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
attah_ has joined #armbian
attah has quit [Read error: Connection reset by peer]
attah has joined #armbian
attah_ has quit [Ping timeout: 244 seconds]
slacker_HD has joined #armbian
attah_ has joined #armbian
attah has quit [Ping timeout: 244 seconds]
attah_ is now known as attah
slacker_HD has quit [Quit: Connection closed]
attah_ has joined #armbian
attah has quit [Read error: Connection reset by peer]
<stipa>
i've heard stories of fakes being better than the original
attah has joined #armbian
<stipa>
better performance and less power waste
<stipa>
not all ofcourse
<stipa>
but there are stories
attah_ has quit [Ping timeout: 244 seconds]
archetech has quit [Quit: Konversation terminated!]
<milkii>
What might be the situation regarding video hardware acceleration on the odroid n2?
Unit193 has quit [Remote host closed the connection]
Unit193 has joined #armbian
CrashTestDummy2 has joined #armbian
CrashTestDummy3 has quit [Ping timeout: 244 seconds]
<ArmbianTwitter>
@Superna9999 (Neil Armstrong): @TonyMac_32 @lanefu @armbian @haerwu @khadas_official @AmlogicPR @IloveRockchip Exact, the initial blobs are the only ones not open source (bl2, bl301 & bl31), but the rest is mainline and supported (even PCie + NVMe boot from U-boot for the VIM3/L) https://tinyurl.com/yk4ablz8 (20s ago)
CrashTestDummy3 has joined #armbian
cheakoirccloud has quit [Quit: Connection closed for inactivity]
CrashTestDummy2 has quit [Ping timeout: 264 seconds]
kilobyte_ch has joined #armbian
<ArmbianTwitter>
@kravietz_ (Paweł Krawczyk 🦇 @kravietz@social.privacytools.io): @Jacek85281417 I'm running NanoPi R2S as a firewall with Armbian and it's been working great for the last year https://tinyurl.com/ydnycbze (21s ago)
<ArmbianTwitter>
@armbian (armbian): RT @kravietz_: @Jacek85281417 I'm running NanoPi R2S as a firewall with Armbian and it's been working great for the last year https://tinyurl.com/yk4s6f5ehttps://tinyurl.com/yj8n6kyf (14s ago)
CrashTestDummy3 has quit [Ping timeout: 272 seconds]
CrashTestDummy has joined #armbian
CrashTestDummy2 has quit [Ping timeout: 244 seconds]
CrashTestDummy2 has joined #armbian
CrashTestDummy has quit [Ping timeout: 244 seconds]
oida has quit [Ping timeout: 244 seconds]
oida has joined #armbian
lanefu has quit [Read error: Connection reset by peer]
Kamilion has joined #armbian
lanefu has joined #armbian
lanefu has quit [Quit: WeeChat 3.2-dev]
lanefu has joined #armbian
lanefu has quit [Ping timeout: 244 seconds]
lanefu has joined #armbian
rneese has joined #armbian
rneese has left #armbian [Leaving]
BCMM has quit [Quit: Konversation terminated!]
CrashTestDummy3 has joined #armbian
CrashTestDummy2 has quit [Ping timeout: 272 seconds]
cheakoirccloud has joined #armbian
diederik has quit [Ping timeout: 272 seconds]
BCMM has joined #armbian
diederik has joined #armbian
diederik has quit [Client Quit]
diederik has joined #armbian
cheakoirccloud_ has joined #armbian
cheakoirccloud has quit [Ping timeout: 265 seconds]
cheakoirccloud_ is now known as cheakoirccloud
FizzBuzz has joined #armbian
<cheakoirccloud>
I've been thinking about the love ebin receives. What if the approach was more like whatever the Nouveau project uses? They obviously don't get any support from NV. Also, nobody has pointed out the open questions, so it's hard to judge Marvel's involvement... One thought I had is if the hardware has a bug, there obviously isn't a register somewhere to fix that bug... So asking Marvel where that flag is, obviously will
<cheakoirccloud>
result in Marvel seemingly having poor support when actuality it's a flaw elsewhere. All anyone has said is that Marvel offers no support, and that's obviously far from the whole story.
