connection]
<Tonymac32> erg drivers/net/wireless/rtl8811cu/include/rtw_security.h:252:8: error: redefinition of โ€˜struct sha256_stateโ€™
<Tonymac32> until tomorrow friends
johnnynitwits has quit [Quit: Bridge terminating on SIGTERM]
threenp has quit [Quit: Bridge terminating on SIGTERM]
johnnynitwits has joined #armbian
johnnynitwits has quit [Excess Flood]
threenp has joined #armbian
johnnynitwits has joined #armbian
threenp has quit [Quit: Bridge terminating on SIGTERM]
johnnynitwits has quit [Quit: Bridge terminating on SIGTERM]
threenp has joined #armbian
johnnynitwits has joined #armbian
<IgorPec> tonymac32: hehe. symlinks are a part of the improvements ;)
Manouchehri has quit [Read error: Connection reset by peer]
lvrp16 has quit [Ping timeout: 258 seconds]
Manouchehri has joined #armbian
Manouchehri has quit [Excess Flood]
lvrp16 has joined #armbian
Manouchehri has joined #armbian
Fleck has joined #armbian
sunshavi has quit [Read error: Connection reset by peer]
sunshavi has joined #armbian
marco44 has quit [Disconnected by services]
marco44_ has joined #armbian
prefixcactus has joined #armbian
marco44_ is now known as marco44
f476_ has quit [Ping timeout: 252 seconds]
f476 has joined #armbian
f476_ has joined #armbian
f476 has quit [Ping timeout: 240 seconds]
stipa_ has joined #armbian
stipa has quit [Ping timeout: 240 seconds]
stipa_ is now known as stipa
attah has joined #armbian
stavrakis has joined #armbian
chewitt has joined #armbian
stipa_ has joined #armbian
Redentor has joined #armbian
stipa has quit [Ping timeout: 240 seconds]
stipa_ is now known as stipa
Net147 has quit [Ping timeout: 276 seconds]
Net147 has joined #armbian
Redentor has quit [Quit: Leaving]
sunshavi has quit [Quit: nil]
sunshavi has joined #armbian
<solderfumes[m]> /join !iyIlInqJyxXrRmRHFx:matrix.org
<solderfumes[m]> NO
<solderfumes[m]> this was not supposed to happen, sorry
<Tonymac32> too many solderfumes? :)
<solderfumes[m]> just finished recapping a late '70s Sony stereo receiver, it could be related
<stipa> kprasadvnsi[m]: RK3566 has 32Bit DDR Controller, doesn't it has a 4GB limit ? / https://www.cnx-software.com/wp-content/uploads/2020/12/RK3566-vs-RK3568.jpg
<stipa> if thex do some tricks to have 8GB then it for sure lowers speed of ram
<kprasadvnsi[m]> <stipa "if thex do some tricks to have 8"> There is no trick. The DRAM controllers these days can access large address space with fewer pins
<stipa> hmmm
<kprasadvnsi[m]> <stipa "kprasadvnsi: RK3566 has 32Bit DD"> 32bit is the bus length. You will not find a separate data/address bus
<kprasadvnsi[m]> stipa I can point you to some tutorial on this topic. Some articles explain how all this black magic happens
<stipa> so, how to they squeeze 64bit location into 32bit bus?
stavrakis has quit [Quit: Leaving]
<stipa> and 64 bit number
<stipa> in one instance
<stipa> that's possible only in few steps and it takes time
<kprasadvnsi[m]> There is a translation layer in DRAM controller that translation from CPU logical memory address to physical DRAM location.
<kprasadvnsi[m]> And vice versa
<stipa> that's some heavy voodo
<stipa> sounds like compression
<kprasadvnsi[m]> watch all videos on this playlist and then come to me.
<stipa> i'll take a look
Hokedli has joined #armbian
Elpaulo has quit [Quit: Elpaulo]
Hokedli has quit [Quit: Konversation terminated!]
