<_mamalala_>
tkaiser: wow, they have banned you on the pine64 forum, i see... what a bunch of whiners ...
<tkaiser>
_mamalala_: it's only one person and not important at all :)
<_mamalala_>
hehe
<tkaiser>
But according to a few others that wrote me via PM he does really great damage their. Today someone produced a set of 5 youtube videos to convince him that GbE has real world advantages over Fast Ethernet
<_mamalala_>
rather sad that such videos are even needed
<tkaiser>
Stuff everyone knows needs to be explained to one moderator since otherwise he deletes/edits all post so that they contain only irrelevant stuff any more.
<_mamalala_>
btw, i played around with cpuburn-a7 yesterday, and noticed something ...
<tkaiser>
_mamalala_: Even more since the whole 'performance' issue the people have is due to using a Debian image with the wrong settings. By using longsleep's OS image the problem would not even exist :)
<_mamalala_>
it only creates n threads for n cores, but it doesn't set any affinity for the threads, so they end up on random cores, often several on the same core
<_mamalala_>
so, to get the cores really loaded one should use taskset to tie the threads to the cores ... makes it go much warmer in the process :P
<tkaiser>
_mamalala_: Really?
<_mamalala_>
yep
<tkaiser>
Or is this result of your RL kernel patches? ;)
<tkaiser>
s/RL/RT
<_mamalala_>
see yourself by doing the following: start cpuburn in one console, then start top in another ... in top, type "f" and select "last used cpu" to disaply as well
<_mamalala_>
it well then show you what core each thread/process runs on
<tkaiser>
Well I use htop normally and thought it would display utilization on each core individually. Will check later.
<_mamalala_>
you can then use "taskset -p -c <core> <PID>" to lock the thread with <PID> to <core>
<_mamalala_>
so far i tried only my rt-kernel as well as the stock image that friendlyarm links to, and both show the same behaviour
enrico_ has joined #linux-sunxi
<tkaiser>
_mamalala_: Yeah, I know taskset, use it regularly to see if things improve if Ethernet IRQs and application perform better when using the same core. But never experienced this behaviour or overlooked it since htop displays stuff wrong
<_mamalala_>
hmm, well, there is always a chance that i do something wrong ....
<KotCzarny>
_mamalala_: if you press '1' in top you will see cores load
<tkaiser>
_mamalala_: quick check, apt-get install htop and look what's happening. When it shows all CPU cores are busy while top shows the opposite...
sarietta has quit [Remote host closed the connection]
jonkerj has quit [Remote host closed the connection]
<_mamalala_>
hmmm, this is strange ... just re-tested it, and now it does indeed spread the threads over all the cores ...
<_mamalala_>
so i take back what i said and blame myself instead
<_mamalala_>
dunno what made it do what it did yesterday, though
<KotCzarny>
maybe something used one core in themoment you started it
<KotCzarny>
then ended
pg12 has quit [Ping timeout: 240 seconds]
<KotCzarny>
but sticking threads tocores might be good idea anyway, ssvb
pg12 has joined #linux-sunxi
jonkerj has joined #linux-sunxi
iamfrankenstein1 has quit [Quit: iamfrankenstein1]
iceangel666 has joined #linux-sunxi
Winston_Wang has quit [Ping timeout: 255 seconds]
iceangel666 has quit [Ping timeout: 250 seconds]
Mr__Anderson has joined #linux-sunxi
apritzel has quit [Read error: Connection reset by peer]
<ssvb>
_mamalala_: a linux kernel scheduler glitch?
<_mamalala_>
ssvb: dunno what it was, i can't reproduce it now
<ssvb>
my understanding is that setting affinity is mostly important to avoid the task migration from one core to another unnecessarily thrash L1 caches
paulk-collins has joined #linux-sunxi
<ssvb>
but the cpuburn program is just a simple loop which does not access much
<_mamalala_>
yea, it could be that the threads nigrated while it was running ... as said, so far i'm unable to reproduce it ... don't want to spend too much time on that either ;)
utente_ has joined #linux-sunxi
<lvrp16>
hey guys
<lvrp16>
i got the de code and gpu code
<plaes>
GPU code?
