Longsleep magic !

[Lutinmalin]

Hi
Just wanted to show you guys something interesting.

For the past days I was running Debian image from 2016-05-01 on my 1GB board and tonight I update the kernel to 3.10.101 with the pine64_update_kernel.sh script.
Then I noticed this on pi-monitor (thanks to tkaiser for that, I've been reading a lot of your stuff here and on other forums):



That can't be just a coincidence

I guess it illustrates the efforts guys like Longsleep put on to make things better, so thank you very much !

by the way for people wondering what is my cooling setup, I have 2 heatsinks salvaged from an old motherboard and a fan right above. Before putting heatsink when running a stress test I was reaching 80 celcius and cooling stage 1, now It's stable at 57 celcius under full load.



Not a lot of space between the heatsink and the fan but I had not found the heatsink yet when I built the enclosure, I though I was to use smaller heatsink like those cheap copper we see everywhere on ebay but these 2 are doing a great job. It may be a bit overkill but I like the results look and performance wise.

[tkaiser]

That can't be just a coincidence

Sorry but by looking at the small graph you got these temperatures before already and they increased starting 8 hours before maybe caused by some background activity (can be also seen at the cpufreq graph that the system wasn't completely idle but even increased clockspeed from time to time!). So what you're experiencing is what happens after rebooting. Unwanted background activity has gone.

We improved the thermal settings in March so only OS images that are horribly outdated (the crappy XFCE Arch image the Pine64 folks provided a long time for example) could benefit from updates now in this area (both better temperature AND performance behaviour).

But in case you would use an OS image made in early March and would now update kernel (and more importantly .dts file used by u-boot/kernel -- here the THS stuff lives inside) then you could really see such improvements.

[Lutinmalin]

Yeah this morning when I woke up I remembered that I might have left a mate ressource monitor opened in a xrdp session before the reboot 

But I'm still thankful for your work 


Have a nice day

[tkaiser]

Yeah this morning when I woke up I remembered that I might have left a mate ressource monitor opened in a xrdp session before the reboot 

But that's the nice thing when you already enabled monitoring: All data sources to get a clue what happened are in round robin databases and can be queried or used for graphing. And again: The conclusion you draw (community members improved thermal settings and after updating kernel now the boards run both faster and stay more cool -- might not be true for Android/RemixOS that still use outdated/bad Allwinner settings) is true. But it happened already in March.

[texadactyl]

Yeah this morning when I woke up I remembered that I might have left a mate ressource monitor opened in a xrdp session before the reboot 

But that's the nice thing when you already enabled monitoring: All data sources to get a clue what happened are in round robin databases and can be queried or used for graphing. And again: The conclusion you draw (community members improved thermal settings and after updating kernel now the boards run both faster and stay more cool -- might not be true for Android/RemixOS that still use outdated/bad Allwinner settings) is true. But it happened already in March.

What Linux package are you using for temperature monitoring?  Thanks.

[Lutinmalin]

Yeah this morning when I woke up I remembered that I might have left a mate ressource monitor opened in a xrdp session before the reboot 

But that's the nice thing when you already enabled monitoring: All data sources to get a clue what happened are in round robin databases and can be queried or used for graphing. And again: The conclusion you draw (community members improved thermal settings and after updating kernel now the boards run both faster and stay more cool -- might not be true for Android/RemixOS that still use outdated/bad Allwinner settings) is true. But it happened already in March.

What Linux package are you using for temperature monitoring?  Thanks.

I'm using RPI Monitor configured for Pine64
I got that from tkaiser at the following url: http://linux-sunxi.org/User:Tkaiser#Reli..._on_Pine64
you can just WGET the script http://kaiser-edv.de/tmp/4U4tkD/install-...for-a64.sh run it and then access the web interface using <ip of pine64>:8888

[texadactyl]

Yeah this morning when I woke up I remembered that I might have left a mate ressource monitor opened in a xrdp session before the reboot 

But that's the nice thing when you already enabled monitoring: All data sources to get a clue what happened are in round robin databases and can be queried or used for graphing. And again: The conclusion you draw (community members improved thermal settings and after updating kernel now the boards run both faster and stay more cool -- might not be true for Android/RemixOS that still use outdated/bad Allwinner settings) is true. But it happened already in March.

What Linux package are you using for temperature monitoring?  Thanks.

I'm using RPI Monitor configured for Pine64
I got that from tkaiser at the following url: http://linux-sunxi.org/User:Tkaiser#Reli..._on_Pine64
you can just WGET the script http://kaiser-edv.de/tmp/4U4tkD/install-...for-a64.sh run it and then access the web interface using <ip of pine64>:8888

Thanks.  Will look into those references.

Also, I just looked in pine64_health.sh and saw a simple way to gather hardware information.  Very similar to what psensor does for AMD/Intel CPUs.