Greetings!
This thread may be related, though it's using Bookworm instead of Buster.
viewtopic.php?t=376584
System:
A Pi-4 4g, running Buster on a GoPiGo robot, using the default desktop and VNC to access the robot.
I have installed the Adafruit 128x32 2.23" OLED Bonnet and the associated drivers, including a test-script called Stats.py, as noted here https://learn.adafruit.com/adafruit-2-2 ... led-bonnet.
I am assuming that this version of Buster, (that is specifically configured for the robot), is based on the Super-Deluxe, Action-Packed, Full-Monty version of Raspberry Pi O/S with all the raspberries, whipped-cream, sprinkles and cherries on top.
I am running Chromium "Version 92.0.4515.98 (Official Build) Built on Raspbian , running on Raspbian 10 (32-bit)".
=================================================
Issue:
I am testing an Adafruit 128x32 2.23" OLED Bonnet for the Raspberry Pi on my GoPiGo robot as a (semi) real-time status display.
One of the things this status display shows is the CPU load factor taken from "top" - and it was horrible! (Right now it's over 2.0 on a 4g Pi-4 running Buster.)
Of course, I immediately suspected the status python script of hogging all the processors (and actually opened a topic over on the Adafruit forums about it), until I discovered that it didn't seem to matter if the status script was running or not. (Then I deleted the thread. . .)
Further investigation disclosed that the culprit seems to be Chromium.
* When Chromium launches it spawns several dozen processes. (!!!)
* If I launch a web page like this one, and begin interacting with it - by typing into a forum thread for example - the load factor climbs to over 2.0. Sometimes well over 2.0
* If I stop doing anything, the load factor - very, very, slowly - drops to about 0.9/0.8.
* Completely stopping Chromium drops the processor usage to below 0.5.
* Viewing the system stats using htop, I notice that Chromium, (and init), are using the lion's share of the processor, with occasional visits from pigpiod.
I noticed that closing Chromium causes the processor load to (very slowly) drop to reasonable values.
I have also noticed that, though Chromium seems to be the chief offender, the load factor seems to vary over relatively wide limits, (plus or minus 0.5), for no seeming reason whatsoever at times.
In the past I haven't really worried about it - it wasn't on my radar to be honest - until I shared a picture of my robot on the Dexter Industries forums with the status display showing a processor load of 1.5-something and one of my friends there asked me if I was crypto-mining!
I've also tried this on an 8g Pi-4 with the same results.
It should also be noted that there are significant differences in the way our respective robots are configured - his is headless and he doesn't use a desktop, whereas I use the desktop almost exclusively. Because of this, he doesn't use graphical tools, (gparted, Thonny, Chromium, pcmanfm, etc. etc. etc.), and does all the graphical stuff on his remote system.
===================================================
Questions:
1. Why does Chromium seem to be such a resource hog? Is there anything I can do to help mitigate this?
2. Is there that much of a difference running headless as opposed to running Buster with what I am assuming is "The Full Monty" of installed goodies?
3. What else can I do to try to find the resource hogs on the system? I figure that I can't fix the problem until I know exactly what it is.
Thanks!
This thread may be related, though it's using Bookworm instead of Buster.
viewtopic.php?t=376584
System:
A Pi-4 4g, running Buster on a GoPiGo robot, using the default desktop and VNC to access the robot.
I have installed the Adafruit 128x32 2.23" OLED Bonnet and the associated drivers, including a test-script called Stats.py, as noted here https://learn.adafruit.com/adafruit-2-2 ... led-bonnet.
I am assuming that this version of Buster, (that is specifically configured for the robot), is based on the Super-Deluxe, Action-Packed, Full-Monty version of Raspberry Pi O/S with all the raspberries, whipped-cream, sprinkles and cherries on top.
I am running Chromium "Version 92.0.4515.98 (Official Build) Built on Raspbian , running on Raspbian 10 (32-bit)".
=================================================
Issue:
I am testing an Adafruit 128x32 2.23" OLED Bonnet for the Raspberry Pi on my GoPiGo robot as a (semi) real-time status display.
One of the things this status display shows is the CPU load factor taken from "top" - and it was horrible! (Right now it's over 2.0 on a 4g Pi-4 running Buster.)
Of course, I immediately suspected the status python script of hogging all the processors (and actually opened a topic over on the Adafruit forums about it), until I discovered that it didn't seem to matter if the status script was running or not. (Then I deleted the thread. . .)
Further investigation disclosed that the culprit seems to be Chromium.
* When Chromium launches it spawns several dozen processes. (!!!)
* If I launch a web page like this one, and begin interacting with it - by typing into a forum thread for example - the load factor climbs to over 2.0. Sometimes well over 2.0
* If I stop doing anything, the load factor - very, very, slowly - drops to about 0.9/0.8.
* Completely stopping Chromium drops the processor usage to below 0.5.
* Viewing the system stats using htop, I notice that Chromium, (and init), are using the lion's share of the processor, with occasional visits from pigpiod.
I noticed that closing Chromium causes the processor load to (very slowly) drop to reasonable values.
I have also noticed that, though Chromium seems to be the chief offender, the load factor seems to vary over relatively wide limits, (plus or minus 0.5), for no seeming reason whatsoever at times.
In the past I haven't really worried about it - it wasn't on my radar to be honest - until I shared a picture of my robot on the Dexter Industries forums with the status display showing a processor load of 1.5-something and one of my friends there asked me if I was crypto-mining!
I've also tried this on an 8g Pi-4 with the same results.
It should also be noted that there are significant differences in the way our respective robots are configured - his is headless and he doesn't use a desktop, whereas I use the desktop almost exclusively. Because of this, he doesn't use graphical tools, (gparted, Thonny, Chromium, pcmanfm, etc. etc. etc.), and does all the graphical stuff on his remote system.
===================================================
Questions:
1. Why does Chromium seem to be such a resource hog? Is there anything I can do to help mitigate this?
2. Is there that much of a difference running headless as opposed to running Buster with what I am assuming is "The Full Monty" of installed goodies?
3. What else can I do to try to find the resource hogs on the system? I figure that I can't fix the problem until I know exactly what it is.
Thanks!
Statistics: Posted by jharris1993 — Sat Oct 19, 2024 4:44 pm — Replies 0 — Views 33