<plaes>
does it have proper copyrights?
Mr__Anderson has quit [Remote host closed the connection]
orly_owl_ has joined #linux-sunxi
Andy-D has quit [Ping timeout: 265 seconds]
apritzel has quit [*.net *.split]
orly_owl has quit [*.net *.split]
NiteHawk has quit [*.net *.split]
tuxillo has quit [*.net *.split]
formruga has quit [*.net *.split]
akaizen has quit [*.net *.split]
miasma has quit [*.net *.split]
sW` has quit [*.net *.split]
indy has quit [*.net *.split]
avph has quit [Ping timeout: 264 seconds]
pietrushnic has quit [Ping timeout: 240 seconds]
pietrushnic has joined #linux-sunxi
NiteHawk has joined #linux-sunxi
NiteHawk has quit [Changing host]
NiteHawk has joined #linux-sunxi
apritzel has joined #linux-sunxi
miasma has joined #linux-sunxi
tuxillo has joined #linux-sunxi
sW` has joined #linux-sunxi
akaizen has joined #linux-sunxi
formruga has joined #linux-sunxi
indy has joined #linux-sunxi
avph has joined #linux-sunxi
Da_Coynul has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
IgorPec has quit [Ping timeout: 250 seconds]
orly_owl_ has quit [Quit: leaving]
orly_owl has joined #linux-sunxi
bfree has joined #linux-sunxi
<jelle>
for the h3?
IgorPec has joined #linux-sunxi
utente_ has quit [Ping timeout: 240 seconds]
premoboss has joined #linux-sunxi
tkaiser has joined #linux-sunxi
<montjoie>
apritzel: could you test as you proposed my emac driver on a big endian kernel ?
afaerber has quit [Quit: Ex-Chat]
avph has quit [Ping timeout: 260 seconds]
afaerber has joined #linux-sunxi
avph has joined #linux-sunxi
avph has quit [Ping timeout: 264 seconds]
Mr__Anderson has joined #linux-sunxi
IgorPec has quit [Ping timeout: 265 seconds]
afaerber has quit [Ping timeout: 250 seconds]
IgorPec has joined #linux-sunxi
enrico_ has quit [Ping timeout: 276 seconds]
The_Loko has joined #linux-sunxi
premoboss has quit [Ping timeout: 244 seconds]
afaerber has joined #linux-sunxi
<longsleep>
muhahah this video of terminals showing pine64 network speed - so funny and in category 'Comedy' on Youtube
<apritzel>
montjoie: Sure (but only later tonight). Where can I find the latest and greatest version of your driver?
<apritzel>
tkaiser: well, I don't touch PLL_CPUX, so it's what boot0 and/or U-Boot set up
<jmcneill>
tkaiser: did manu ever follow up to your email about h3?
<apritzel>
tkaiser: tbh: I didn't look at performance so far, the board seems to be limited by the 115200 bps of the serial terminal to me ;-)
<tkaiser>
apritzel: Ok, then I know that I have to look at u-boot now ;) I really hate sysbench but at least it's ok for calculating cpufreq based on sysbench scores (since no dependency on memory bandwidth)
nereeb has joined #linux-sunxi
<jmcneill>
Pine64 is definitely 408MHz coming out of U-Boot
iamfrankenstein has quit [Ping timeout: 240 seconds]
<tkaiser>
jmcneill: thanks for confirmation and no, no feedback so far.