plntyk has quit [Quit: Leaving]
xtron has joined #armbian
attah has quit [Remote host closed the connection]
xtron has quit [Ping timeout: 265 seconds]
tparys has joined #armbian
<tparys> evening all
<IgorPec> hey
<IgorPec> whats' up
<IgorPec> evening is only here :)
<tparys> not too much, trying to decide where to put some simple how tos to build Zynq FPGA images for the interested
<tparys> I tossed a working build up on google drive, but I suspect no one has the MicroZed board to try it
<IgorPec> yeah, its probably rareity
<IgorPec> we can put image to our servers if you like
<IgorPec> or i can give you users.armbian.com space
<tparys> the second would work, I saw that balbes was putting jetson nano images there
<IgorPec> ok, send ssh key
<tparys> I also saw this board which looked interesting: http://linuxgizmos.com/raspberry-pi-like-zynq-7020-sbc-sells-for-73/
<tparys> looks cheap and accessible for a Zynq, but I don't see a lot of I/O, and apparently the documentation is terrible
<[TheBug]> tparys also if you end up needing a place for a site let me know and I can setup a subdomain of systemonachip.net and give you a place to store files / a site as well if you want
<tparys> I appreciate that, thank you
<[TheBug]> I have similar things setup like atomicpi.systeonachip.net
<[TheBug]> I have similar things setup like atomicpi.systemonachip.net *
Fleck has quit [Quit: GOSUB w/o RETURN]
<IgorPec> tparys should work now
<tparys> thank you, I'll give it a try
<tparys> Permission denied (publickey) ...
<IgorPec> hah, it would be amazing if worked in first try :)
<tparys> permission issue on .ssh/authorized_keys file?
<IgorPec> i don't have scripted procedure ... let me check
<IgorPec> yes, wrong key permission
<IgorPec> try now
<tparys> still permission denied
<IgorPec> damn
<tparys> permission issue on .ssh ? I've hit that before
<tparys> no worries if it's not working, it's gotta be near midnight your time
<IgorPec> no, i'll fix this
<IgorPec> had a long nap in the afternoon, so i plan to be around for another 2h
<IgorPec> now?
<IgorPec> it works for my username
<tparys> still no
<IgorPec> errors is "permission denied" you sent me correct key?
<IgorPec> now?
<ArmbianTwitter> @TvPrivacy (Privacy & Tech Tips): @MisterTechBlog @codingfield + anyone else joining: don't upgrade to latest Armbian yet - issue persists in upgrade leads to failure to boot. Going to see if I can fix image & w/reply *here* w/fix after verifying upgrading fix works.๐Ÿ™‚ til then this works: https://t.co/xkcuspXOFv https://tinyurl.com/yzy2z76o (20s ago)
<tparys> still no, emailed you my file in case IRC ate a byte
<tparys> wait .. I'm a dummy
* tparys was using a different username at home
<tparys> it works
<IgorPec> haha
<IgorPec> ok
<tparys> you'd think I'd know better
<tparys> greatly appreciated. thanks.
<IgorPec> now i can grab a beer
<IgorPec> try to upload
<IgorPec> to see if you get here https://users.armbian.com/tparys/
<tparys> it works
<IgorPec> gr8!
<IgorPec> mechanical 15T 4,7T 9,5T 33% /tank/mechanical
<tparys> 10 TB is respectable, but I've grown numb to those big numbers
<IgorPec> hehe, this is not even ssd
<IgorPec> sorry, actually its not on this pool
<IgorPec> ssd 3,6T 755G 2,8T 21% /tank
<[TheBug]> authorized_keys should be 600
<[TheBug]> .pub files 644
<IgorPec> yes, they are now
<IgorPec> that was actually the only problem :)
<[TheBug]> arg sorry my silly buffer
<[TheBug]> I was stuck up above, sorry
<[TheBug]> thats why it seemed my response was really delayed, lol
<IgorPec> haha, there was some struggle in between, but that was it :)
<tparys> it's alright, there's an record of me being an idiot in multiple IRC logs :P
<IgorPec> so 3 x 100TB of exa drive
<[TheBug]> you need a place to ship them?
<[TheBug]> :D
<IgorPec> i have some spare place in the casing too :)
<tparys> I have one at work. Its comparable to the cost of a family car.
<[TheBug]> those crazy 100TB SSD drives they are making now which are a card with multiple 25TB SSDs looks amaziiing
<[TheBug]> yeah hook me up
<[TheBug]> I would take one :D
<IgorPec> i am done with storage sage for a while
<[TheBug]> (to be clear I wouldn't have the money to pay for it, but I would accept one :)
<tparys> so, they're big enough that mkfs.ext4 actually freaks out and doesn't print to screen correctly
<[TheBug]> um if I got about 3 of those I could shutdown 3 servers LOL
<[TheBug]> I could run my whole storage on 3 drives
<[TheBug]> haha
<IgorPec> well, i actually split my main storage pool into two
<IgorPec> double parity into single parity on two servers
<[TheBug]> IgorPec: can you imagine a time where people withh start either 'leasing' or 'mortgaging' a hard drive, LOL?