<lanefu>
yeah i mean that's up for the community to determine
<lanefu>
it could be global scale ti could be marvel
akanouras has quit [Quit: Client limit exceeded: 10000]
rneese has joined #armbian
<ArmbianTwitter>
@xahteiwi (Florian Haas): Completely painless apt upgrade from @Ubuntu Focal to Hirsute with @armbian on the @thepine64 Pinebook Pro. Thanks, devs and packagers! https://tinyurl.com/yz6n5bzb (21s ago)
akanouras has joined #armbian
<lanefu>
rneese: ^ 😛
<rneese>
hahah
<rneese>
lets say not
<rneese>
heheh
<rneese>
1001 ways to kill a device
<rneese>
lol
<pkal>
how come firefox on armbian always resets the homepage? Is that intentional or should it be reported as a bug?
<[TheBug]>
as in when you upgrade?
<[TheBug]>
or like everytime you start the application
<pkal>
every time I start it
<rneese>
thats a firefox thing not a armbian thing
<rneese>
I know chromium and falkon and vavaldi dont
<pkal>
ok, but that doesn't mean that the firefox package hasn't been patched to do this
<ArmbianTwitter>
@armbian (armbian): RT @xahteiwi: Completely painless apt upgrade from @Ubuntu Focal to Hirsute with @armbian on the @thepine64 Pinebook Pro. Thanks, devs and… https://tinyurl.com/yeuao9glhttps://tinyurl.com/yzk2fttb (26s ago)
<[TheBug]>
well luckily the board was in another box inside the original box
<[TheBug]>
but the heatsink and emmc were just bouncing around in there
<[TheBug]>
and the whole side of the box was smashed in (outside box)
<nekomancer[m]>
+ $183.22 Shipping & Import Fees.
<nekomancer[m]>
so sute
<[TheBug]>
considering I spent over 100$ on RockPro64 that isn't so bad
<[TheBug]>
ohh you mean just the shipping
<[TheBug]>
hahaha
<[TheBug]>
I was gonna say for whole thing 200$ sounds cheap
<[TheBug]>
0
<lanefu>
nekomancer[m]: free 1 day delivery for me lol
<nekomancer[m]>
lol
<[TheBug]>
so
<[TheBug]>
poll
<[TheBug]>
should I use the provided thermal pad for the heatsink or Artic Mx4?
<lanefu>
mx4
<[TheBug]>
yeah I leanrned my lesson, no thermal glue on this one
<[TheBug]>
:Z
<rneese>
the pad is ok but if the heatsync can sit right on the chip paste is better
<rneese>
mx4 is the best
<[TheBug]>
actually I lied somewhat I have only Artic Silver 5 left
<[TheBug]>
but it should be just as good
<FizzBuzz>
AS5? They still sell that?
* rneese
graps [TheBug] toothpaste and a heatsync
<rneese>
this should do the job
<rneese>
lol
<rneese>
this should keep you cooler for now
archetech has joined #armbian
<[TheBug]>
lol
<[TheBug]>
actually I have some good thermal plaster but the last board I applied it to with the SoC and memory in this orientation I ended up pulling a memory chip off the board when I removed the heatsink :(
<[TheBug]>
so no more of that
<[TheBug]>
AS5 I have had for a while its old stock lol
<[TheBug]>
need to order some new thermal paste actually
<stipa>
sometimes going dry without lubricant works ok too
<stipa>
maybe a little higher temps but ntothing to write home about
<[TheBug]>
well with an rk3399
<[TheBug]>
you need the heatsink and even a fan in some conditions
<[TheBug]>
if you really want to get performance out and not constantly throttle
<[TheBug]>
I got RockPro64 booted
flyback has quit [Quit: Leaving]
<[TheBug]>
seems if you have a blank emmc is still tries to check emmc before sdcard in their uboot which is a bit annoying
<[TheBug]>
wasn't booting from sdcard and got me scratching my head for a minute
<[TheBug]>
hmmm
<[TheBug]>
even more interesting
<[TheBug]>
the emmc has something on it O_o
<stipa>
maybe it's something worthless
<[TheBug]>
well its odd cause I didn't order the module preloaded
<[TheBug]>
but maybe thats what they do
<[TheBug]>
just to avoid support issues
<stipa>
that card looks good
archetech has quit [Quit: Konversation terminated!]