<jmcneill>
i'll bug him
jernej has quit [Ping timeout: 264 seconds]
<tkaiser>
So to get any reasonable EMAC driver results with iperf adjusting that to 816 MHz seems mandatory to me
<tkaiser>
jmnceill: Not necessary, he should do what he thinks is ok. Already too much involved in stuff i'm not that much interested in ;)
<apritzel>
tkaiser: the problem seems to be that U-Boot does some clock setup in the SPL, but that's not used when using boot0
<apritzel>
so us hackers using FEL and the libdram SPL hack get different numbers than those people using boot0 on a SD card
sunxi_fan1 has quit [Remote host closed the connection]
<tkaiser>
apritzel: I fear I don't understand that much but at least someone else knows maybe where to look for when it's about judging EMAC performance with mainline on A64. It's CPU limited so increasing cpufreq is necessary.
<apritzel>
should be history once I find this nasty bug in my 64-bit SPL-FEL branch (with jemk's patches without libdram)
<apritzel>
then we get a pure FOSS stack and everyone uses a pure AArch64 U-Boot with SPL
nereeb_ has joined #linux-sunxi
<apritzel>
so save a date for that boot0/boot0img funeral ;-)
<tkaiser>
apritzel: ok, will remind you daily from now on ;)
<apritzel>
tkaiser: it kind of works already, but FEL is still broken, it crashes on returning to the FEL stack
<apritzel>
and I am too lazy to flash a SD card (because FEL is so convenient)
<tkaiser>
apritzel: Damn! Why did we told you? ;)
cnxsoft has quit [Quit: cnxsoft]
<tkaiser>
longsleep: How many Pine64+ were sent to you in total and cured by you from their GbE disease?
<apritzel>
FEL is admittedly a bit crazy with a 64-bit SPL, because we switch to 64-bit, run the SPL, then switch back to 32-bit
cptG has joined #linux-sunxi
<longsleep>
tkaiser: well i got one from some guy from Germany - thats the only one and it worked fine in my network
IgorPec has joined #linux-sunxi
<longsleep>
tkaiser: also i have 10 Pine64 in use in various networks and they all work
<apritzel>
longsleep: are there desperate pilgrims coming to you with their Pine64s in hoping to get them miraculously cured by being touched by you?
<tkaiser>
longsleep: ok, only one cured. I tested my 2 in all combinations even with CAT5 cables from 20 years ago. Not a single problem. Really strange.
<longsleep>
apritzel: send women and i am happy to touch :P
jernej has joined #linux-sunxi
<longsleep>
tkaiser: yeah .. i also got a very cheap switch from Amazon and the same guy bought that same switch and it did not work for him. So its not the switch but the peer - thats pretty much all i can tell.
<apritzel>
longsleep: sorry, I meant you touch the boards ;-)
cptG_ has quit [Ping timeout: 264 seconds]
* longsleep
prefers women over boards
<lvrp16>
i got the code for the h3 display engine and h3 gpu, how does one disclose responsibly?
Nacho has quit [Remote host closed the connection]
<longsleep>
lvrp16: look at the license, it should tell :)
<apritzel>
lvrp16: put on a USB stick and drop it at some conference ;-)
<_mamalala_>
or make an anonymous github account and check it in there
<lvrp16>
apritzel: just one?
<apritzel>
_mamalala_: sure, an anonymous account after having asked that question in a public forum ;-)
Nacho has joined #linux-sunxi
<lvrp16>
;)
arossdotme-planb has joined #linux-sunxi
<_mamalala_>
well, i'd be be happy to do the dirty deed, if needed .... :P
arossdotme has quit [Ping timeout: 250 seconds]
<apritzel>
_mamalala_: I think lvrp16 is asking for a legally sane way to do so
<apritzel>
lvrp16: is there a license on this code?
<apritzel>
lvrp16: did it come with a NDA agreement?
<jernej>
lvrp16: Is H3 DE code substantially different from that in BSP kernel? Is GPU code only for kernel?
