jstein has quit [Remote host closed the connection]
TheSeven has quit [Disconnected by services]
[7] has joined #linux-sunxi
apritzel has quit [Ping timeout: 244 seconds]
egbert has quit [Disconnected by services]
egbert has joined #linux-sunxi
ninolein_ has joined #linux-sunxi
ninolein has quit [Ping timeout: 276 seconds]
cnxsoft has joined #linux-sunxi
cnxsoft has quit [Read error: Connection reset by peer]
cnxsoft has joined #linux-sunxi
<wens>
montjoie: fyi i would drop the regulator bindings for now, since a) we don't support the pmic, and b) there's discussion on how to do power sequencing for various busses
<wens>
that would also include stuff like enable and reset gpios
leoncito81 has joined #linux-sunxi
leoncito81 has quit [Remote host closed the connection]
<apritzel>
yeah, so that's exactly half of the advertised frequency!
clonak_ has joined #linux-sunxi
<wens>
:(
clonak has quit [Ping timeout: 240 seconds]
<tkaiser>
apritzel: With a really heavy CPU bound workload
diego71 has quit [Ping timeout: 246 seconds]
<apritzel>
so how does this show on the BSP kernel? Does the system crash?
<apritzel>
or does the arisc kicks in and turns everything off?
<tkaiser>
When the Pine64 folks made their marketing video stuff showing 38¡C and low consumption I would assume they already were running on a single CPU core since back then BSP throttling settings were as broken as we already knew from H3/A83T
Nacho has joined #linux-sunxi
diego71 has joined #linux-sunxi
<tkaiser>
apritzel: Nope, we improved the THS settings back in March. Before with longsleep's kernel he also lost CPU cores all the time. With the new settings throttling works reliable and killed CPU cores only happen with weird stuff like.... cpuburn-a53
<apritzel>
tkaiser: I'd guess "impressive" Android games may stress more the GPU than all CPU cores
<tkaiser>
Exactly. They still run nicely even if 3 CPU cores were killed already due to short load peaks before
afaerber has quit [Ping timeout: 250 seconds]
<apritzel>
so you give it more saner operating points then?
<tkaiser>
apritzel: Yes, we increased both count of cpufreq steps to make throttling more efficient, added also more dvfs operating points and tuned the THS settings
<tkaiser>
Now BSP kernel starts a shutdown when reaching 108¡C
<tkaiser>
And behaves more sane before that happens trying to throttle down instead of killing CPU cores
<tkaiser>
apritzel: Temperature between 88¡C and 92¡C at 600 MHz. So with 816 MHz @ 1.1V you can harm to A64 for sure :) At least with cpuburn-a53 or cpuminer
<wens>
why would people not put a heatsink on it...
<tkaiser>
wens: The only reason my remaining Pine64 doesn't wear one is to compare. But I sent my other Pine64 to another dev in the meantime so that will change soon
<tkaiser>
The 'el cheapo' heatsinks for A20 (web search for '20x20 cubie heatsink') for 0.5$ reduce upper temperatures by 10¡C-15¡C
<apritzel>
has anyone tried to do those tests without throttling / core-killing? I wonder if the system would crash (so removing the load) or whether the CPU would actually be harmed due to overheating
hansg has joined #linux-sunxi
montjoie has quit [Ping timeout: 260 seconds]
montjoie has joined #linux-sunxi
<tkaiser>
apritzel: I had sudden power-offs back then but still no clue what the reason was (PMIC responsibly cutting power or something else)
paulk-leonov has joined #linux-sunxi
<tkaiser>
But so far either those mysterious power-offs or shutdown initiated by kernel
buZz has joined #linux-sunxi
enrico_ has joined #linux-sunxi
buZz is now known as Guest88239
<tkaiser>
apritzel: How many Pine64 do you have lying around?
enrico__ has quit [Ping timeout: 260 seconds]
<apritzel>
tkaiser: ;-) Enough to risk one of them
<tkaiser>
apritzel: Nice, then give it a go :)
* apritzel
looks for the next fire extinguisher ;-)
afaerber has joined #linux-sunxi
enrico_ has quit [Ping timeout: 244 seconds]
enrico_ has joined #linux-sunxi
cnxsoft has quit [Quit: cnxsoft]
Guest88239 is now known as buZz
<tkaiser>
apritzel: With 'stress -m 3 -c 3' temperature at 816MHz@1.08V doesn't exceed 70¡C so I would consider 816MHz as 'safe' unless people want to run special code
<apritzel>
tkaiser: thanks a lot for those tests
<apritzel>
so I rather keep it at 816 MHz for the time being, until I have some DVFS or throttling in place
enrico_ has quit [Ping timeout: 246 seconds]
enrico_ has joined #linux-sunxi
<tkaiser>
apritzel: Or if it's just for testing purposes limit CPU cores to 2 and allow higher clockspeeds?
<apritzel>
Personally I rather have 4 lower clocked cores than 2 higher clocked ones
<tkaiser>
But there we make use of dvfs already and then all cores at lower clockspeeds are more efficient than less cores at higher clockspeeds
Amit_t_ has joined #linux-sunxi
<tkaiser>
apritzel: Did another test with 'sysbench --test=cpu'. This one is that light that I get 88¡C @ 1056MHz/1.24V, so maybe 816 MHz is really too conservative for 'normal' workloads
reinforce has quit [Quit: Leaving.]
