09:27 <jernej> DarkNeutrino: I only have OrangePi Zero2 and T95 box. Aren't OPP tables same?

09:29 <DarkNeutrino> Not really. There are 4 variants of h616. 2 of them have higher mA to account for the worse chips and i also need to check the boost clocks which h616 supports

09:35 <jernej> this sounds similar to H6, where there are 3 bins and driver selects correct OPP table based on the bin

09:35 <jernej> I just looked at dumped DT from T95 and there are two OPP tables combined into one

09:36 <jernej> I can upload this file if you with

09:36 <DarkNeutrino> That would be great

09:37 <jernej> http://ix.io/2MBN

09:51 <DarkNeutrino> yep that looks to be with the revisions as well. lovely. tho there are also boost clocks. i saw one opp table that contained them. have to find itn

09:53 <DarkNeutrino> since we will need ths and sid before tho i will add it to bindings and add their configs to drivers. then we can add opp tables. tho i havent found a table for gpu yet. which you mentioned you also wanted

13:12 <apritzel> DarkNeutrino: with boost clocks you mean rates beyond 1.5 GHz?

13:14 <DarkNeutrino> yep h616 supports those via turbo mode

13:14 <DarkNeutrino> so they are disabled normally but once you enable turbo mode they will activate and cpu will scale beyond 1.5Ghz

13:15 <apritzel> "once you enable turbo mode"> who does that? and when?

13:15 <DarkNeutrino> nobody does it. its up to user

13:15 <DarkNeutrino> they do not get activated automatically

13:16 <apritzel> so is there already kernel code to let users activate "advanced" OPPs?

13:17 <DarkNeutrino> yea. let me check the file you can write 1 to enable it. dont remember the path from head

13:19 <DarkNeutrino> - turbo-mode: Marks the OPP to be used only for turbo modes. Turbo mode is available on some platforms, where the device can run over its operating frequency for a short duration of time limited by the device's power, current and thermal limits.

13:20 <DarkNeutrino> and the file ....

13:20 <DarkNeutrino> is /sys/devices/system/cpu/cpufreq/boost

13:21 <DarkNeutrino> write 1 to that and you will activate the "advanced" opp tables

13:27 <DarkNeutrino> at least thats the normal cpufreq-dt driver

13:28 <DarkNeutrino> im wondering if to use that one and adjust opp tables to account for the other binned chips or do it the h6 way where we read the efuse value and adjust opp according to that

13:30 <apritzel> have you figured out what the opp-supported-hw value means? Is that just some kind of bin indicator?

13:30 <DarkNeutrino> yes

13:30 <DarkNeutrino> there are 4 bins

13:31 <DarkNeutrino> 2 of those use the opp table with bumped up mA but same clocks

13:33 <apritzel> bumped up mA? And why do bins 1 and 2 only provide one OPP?

13:34 <DarkNeutrino> wdym provide 1 opp ?

13:36 <apritzel> I mean in this table that jernej linked to

13:36 <DarkNeutrino> as for the bumped up mA. thats kinda obvious no? the chip is not stable at those clocks with lower power so they added a bit more to it, to make it stable

13:36 <apritzel> you mean voltage?

13:37 <DarkNeutrino> omg i say mA.... my bad. mV

13:37 <DarkNeutrino> as for the 1 2 bin no clue

13:37 <apritzel> right, that somewhat confused me ;-)

13:37 <DarkNeutrino> yea sorry :)

13:38 <DarkNeutrino> its as if the first and second bin didnt have scaling

13:38 <DarkNeutrino> so they just run at 1 set speed

13:38 <apritzel> yeah, still kind of odd

13:39 <apritzel> do you have an idea what bins are out there?

13:39 <apritzel> have you tried to read the value on your boards?

13:39 <DarkNeutrino> didnt have the chance yet no

13:39 <HackerKkillinghisLGA775cpuPent> <ungeskriptet[m]> "Hi all, does someone know if the..." <- maybe

13:41 <DarkNeutrino> mainly cause the kernel provided by biqu is kinda meh and they didnt even bother with the bins

13:41 <DarkNeutrino> they dont even read the value from the efuse

13:42 <HackerKkillinghisLGA775cpuPent> test

13:42 <DarkNeutrino> HackerKkillinghisLGA775cpuPent: test works

13:42 <HackerKkillinghisLGA775cpuPent> DarkNeutrino: thank

14:26 <DarkNeutrino> h616 sid and thermal sensor https://github.com/SoMainline/linux/commits/martin/cb1-thermal-opp

14:28 <DarkNeutrino> jernej: apritzel: would love to hear if you guys (or anyone else really) see anything wrong with it ?

14:30 <HackerKkillinghisLGA775cpuPent> ungeskriptet[m]: https://linux-sunxi.org/Format_Q8

14:30 <HackerKkillinghisLGA775cpuPent> HackerKkillinghisLGA775cpuPent: ungeskriptet pls took a look of this

14:49 <jernej> DarkNeutrino: post it and see? :)

14:50 <DarkNeutrino> thats... not a bad point. will add the thermal table and etc and will post

15:10 <jernej> DarkNeutrino: fyi, if SID is same as in A64, as your commits suggests, then you should add compatible pair (first h616 specific, then a64 one) in bindings and you don't need any driver changes.

15:35 <DarkNeutrino> gotcha