<_mamalala_>
if either of those are the case, and the license say not to hand it to anyone else, then i guess there is no "sane" way to disclose it .... at least not easily
<apritzel>
_mamalala_: agreed
<lvrp16>
i left it at the office, i only took a glance at it
<lvrp16>
will take a deeper look tomorrow
<_mamalala_>
one could write a poem about the important bits and pieces of the contents, though .... :P
_whitelogger has joined #linux-sunxi
nereeb_ has quit [Quit: Ik ga weg]
apritzel1 has joined #linux-sunxi
<_mamalala_>
personally i wouldn't give a damn and just upload it somewhere anonymously ... it's not as if that stuff is the source and launch codes for some nuclear arsenal or the likes ... but then, that's just me
<_mamalala_>
about two years ago we used that method to get espressif open-source their code for the esp8266 chips, for example
JohnDoe_71Rus has joined #linux-sunxi
<apritzel>
_mamalala_: the problem is that if someone pushes this code into a proper OpenSource project like the Linux kernel
joedj has quit [Ping timeout: 240 seconds]
<apritzel>
_mamalala_: then you can't really put your "Signed-off-by:" on it
<_mamalala_>
yea, 1:1 copying of that code into the kernel would be stupid, sure ... but using as basically some sort of documentation to re-implement it should be ok, i think
<_mamalala_>
in case a nda was signed, it's worth reading it ... there's a good chance that there is a clause in it that says "not allowed to disclose anything, except for the parts that otherwise came into the public"
<apritzel>
for clean room engineering one would need to translate that code, into documentation, for instance
<_mamalala_>
then, with someone having it uploaded somewhere, it becomes possible to legally disclose the stuff because by then it is available somewhere else already
<mripard>
_mamalala_: being public doen't mean the license is compatible
joedj has joined #linux-sunxi
<_mamalala_>
mripard: i'm not talking about any license here, just a pssible nda ... and as said, copying the code 1:1 would be a no-go still ...
apritzel1 has quit [Ping timeout: 244 seconds]
<mripard>
you CANNOT use some code under a given license to get some "inspiration" and implement something under a different license
<_mamalala_>
i'm looking at it from the "makes some good refence/docu for re-implementation" angle
<mripard>
it's not even good for that
<mripard>
if you ever come across proprietary code
<mripard>
don't even look beyond the license
<mripard>
and certainly don't talk about doing that in public, archived, IRC channels
<jernej>
mripard: what about NDA datasheets? Can they be used for legaly implementing a driver, if they are found somewhere on the net?
ninolein has quit [Remote host closed the connection]
reinforce has quit [Quit: Leaving.]
Mr__Anderson has joined #linux-sunxi
azend|vps has quit [Remote host closed the connection]
azend|vps has joined #linux-sunxi
vagrantc has joined #linux-sunxi
<lvrp16>
i did not check the license
<lvrp16>
will report back tomorrow
ninolein has joined #linux-sunxi
Andy-D has joined #linux-sunxi
iamfrankenstein has joined #linux-sunxi
avph has joined #linux-sunxi
ninolein has quit [Remote host closed the connection]
ninolein has joined #linux-sunxi
iamfrankenstein has quit [Ping timeout: 252 seconds]
premoboss has quit [Ping timeout: 250 seconds]
ninolein has quit [Remote host closed the connection]
ninolein has joined #linux-sunxi
reinforce has joined #linux-sunxi
Mr__Anderson has quit [Remote host closed the connection]
netlynx has joined #linux-sunxi
<mripard>
jernej: datasheet are not code
<mripard>
but I guess it depends on the license and NDA.
