Electron CPU usage

[wtidwell]

@TiemenMoesker I tried two pies with exactly the same result. My spikes are way higher than 55%, on the order of max processing.

[A Former User]

I have a very stupid question.
In my new ATB, transition animation caused CPU usage around max 70%.(by nmon. htop shows less than that… around 30~50%). Without transition, near 10~20%.
But… I had not felt the performance issue before this thread. Will it be a real problem? I’m not an expert, but I think, temporally CPU usage could be high than we’ve thought. Should I really worry about this?

[djsunrise19]

@wtidwell said in Electron CPU usage:

@TiemenMoesker I tried two pies with exactly the same result. My spikes are way higher than 55%, on the order of max processing.

Hm, okay. But that seems to be a little bit weird. Imagine there are f.e. 8 modules. 3 of them make a fade at the same time (compliments, feed and calendar / netatmo, …). Then max processing is over 100%. Thats the problem in my first screenshot. 320% max cpu. This can not be right.

[djsunrise19]

@Sean said in Electron CPU usage:

I have a very stupid question.
In my new ATB, transition animation caused CPU usage around max 70%.(by nmon. htop shows less than that… around 30~50%). Without transition, near 10~20%.
But… I had not felt the performance issue before this thread. Will it be a real problem? I’m not an expert, but I think, temporally CPU usage could be high than we’ve thought. Should I really worry about this?

My ATB is coming today. I’ll made a new installation with all my modules and settings from my actual raspi. I’m very excited and I’ll give a report.

All in all it seems crazy to me, that those transitions are so cpu heavy. And in some cases (look at those nmon screenshots here) the cpu load is way to high and the mirror becomes unusable. So in that case, we should worry about that.

[Mykle1]

@Snille said in Electron CPU usage:

The was originally @Snille idea!

Run in Pi3 terminal (or in SSH): cat /proc/cpuinfo
Compare hardware and revision… :)
Hardware : BCM2835
Revision : a02082

Pi 3 Model B 1GB a02082 (Sony, UK)
Pi 3 Model B 1GB a22082 (Embest, China)

Does anyone with Hardware :BCM2709 have this problem?

Does anyone with Hardware :BCM2835 NOT have this problem?

[TiemenMoesker]

@Mykle1

Maybe we are on to somthing here.
My hardware =BCM2835
Revision= a02082

And I DO have this problem.

[strawberry 3.141]

if i remember correct there is also a possibility to increase performance of transformations, but I’m not sure. Have to research on this

EDIT: I can’t do any testing for a while, if someone wants to get a hand on it I think this should make an effect as it will optimize the rendering from cpu usage to gpu. It’s still a new stuff though and I’m not sure if it is supported in electron yet.

.MODULENAME {
  will-change: opacity;
}

replace modulename with the actual module name

[Mykle1]

@TiemenMoesker said in Electron CPU usage:

Maybe we are on to somthing here.
My hardware =BCM2835
Revision= a02082
And I DO have this problem.

My hardware = BCM2709

I do NOT have this problem

I’ve long suspected that this was/is a hardware issue. Why? I’ve done nothing special or different to my Pi 3 (except for adding a fan), yet it runs flawlessly, smoothly, and without the cpu overload experienced by a few select users. Assuming they have done nothing out of the ordinary (most have done fresh installs of the OS and default install of MM) yet their Pi 3’s exhibit inordinate amounts of cpu load at rest and extreme cpu load with just a simple function like the compliments changing.

@strawberry-3-141 There’s no need for me to try your suggestion but perhaps someone experiencing this problem will test it and report back their findings.

[TiemenMoesker]

@strawberry-3.141

Ill hook up my PI in a sec and test it. I am supposed to put that line of code in the main.css file right? (Just to be sure)

[Mykle1]

@TiemenMoesker

custom.css probably