<stipa>
sbc*
<[TheBug]>
hmm must be dummy info or something cause I can't mount any of the partitions
<[TheBug]>
will just go ahead and dd over armbian I suppose
<stipa>
probably nothing you can't download from net
<stipa>
yeah, it screams for armbian for sure
flyback has joined #armbian
<stipa>
give it to it
flyback has quit [Read error: Connection reset by peer]
flyback has joined #armbian
flyback has quit [Read error: Connection reset by peer]
flyback has joined #armbian
FizzBuzz has quit [Ping timeout: 240 seconds]
<[TheBug]>
wow RockPro64 SoC runs much cooler at idle than on RockPi 4c, they must use the PCB for some of the heat disipation
<stipa>
on opi lite 2 heat goes through mounting terminals too
<stipa>
so the heat goes through scres
<stipa>
screws*
<stipa>
i have plastic housing for it, so no joy there
<stipa>
but i may be mistaken, that's just my observation
MaximillianGreat has quit [Quit: Going offline, see ya! (www.adiirc.com)]
<[TheBug]>
well PCB on 4c is RPi form factor
<[TheBug]>
on RockPro64 PCB is much larger
<[TheBug]>
like almost 2x the area
<[TheBug]>
so probably have copper in the PCB and it helps with heat disipation
<stipa>
yeah, that could contribute to what you're experiencing too
<stipa>
using whole groundplane for heatsink
<stipa>
and it hoes through screws since they're grounded too i guess
<stipa>
so the touch that plane
<stipa>
so they*
<stipa>
+ heatsink
<stipa>
i know smartphones work like that
<stipa>
and its blody hard to solder something to a ground plane and desolder
<stipa>
beacuse it's huge
<[TheBug]>
weee
<[TheBug]>
<3 1080p playback with VPU accell
<stipa>
fancy
<[TheBug]>
Thanks to Nicod and that package he helped put out
<[TheBug]>
works nicely on legacy
<[TheBug]>
(Armbian) Legacy
<[TheBug]>
man I like this board and the form factor
<[TheBug]>
need to test out PCIe yet though
<[TheBug]>
but not bad
<stipa>
yeah, that could be a cool part if it works
<[TheBug]>
need to test 4k output and also USB-C to DP
<[TheBug]>
if it works as expected may even be able to trade out my RPi4 as thin client for one
<nekomancer[m]>
helios4 idle with big heatsing gets 62°C by `htop`
<nekomancer[m]>
anti-ekological
<nekomancer[m]>
maybe it mines
<[TheBug]>
when idle?
pkal has left #armbian [rcirc on GNU Emacs 28.0.50]
<[TheBug]>
I mean max temp I see on RockPi 4c with 20mmx20mm heatsink and 5v fan is 62c under full load
<[TheBug]>
at idle usually around 45-46c
<[TheBug]>
the RockPro64 idled at 30c
<[TheBug]>
haven't fully loaded it yet to test
<[TheBug]>
I am using their fan heatsink they sell directly for RockPro64
<nekomancer[m]>
yes. just powered on, fresh armbian, no any workload
<[TheBug]>
nekomancer[m]: on all my RockPi 4c I have at least heatsink and 5v fan of same size attached
FizzBuzz has joined #armbian
<[TheBug]>
rk3399 runs very hot
<nekomancer[m]>
and when no boot souce looks the same
<nekomancer[m]>
but why when idle?
<[TheBug]>
without heatsink rk3399 is very hot in general
<[TheBug]>
so it will not suprise me that it is idle at 60c with little hear removal
<[TheBug]>
heat*
<[TheBug]>
do they provide a heatsink?
<nekomancer[m]>
I think they do something wrong. Idle shold keep cool.
<[TheBug]>
how big is it?
<[TheBug]>
do you have the performance govenor set to performance?
