Go Down

Topic: IDE causes high cpu / fan usage (Read 9667 times) previous topic - next topic

MatthewDriver

Hi WacKEDmaN,

I am seeing the same problem as you.


40% CPU utilisation by IDE/Java, mostly after a compile, sometimes before a compile. Then the CPU never drops back down.

System: Windows 10 Pro 64bit, i7, 8Gb, 512Gb SSD, HP notebook

I develop for four different chips: ATTINY45, ATMEGA328P, ATMEGA1284P, ESP8266.

To make the board configurations simpler I have a portable IDE for each chip.

I discovered that only the IDE's on version 1.8.7 had the high CPU problem and those on 1.8.5 were fine.

If I had two IDE's running that were both 1.8.7, I would end up with 80% CPU utilisation (until I closed the IDE's).

My solution for the moment is to roll back all my environments to 1.8.5.

I have not tried 1.8.6.

I total agree there is a bug in 1.8.7.

Cheers,

Matthew.




cb88

#46
Oct 23, 2018, 04:26 pm Last Edit: Oct 23, 2018, 04:31 pm by cb88
@ballscrewbob stop telling people the problem is them. I'm a developer and there is obviously a bug in the Arduino code. I've read you dismissing people's problems like this for many months and it is very insulting.

I've attached a photo of this occuring on my PC you can download JProfiler with a trial license and get the same thing, this is the call graph that is generated by profiling the cpu hogging JVM. This occurs right after hitting compile. It appears to be a broken graphics event handler...

This has absolutely nothing to do with any other thing on the computer, the rest of the OS and all other programs are completely idle. I also have a very light install of windows 7 that runs very well. That said any suggestion that it could ever have been any of those things, when you close out Arduino and it stops is blatantly bogus. Also you apparently have no idea how anti virus software works... there is no active scanning as you type going on etc... the arduino software doesn't even do anything remotely suspicious anyway except perhaps when it checks for updates.

Such suggestions are fine perhaps the first time you hear about a problem... but as I can see with a quick forum search you've been giving these response to people dealing with this Arduino bug for many months at least since January... stop!

ballscrewbob

@cb88

It is NOT the people using the IDE that is the problem that suggestion is clearly BOGUS on your part.
There are a MULTITUDE of possible causes and you as a "developer" will be extremely aware of that.

As far as AV causing issues with CPU spikes that case has been proven on multiple occasions and not just from my advice.
A forum search will bear that out.

Do you have a precise answer ?
Are you offering any forms of alternative answer apart from those offered by many besides myself  ?

I already stated on at least two occasions in the forums that there is a SPIKE WHEN YOU HIT compile.
That in most cases tails off quite quickly.

I can get that to occur on most of the IDE's including NON ARDUINO and NON JAVA based programs which you as a dev will also be aware of.

if you wish to take this further you are welcome to go PM rather than direct a personal attack.



It may not be the answer you were looking for but its the one I am giving based on either experience, educated guess, google or the fact that you gave nothing to go with in the first place so I used my wonky crystal ball.

WacKEDmaN

#48
Oct 23, 2018, 10:03 pm Last Edit: Oct 23, 2018, 10:52 pm by ballscrewbob
Hi WacKEDmaN,

I am seeing the same problem as you.


40% CPU utilisation by IDE/Java, mostly after a compile, sometimes before a compile. Then the CPU never drops back down.


glad its not just me...mine does it randomly tho..it'll just start this 'race' condition out of the blue.. compile usually does spike the CPU as its doing work..

but i have found something interesting.. i go File -> Save the 'race' condition stops :) ..it still kicks in randomly tho later on tho :(

EDIT: and just after i posted this i triggered the 'race' condition!... simply minimizing the IDE window to taskbar (serial monitor window open in a window).. again, Save did the trick to stop the 'race'...(cant save when minimized tho!)

what say you now BALLSCREWBOB < profanity adjusted

ballscrewbob

Apples and oranges.

Almost no two systems are the exact same be it hardware, OS, installed programs etc.
What can I tell you I see a spike at the start of a compile but that draws down quite rapid.

It is an age old thing that crops up once in a blue moon on here but some take it as the worst thing since the last ice age. ;)

It may not be the answer you were looking for but its the one I am giving based on either experience, educated guess, google or the fact that you gave nothing to go with in the first place so I used my wonky crystal ball.

WacKEDmaN

....theres nothing to compare ..the common theme here is the IDE..not hardware or software or whatever else you want to come up with today...just coz you are not seeing it doesnt mean its not there

we are trying to explain a problem(i specifically signed up to post about this issue), possibly find the cause, and either work around the issue or find a fix.. but you keep providing the same old same old..i suggest you just moderate the forum and keep out of this thread if you cannot provide any new info... SMH...