Amit_t_ has joined #linux-sunxi
Nacho has quit [Ping timeout: 264 seconds]
Nacho has joined #linux-sunxi
<apritzel>
some companies have strict rules on those NDA docs, so employees cannot use those without getting into trouble
ninolein has quit [Remote host closed the connection]
sunxi_fan has joined #linux-sunxi
ninolein has joined #linux-sunxi
ninolein has quit [Remote host closed the connection]
perr has joined #linux-sunxi
perr has joined #linux-sunxi
perr has quit [Changing host]
ninolein has joined #linux-sunxi
rotcpy has joined #linux-sunxi
nereeb has quit []
enrico_ has quit [Quit: Bye]
perr has quit [Quit: Leaving]
ninolein has quit [Remote host closed the connection]
ninolein has joined #linux-sunxi
ninolein has quit [Remote host closed the connection]
ninolein has joined #linux-sunxi
IgorPec has quit [Read error: Connection reset by peer]
premoboss has joined #linux-sunxi
ricardocrudo has quit [Remote host closed the connection]
ninolein has quit [Remote host closed the connection]
ninolein has joined #linux-sunxi
ninolein has quit [Remote host closed the connection]
ninolein has joined #linux-sunxi
iamfrankenstein has joined #linux-sunxi
popolon has quit [Quit: WeeChat 1.4]
ninolein has quit [Remote host closed the connection]
ninolein has joined #linux-sunxi
JohnDoe5 has joined #linux-sunxi
ninolein has quit [Remote host closed the connection]
ninolein has joined #linux-sunxi
JohnDoe_71Rus has quit [Ping timeout: 260 seconds]
ninolein has quit [Remote host closed the connection]
ninolein has joined #linux-sunxi
Mr__Anderson has joined #linux-sunxi
ninolein has quit [Remote host closed the connection]
ninolein has joined #linux-sunxi
apritzel has quit [Ping timeout: 244 seconds]
ninolein has quit [Remote host closed the connection]
Amit_t_ has quit [Quit: Page closed]
akaizen has quit [Read error: Connection reset by peer]
ninolein has joined #linux-sunxi
akaizen has joined #linux-sunxi
ninolein has quit [Remote host closed the connection]
ninolein has joined #linux-sunxi
florianH has quit [Quit: Connection closed for inactivity]
montjoie has quit [Ping timeout: 252 seconds]
Mr__Anderson has quit [Remote host closed the connection]
montjoie has joined #linux-sunxi
afaerber has quit [Quit: Ex-Chat]
jernej has quit [Quit: Konversation terminated!]
ninolein has quit [Remote host closed the connection]
ninolein has joined #linux-sunxi
ninolein has quit [Remote host closed the connection]
ninolein has joined #linux-sunxi
<jonkerj>
mripard: per your reply: when the pins (which have functions for other peripherals as well) are associated with a disabled peripheral (say uart2), users are still free to choose how to use them, right?
<jonkerj>
(ie, I seem to miss some background to understand your point)
<jonkerj>
I thought that as long as the peripherals in my patch have "status = disabled", they could be used for other stuff
ninolein has quit [Remote host closed the connection]
ninolein has joined #linux-sunxi
mzki has quit [Quit: leaving]
ninolein has quit [Remote host closed the connection]
ninolein has joined #linux-sunxi
_whitelogger has joined #linux-sunxi
ninolein has quit [Read error: Connection reset by peer]
ninolein has joined #linux-sunxi
ninolein has quit [Remote host closed the connection]
ninolein has joined #linux-sunxi
foxx_ has quit [Ping timeout: 255 seconds]
tkaiser has joined #linux-sunxi
tkaiser has quit [Client Quit]
afaerber has joined #linux-sunxi
Nacho has quit [Ping timeout: 265 seconds]
ninolein has quit [Remote host closed the connection]
ninolein has joined #linux-sunxi
netlynx has quit [Quit: Ex-Chat]
sarietta has joined #linux-sunxi
Andy-D has quit [Ping timeout: 276 seconds]
apritzel has joined #linux-sunxi
Nacho has joined #linux-sunxi
apritzel has quit [Ping timeout: 244 seconds]
Mr__Anderson has joined #linux-sunxi
jernej has joined #linux-sunxi
Putti has joined #linux-sunxi
mnemoc has quit [Read error: Connection reset by peer]