<[TheBug]>
if you want it to idle low you have to use ondemand govenor
<[TheBug]>
which should be the default
<[TheBug]>
my numbers on RockPi 4c are with it in performance on all cores
<[TheBug]>
I haven't put rockpro64 in performance yet
<[TheBug]>
will be one of my next test
<nekomancer[m]>
ho! armbian-config crashed when I try to see a governor
<[TheBug]>
check htop
<[TheBug]>
you will know
<[TheBug]>
it shows cpufreq and temp
<[TheBug]>
and if its in ondemand you will see core drop to like 408mhz
<nekomancer[m]>
heatsink on helios4 looks industrial
<[TheBug]>
if in performance it will never go under max setting
<[TheBug]>
cat: /sys/devices/system/cpu/cpufreq: Is a directory
<[TheBug]>
does exist on my ebin though
<[TheBug]>
maybe it is in a different place on that board?
<[TheBug]>
have to go look
<[TheBug]>
or did you build with that stuff enabled in your kernel if you built it?
<[TheBug]>
I believe that is optional
<[TheBug]>
may also tie to your issue?
<[TheBug]>
because if you don't include it being able to manage it may just set all to performance or highest value
<[TheBug]>
with the default kernel govenor
<[TheBug]>
I forget what section that is unfortunately to just tell you outright
<[TheBug]>
I just can recall it being an option
<IgorPec>
on ebin we have disabled cpu freq scaling to get stability, but now i think its back since it stable
<IgorPec>
for n2 i don't know atm
<[TheBug]>
he is actually on Helio4
<[TheBug]>
it is armada 388 instead
<IgorPec>
aha, there too
<[TheBug]>
but similar
<IgorPec>
yes
<IgorPec>
there is explanation in commit history why
<IgorPec>
tl;dr; implementation of cpu jumping is unstable
<[TheBug]>
make sense why it gets all confused under high load for long period
<[TheBug]>
probably a lot of my stability issues there tied to that as well
<[TheBug]>
guessing poorly optimized opp table?
<[TheBug]>
nekomancer[m]: ^^
<IgorPec>
those numers are hard to put together
<IgorPec>
especially if you only have one or two devices
<rneese>
ok
<[TheBug]>
IgorPec: I think his primary concern there is whatever is set may be too high for his device, he was saying at boot and idle it would sit 60c and lock up
<[TheBug]>
IgorPec: so I don't know if maybe that is in something he compiled, or is dev, or what
<IgorPec>
its default
<IgorPec>
fixed speed
<IgorPec>
the only way to gain stability on modern kernel. dvfs implementation is broken
<IgorPec>
and nobody seems to work on that
<nekomancer[m]>
I choose available via armbian-config kernels. This function was lost after `5.8.18-mvebu #20.11 SMP Mon Nov 23 13:40:35 CET 2020`
<IgorPec>
we already have hard time to bring this device up to 5.12
<IgorPec>
yes, its know to be lost
<IgorPec>
its a trade for stability
<IgorPec>
you can have it back, but it will crash ;)
<nekomancer[m]>
> the only way to gain stability on modern kernel. dvfs implementation is broken
<nekomancer[m]>
omg.
* nekomancer[m]
prefer stability for sure.
<[TheBug]>
nekomancer[m]: maybe you should find marvell mailing list and write to them about fixing dvfs on the devices
<[TheBug]>
s/should/could
<ArmbianHelper>
[TheBug] meant to say: nekomancer[m]: maybe you could find marvell mailing list and write to them about fixing dvfs on the devices
<[TheBug]>
;p
<[TheBug]>
see if you can make someone care
<IgorPec>
marvell will don't give a damn about A380 and mainline kernel
<[TheBug]>
cmon, we gotta be able to arrange an escort for the rigfht lonely developer..
<[TheBug]>
right*
<IgorPec>
they never did anything, why should they do it now?
<[TheBug]>
escorts?
<[TheBug]>
hahaha
<IgorPec>
for their 10y or something old cpu?
<[TheBug]>
j/k
<IgorPec>
i know
<[TheBug]>
gotta be someone over there we could convince
<[TheBug]>
just gotta know who to reach out to
<IgorPec>
this is purely community effort
<nekomancer[m]>
and annoy devs like strange man there 2 days ago.. sure they know they broke. and they not fix because they can't. hope can't for a while.
<[TheBug]>
won't know if you don't try
<IgorPec>
most of those vendors doesn't support mainline kernel
<IgorPec>
they have their BSP and that's it
<IgorPec>
especially not for 32bit cpu
<IgorPec>
i only notice some marvell activitiy on their latest chips
<Tables>
Could try to get them on GPL violations for products that use their SoCs.