MatthewDriver

Hi WacKEDmaN,

What IDE version are you seeing this problem with and have you tested any others?

I see the fault with 1.8.7
I don't see the fault with 1.8.5

Cheers,

Matthew.



Coding Badly


Are there any Java developers in the pool of people having CPU usage trouble?


WacKEDmaN

Hi WacKEDmaN,

What IDE version are you seeing this problem with and have you tested any others?

I see the fault with 1.8.7
I don't see the fault with 1.8.5

Cheers,

Matthew.



yup 1.8.7...i might roll back.. these fans are getting annoying while im trying to code...for no good reason at all!

ballscrewbob

A short video of CPU load

Security is OFF for this test.
A small amount of settle time was allowed but is not a true test.

The initial low is almost system idle.

The first PEAK is starting the screen recorder to avoid being accused of picking and choosing a screenshot.
(AVG 15% CPU load)

The second peak is startup of IDE 1.8.7.
(NOTE this will add to the cumulative JAVA and CPU load)

Third peak is COMPILE. This will peak out at all 4 cores but that will tail rapidly as can be seen in the video.

I can reproduce similar results with multiple different programs many NON JAVA.

Did confirm the observation that a SAVE from the IDE will drop the java load even further to an average of around 7%

Why a VIDEO ?  It shows how easy it is to get CPU load to peak andf in this case it also shows OPTIMISED cores as being a NON BS factor.

Hope to have a second video up shortly with security enabled to show the effects it also has and put to bed some misconceptions.



It may not be the answer you were looking for but its the one I am giving based on either experience, educated guess, google or the fact that you gave nothing to go with in the first place so I used my wonky crystal ball.

ballscrewbob

CPU LOAD with security enabled

Security is ON for this test. (Avast and MBAM)
A small amount of settle time was allowed but is not a true test.
IDE 1.8.7 once again used and as before a system clean up had been performed to clear out possible cache issues etc.

This time I also extended the process selection panel to get a better idea of what is peaking and not.

The initial low is almost system idle but the load added by security can clearly be seen along with the differing CORE loads !

The first PEAK is starting the screen recorder, again to avoid being accused of picking and choosing a screenshot.
(AVG 12% CPU load)
The second peak is startup of IDE 1.8.7.
(NOTE this will add to the cumulative JAVA and CPU load)
Note additionally the the screen recorder often occupies the top position above Arduino

Third peak is COMPILE. This will peak out at all 4 cores but that will tail rapidly once again.

It may not be the answer you were looking for but its the one I am giving based on either experience, educated guess, google or the fact that you gave nothing to go with in the first place so I used my wonky crystal ball.

MatthewDriver

I have been experimenting with this fault and have discovered a few things.

Firstly, let's define the fault I'm referring to:

When the Windows Arduino IDE is running, the javaw.exe uses approximately 35% CPU usage.
If left unattended, it will never drop.

I have tested difference IDE builds.

The following versions show this fault: 1.8.6 and 1.8.7
The following versions do not show this fault: 1.8.5, Hourly build 25 Oct and Beta 1.9.

Stopping and starting:
The high CPU can be stopped by click File\Save (discovered by WacKEDmaN).
The high CPU can be started by click moving or resizing the Arduino IDE window.

So there is hope that this fault has already been identified and will not exist in the next stable release.

For now, I'll stick with using 1.8.5.

Anyone that already his this fault is welcome to reproduce these finding and confirm it or provided an alternate view.

Cheers,

Matthew.



WacKEDmaN

#57
Oct 27, 2018, 03:25 pm Last Edit: Oct 27, 2018, 06:07 pm by ballscrewbob
that sums up my issue exactly MatthewDriver..can confirm that is the behavior im seeing...

..it would be nice if we didnt have to go around in circles to get there!...

ill be going back to 1.8.5 myself until there is a new update..

Coding Badly

Anyone that already his this fault is welcome to reproduce these finding and confirm it or provided an alternate view.
Confirmed exactly as described.

Is there an issue on Github for this exact problem?


MatthewDriver

Hi Coding Badly,

I have check https://github.com/arduino/Arduino/issues/ back until September 2017 and can not see this specific issue mentioned.

I can't find a list of version release dates, so not sure how far back to look.

Perhaps it's only impacting a minority of users and another bug fix has already fixed this issue in the beta and hourly build.

Matthew.

Go Up