<[TheBug]> I feel like thats what its gonna come to when people have high storage need eventually
<[TheBug]> who needs a house, I need to mortgage a hard drive
<IgorPec> haha
<tparys> the big issue is when managers realize how much time that will take to stream out of a gigabit network connection
<IgorPec> most people don't need this big and fast storage
<tparys> ... which is measured in days
<[TheBug]> not now, but with VR..
<[TheBug]> maybe soon
<[TheBug]> could imagine 25-50TB datasets
<IgorPec> gigabit is also history, especially in storage area
<[TheBug]> yeah will all be direct pcie connect and fiber
<[TheBug]> infinaband is coming into it's prime
<[TheBug]> soon 40Gbit will be cheap enough
<[TheBug]> 2 and 10Gbit isn't so uncommon
<tparys> at that point, it's probably better to stack a bunch of independent controllers and bolt them together with ceph
<IgorPec> 10gb is still expensive for home / small business
<tparys> don't move the data, just access it in place
<[TheBug]> actually, I was watching / reading about some of the new datacenter stuff they are coming out with
<[TheBug]> and now they are to the point with iommu and virtualization
<[TheBug]> that they will start deploying basically modules
<tparys> There are 2.5 and 5.0 GbE links via USB3?
<[TheBug]> amd providing resource ovver 100Gbit connect across network from different chasis
<[TheBug]> one will do all NVME drives
<[TheBug]> one all ssd storage
<[TheBug]> gpus
<[TheBug]> it will be in pools
<[TheBug]> and it will be dynamically assigned and virtualized over network
<IgorPec> yes, you can run 5Gg via usb3 ssd 3,6T 755G 2,8T 21% /tank
<IgorPec> sorry wrong paste
<IgorPec> i have this one it test configuration, attached to one RK3399 board
<[TheBug]> ohh yeah
<[TheBug]> I am using 2x1Gbit on multiple rk3399
<[TheBug]> but not 2.5 or 5
<IgorPec> i also tried 10Gbe
<[TheBug]> and seeing the price I know why
<IgorPec> but that is a bit too much
<[TheBug]> that adapter cost more than the board
<[TheBug]> lol
<[TheBug]> and then the network switch I would need...
<[TheBug]> woo
* [TheBug] puts on his big boy pants
<IgorPec> the idea is that we join our labs, at least this armbian part
<IgorPec> so anyone can access hardware of each other
<[TheBug]> nice
<IgorPec> now all test devices - currently around 40 - are booked and hooked to power delivery units with switch on/off, some have serial consoles, some not, subnets are defined
<IgorPec> this is primarely used for general testings, but also to jump on some device to check something at any moment is possible
xtron has joined #armbian
<lanefu> man i missed all the storager talk
<tparys> sorry?
xtron has quit [Ping timeout: 252 seconds]
<IgorPec> tparys: i assume this https://github.com/armbian/build/pull/2759 is not breaking anything and its done to some degree?
<IgorPec> lanefu: there was 3 x 100TB ssd giveway :)
<tparys> lanefu, it boots, and could be better, but I don't think I touched anything outsize of the Zynq config
<IgorPec> ok, then let's merge
<tparys> [TheBug] won cause I don't know how SSH keys work ;)
<lanefu> haha
<tparys> .. actually, it does add bootgen to lib/compilation.sh?
<tparys> I don't think I added that to the header comment, if that's a detail that's important
<IgorPec> which is not called if this board is not selected
<IgorPec> ah it is in main
<tparys> run no, but it does get compiled alongside sunxi-tools
<IgorPec> well, all those should be done different anyway
<tparys> I'll make a note and look into it in the future
<IgorPec> ok
<lanefu> I drank some night coffee.. so maybe i'll get somethign done this evening
<lanefu> kprasadvnsi[m]: re 8 gigs of 32bit DDR... is that like PAE we had on intel32bit?
<tparys> 32 bit CPU, or 32 bit DRAM address bus?
<[TheBug]> night coffee?
<[TheBug]> that almost sounds a bit ominous
<[TheBug]> you put 2 extra scoops in or something?
<[TheBug]> slip in some taurine or something?
<[TheBug]> lol
<[TheBug]> or is that what you drink to go to bed, mixed with booze?
* IgorPec will slowly finish his beer before runnin hibernate script
<lanefu> tparys: chmod 0600 /path/to/private.key
<lanefu> tparys: haha sorry my screen was crolled up
<lanefu> [TheBug]: sorry not as exotic as you'd like
<lanefu> if i drink the leftover coffee in the from the morning i'm prodiuctive at night
<[TheBug]> hehe