<IgorPec>
you can try anything :) just what is worth doing?
<[TheBug]>
IgorPec: you telling me you don't know someone whom take $500 or something to fix the code? what does it take to get them to give some care, $1000?
<Tables>
Right. I'm sure someone could spare the time to email. :P . Personally I don't care about 32bit anything anymore.
<IgorPec>
no, i don't know
<[TheBug]>
there has got to be enough users to collect something reasonable for a solution
<IgorPec>
i don't send emails out for much more important stuff since I am unable to process that much emailss
<IgorPec>
yes, but organisining things is yet another effort
<IgorPec>
just donate for what you have is enough for us
<IgorPec>
if we are talking for services, therea are companies out there that do this
<IgorPec>
things works so well already because someones are investing their time
<IgorPec>
a380 is stable because we wasted days / weeks to find a workaround
<IgorPec>
why that is not covered first?
<[TheBug]>
but there is always going to remain the outliars which people don't understand enough to work on or its so difficult they won't do it for free
<Tables>
Are any boards/socs actually good/stable?
<Tables>
As in zero problems
<IgorPec>
sw support for any hw is pain
<IgorPec>
and most of the problems are caused by us, providing bad fixes
<IgorPec>
fixes one problem, open another
<IgorPec>
nobody has enough time to make it right
<Tables>
Maybe i'm asking the wrong question. What board 64bit is mainlined to the point where the minimum number of fixes from Armbian are needed.
<IgorPec>
that's also not very correct question
<IgorPec>
some boards we simply don't deal with much and we ship as is
<IgorPec>
so its difficult to answer
CrashTestDummy2 has joined #armbian
rneese has left #armbian [Leaving]
CrashTestDummy has quit [Ping timeout: 272 seconds]
willmore has quit [Ping timeout: 244 seconds]
<lanefu>
Lol im scrolling back and i have like a bunch of things to respond to
<lanefu>
1. Im pretty sure the rk3399 boot order in hardware is spi, emmc, then sdcard. So im surprised if you experienced something different on rockpi4
<[TheBug]>
emmc had something weird on it to start with, so even though I put sdcard in it was still attempting and failing to boot from emmc
<lanefu>
2. I always end up with emmcs preflashed with stuff and its annoying
<[TheBug]>
I removed emmc and just wrote it directly, boots just fine from emmc
<lanefu>
3. Yeah helios4 runs hi temp and at a fixed clock of 2ghz. The fans follow like a temo sensor for whatever the SOM is for helios4 and cool
<lanefu>
4. Im a big fan of the schedutil governor. It's more responsive and granular than ondemand. Honestly wish it was our default
<[TheBug]>
Have Rockpro64 running legacy desktop image with 'RK3399 Legacy Multimedia Framework' and its working nicely
<[TheBug]>
1080p video playback no sweat
willmore has joined #armbian
<[TheBug]>
welcome willmore ;)
<lanefu>
Hi willmore
<[TheBug]>
hehe I think he just pinged out and came back.. maybe he is afk ;p
<lanefu>
Lol Tables you're looking for stable armbian board?
<lanefu>
Fizbuzz was asking same thing earlier
<Tables>
I'm more just curious what the gold standard is atm
<[TheBug]>
I would say H3 is stable for most part but stable doesn't mean all the accel works and such
<lanefu>
Tables: what's use case?
<[TheBug]>
just means a usable desktop / server experience without crashing etc
<Tables>
docker
<[TheBug]>
rk3399 can be pretty darn stable in most cases
<[TheBug]>
depends how blleding edge you keep up
<[TheBug]>
but thats just anecdotal
<lanefu>
So i run a diverse cluster of stuff on arm boards using docker and nomad
<lanefu>
And yeah looking into it ive probably had best luck with rk3399 boards
<lanefu>
My orange pi 4 has been a trooper
<lanefu>
Rockpi4b has been good
<lanefu>
Orange pi 3 was really good
<lanefu>
pine H64 has been pretty good
<lanefu>
man i thoight i had metrics for it
<willmore>
Sorry, changed wifi netowrks and got disconnected.