<tkaiser>
apritzel: Now 'make -j5' with OpenSSL. Throttling limits this to 960MHz@1.16V and temperature slightly exceeds 90¡C
<apritzel>
tkaiser: thanks, I will try some tests myself, though I don't have a suitable thermometer which goes beyond 42C ;-)
<apritzel>
but those laser gun type ones seem to be cheaper than I expected
<tkaiser>
apritzel: All results rely solely on the thermal sensor (to be more precise on the one for the CPU -- according to user manual there are 3 inside. See chapter 2.1.4.8)
<tkaiser>
Since all three are there for only one reason -- throttling and over-temperature protection -- I think we can trust the readouts
<tkaiser>
If NEON is not used and you make wearing a heatsink a requirement then 960MHz@1.16V should be fine
hansg has quit [Quit: Leaving]
IgorPec has quit [Ping timeout: 250 seconds]
enrico__ has joined #linux-sunxi
JohnDoe_71Rus has joined #linux-sunxi
enrico_ has quit [Ping timeout: 240 seconds]
orly_owl has quit [Ping timeout: 244 seconds]
<MoeIcenowy>
Oh seems that recent Allwinner SoCs are all facing heat problem?
<MoeIcenowy>
from H3 to A64
<MoeIcenowy>
(I have now only A31s and A33
<buZz>
MoeIcenowy: heat problems = needing a heatsink ?
<buZz>
or what kinda problems
apritzel has quit [Ping timeout: 244 seconds]
<Amit_t_>
What these address(parameters) to this function sunxi_rsb_config(AXP81X_ADDR, 0x3a3) are calculated ?
apritzel has joined #linux-sunxi
<MoeIcenowy>
buZz: yes
<tkaiser>
MoeIcenowy: I would suspect that's true for most SoCs from this and last year.
<buZz>
then its not really a problem, is it?
<buZz>
cpus get faster -> need cooling
<buZz>
nothing weird, is it?
<MoeIcenowy>
tkaiser: yes
<tkaiser>
buZz: Heatsinks and even small fans won't help with heavy workloads. You need a thermal sensor inside and code to prevent damage
<buZz>
ah, thats not in hw already?
<Amit_t_>
I mean, how these address(parameters to this function) sunxi_rsb_config(AXP81X_ADDR, 0x3a3) are calculated ?
<tkaiser>
And that's the problem with newer SoCs and mainline kernel: If you use the defaults used by 'vendor kernel' and haven't implemented thermal readouts/throttling damage might occur
The_Loko has quit [Quit: Leaving]
<wens>
Amit_t_: they are not calculated
<wens>
Amit_t_: the hardware address is hardwired in the pmic
<wens>
Amit_t_: the runtime address is from a list of valid addresses
<wens>
(there is a description in mainline drivers/bus/sunxi-rsb.c)
drummer_A20 has joined #linux-sunxi
<Amit_t_>
wens: Thanks, for axp803 where should I look to find it ?
<tkaiser>
apritzel: If you don't need further testings I'll shut the board down since I need the PSU within the next hour
<plaes>
KotCzarny: you don't want to do that :P
<KotCzarny>
plaes: you are right, not enough supply, i'll bring more twigs to keep them burning
paulk-leonov has quit [Ping timeout: 240 seconds]
<wens>
mripard: hi, the chip dts lists ldo5 as an 1.8v regulator, however i don't see this in the schematics
avph has joined #linux-sunxi
massi_ has joined #linux-sunxi
reinforce has joined #linux-sunxi
<apritzel>
tkaiser: that much more testing than I could ever hope for, thanks a ton!
<tkaiser>
apritzel: Glad to help
<Amit_t_>
wens: AXP81X_ADDR is set 0x11 , which I could not really find out how ?
<ssvb>
apritzel: just received this W25Q Windbond Serial Flash Memory Module, now only need to solder pin headers to it and do some experiments :-)
<apritzel>
ssvb: yeah!
<apritzel>
ssvb: do you know how to tell Linux about it properly? the kernel has a "driver" for that particular chip, I just didn't find an easy way to access it (lost in MTD translation)
<ssvb>
I don't intend to use Linux, but will rather talk to it directly over SPI
<KotCzarny>
read and bookmark this page for starters
vagrantc has joined #linux-sunxi
<leoncito81>
Ok.
<leoncito81>
Bookmark done...
scream has joined #linux-sunxi
<apritzel>
leoncito81: you could go through the Pine64 wiki page and try to get something to work. Then either fix the Wiki page or report here what's missing or wrong
<leoncito81>
Ok. By try to get something to work, what do you mean?
<apritzel>
try to boot some kernel, for instance
<leoncito81>
Ok
<jelle>
ah my other tablet has an RTL8710AS wifi chip, but I broke android...
<apritzel>
assuming you have a board (and assuming it's the Pine64)
<leoncito81>
I have one of longsleep's Arch distros up on my pine now
<jelle>
*8703AS
diego_ has joined #linux-sunxi
<apritzel>
leoncito81: OK, then you could check whether the instructions are any good
diego_ is now known as diego71
<plaes>
jelle: check what CHIP has
<leoncito81>
Ok. I'll start there.
<apritzel>
leoncito81: I have the feeling we are missing a lot here, the people who care know what to do, but other people might get lost
<jelle>
plaes: that has a different one
<leoncito81>
True
<plaes>
jelle: damn realtek :)
<jelle>
plaes: oh well more things to do!
<MoeIcenowy>
plaes: realtek is much better than espressif :-(
diego71 has quit [Client Quit]
<MoeIcenowy>
I now only want to say something dirty to espressif
<apritzel>
MoeIcenowy: be expressive about espressif?
<jelle>
MoeIcenowy: no docs, no bsp driver?
<plaes>
no market so far?
<MoeIcenowy>
jelle: nearly no docs
<MoeIcenowy>
I have some driver source codes from A33/RK3188 SDKs
<jelle>
MoeIcenowy: ahhh
<jelle>
RDA you can request documentation but hrrm
<jelle>
not sure if that will work
<plaes>
MoeIcenowy: at least they have been doing everything right
dev1990 has joined #linux-sunxi
diego_ has joined #linux-sunxi
<plaes>
esp8266 is using bsd-licensed software stack
diego_ is now known as diego71
<MoeIcenowy>
esp8089's driver is something strange...
<MoeIcenowy>
It will first add a dummy sdio driver to the driver list
<MoeIcenowy>
to check whether the card is probed by mmc_host
<MoeIcenowy>
then it's remove and the real driver is inserted
<MoeIcenowy>
after the first time the real driver initialized
<MoeIcenowy>
the driver will be unloaded and then loaded again
massi_ has quit [Remote host closed the connection]
<MoeIcenowy>
I got drunk.
vickycq has quit [Ping timeout: 244 seconds]
leoncito81_ has joined #linux-sunxi
<KotCzarny>
balmer peak coding?
apritzel1 has joined #linux-sunxi
vickycq has joined #linux-sunxi
leoncito81__ has joined #linux-sunxi
<plaes>
KotCzarny: legacy code review :P
leoncito81 has quit [Remote host closed the connection]
leoncito81__ has quit [Remote host closed the connection]
VargaD_ has joined #linux-sunxi
VargaD- has joined #linux-sunxi
VargaD- is now known as VargaD
<MoeIcenowy>
how to deal with "probe of mmc0:0001:1 failed with error -110" ...
VargaD_ has quit [Ping timeout: 276 seconds]
leoncito81__ has joined #linux-sunxi
VargaD has quit [Ping timeout: 260 seconds]
leoncito81__ has quit [Remote host closed the connection]
apritzel1 has quit [Ping timeout: 244 seconds]
Mr__Anderson has quit [Quit: Leaving.]
Amit_t_ has quit [Quit: Page closed]
VargaD has joined #linux-sunxi
VargaD_ has joined #linux-sunxi
VargaD has quit [Ping timeout: 260 seconds]
VargaD_ is now known as VargaD
leoncito81__ has joined #linux-sunxi
leoncito81__ has quit [Remote host closed the connection]
ricardocrudo has quit [Remote host closed the connection]
IgorPec11119 has joined #linux-sunxi
<MoeIcenowy>
Finally got a wlan1... Hooray.
<MoeIcenowy>
(wlan0 is r8712u usb wlan card on otg)
<MoeIcenowy>
Hooray!!! I got esp8089 work!!!
VargaD has quit [Ping timeout: 276 seconds]
<plaes>
MoeIcenowy: congratz
jernej has joined #linux-sunxi
<MoeIcenowy>
https://github.com/Icenowy/esp8089 contains my modified driver source (only deleted some includes specified to rk platform...) and a device tree for Aoson M751s
<MoeIcenowy>
I'll now go sleep. Good night everyone.
VargaD has joined #linux-sunxi
<MoeIcenowy>
wens: Here's another problem to a "general" q8 dt: some cards (such as esp8089) needs to be re-probed after boot, thus they needs "broken-cd;" in &mmc1 of dt
jstein has joined #linux-sunxi
apritzel has joined #linux-sunxi
p1u3sch1 has quit [Ping timeout: 250 seconds]
leoncito81 has joined #linux-sunxi
p1u3sch1 has joined #linux-sunxi
orly_owl has joined #linux-sunxi
leoncito81__ has joined #linux-sunxi
leoncito81__ has quit [Remote host closed the connection]
leoncito81 has quit [Remote host closed the connection]
Netlynx has quit [Quit: Leaving]
afaerber has quit [Quit: Ex-Chat]
apritzel has quit [Ping timeout: 244 seconds]
sherpa has joined #linux-sunxi
staplr has quit [Read error: No route to host]
vagrantc has quit [Ping timeout: 276 seconds]
vagrantc has joined #linux-sunxi
<KotCzarny>
wow, pxz | gzip results in ~58C on my banana
<KotCzarny>
first time it went over 48C
<willmore>
KotCzarny, got a nice heatsink on there?