Explorer.exe taking up 15-30% CPU usage when moving mou

H2SO4 said:

EDIT: Hold on a tick.... I just realised something :)

Try these steps please and tell me what you see:

0) Fire up task manager and sort by the (badly named) "CPU" column under the processes tab. Put the task manager window somewhere out of the way on your monitor. Exit all apps, all windows, so that the task manager is the only thing on your task bar and you're looking at your (non-animated) desktop.

1) Move the mouse around on the desktop without going over the TM window - what process is consuming processor time?

2) Now start an Explorer window and go into a folder with lotsa little files. Move the pointer around over the files so the blue highlight bar moves under the pointer. What process is eating processor time?

3) Close the explorer window and open a browser instance - IE or FF. Go to a busy page with lots of elements on it. Don't click anything, just move the mouse around within that window. What process is eating processor?

It just struck me (before writing this response) that you really shouldn't be seeing explorer.exe as being responsible unless you're moving your mouse over an explorer window.

Can you confirm what happens on your machine during the steps above?

logicearth said:

kevindd992002 said:

mouse movement DOESN'T add anything to the CPU usage.

Mouse movement always adds to CPU usage...how else do you think the computer will know where the mouse is if the CPU doesn't calculate it?

Btw you do not need to undervolt your processor. Windows is able to that by using Intel's Speedstep. Go into Power Management > Advance.

Intel's Speedstep throttles the CPU's FREQUENCY not the voltage.

H2SO4 said:

H2SO4 said:

EDIT: Hold on a tick.... I just realised something :)

Try these steps please and tell me what you see:

0) Fire up task manager and sort by the (badly named) "CPU" column under the processes tab. Put the task manager window somewhere out of the way on your monitor. Exit all apps, all windows, so that the task manager is the only thing on your task bar and you're looking at your (non-animated) desktop.

1) Move the mouse around on the desktop without going over the TM window - what process is consuming processor time?

2) Now start an Explorer window and go into a folder with lotsa little files. Move the pointer around over the files so the blue highlight bar moves under the pointer. What process is eating processor time?

3) Close the explorer window and open a browser instance - IE or FF. Go to a busy page with lots of elements on it. Don't click anything, just move the mouse around within that window. What process is eating processor?

It just struck me (before writing this response) that you really shouldn't be seeing explorer.exe as being responsible unless you're moving your mouse over an explorer window.

Can you confirm what happens on your machine during the steps above?

Answers:

1) explorer.exe is averaging at 10% and as soon as I stop moving the mouse it drops to 0%.

2) explorer.exe is averaging at 7% and as soon as I stop moving the mouse it drops to 0%.

3) iexplore.exe is averaging at 40% and as soon as I stop moving the mouse it drops to 0%.

kevindd992002 said:

Answers:

1) explorer.exe is averaging at 10% and as soon as I stop moving the mouse it drops to 0%.

2) explorer.exe is averaging at 7% and as soon as I stop moving the mouse it drops to 0%.

3) iexplore.exe is averaging at 40% and as soon as I stop moving the mouse it drops to 0%.

(2) is normal, (1) is borderline, and (3) seems somewhat excessive - my IE does ~20% on a relatively busy page (www.smh.com.au).

At least it shows that your issue is not specific to explorer.exe. Any process whose window(s) is being pelted with WM_MOUSEMOVE messages will consume processor time to deal with the mouse movement in whatever way it sees fit.

What surprises me is that I cannot spot evidence of 3rd-party code or driver interference in your ETLs. Virtually all of that activity is in user-mode, and it's all expected and necessary: sending window messages, invoking window procedures, checking the owner of a particular rectangle... It's just that it all happens slower than it should, given your specs.

Obviously something improved around the time you removed RMclock because the utilisation dropped so drastically. If you're still encountering an unequivocal improvement when running in safe mode, I'd suggest a little basic safe mode troubleshooting:

Fire up MSCONFIG, choose a "diagnostic" startup, reboot... any improvement?

H2SO4 said:

kevindd992002 said:

Answers:

1) explorer.exe is averaging at 10% and as soon as I stop moving the mouse it drops to 0%.

2) explorer.exe is averaging at 7% and as soon as I stop moving the mouse it drops to 0%.

3) iexplore.exe is averaging at 40% and as soon as I stop moving the mouse it drops to 0%.

(2) is normal, (1) is borderline, and (3) seems somewhat excessive - my IE does ~20% on a relatively busy page (www.smh.com.au).

At least it shows that your issue is not specific to explorer.exe. Any process whose window(s) is being pelted with WM_MOUSEMOVE messages will consume processor time to deal with the mouse movement in whatever way it sees fit.

What surprises me is that I cannot spot evidence of 3rd-party code or driver interference in your ETLs. Virtually all of that activity is in user-mode, and it's all expected and necessary: sending window messages, invoking window procedures, checking the owner of a particular rectangle... It's just that it all happens slower than it should, given your specs.

Obviously something improved around the time you removed RMclock because the utilisation dropped so drastically. If you're still encountering an unequivocal improvement when running in safe mode, I'd suggest a little basic safe mode troubleshooting:

Fire up MSCONFIG, choose a "diagnostic" startup, reboot... any improvement?

I'll try that now. So, this is just software problem right? Never with hardware?

EDIT: Tried Diagnostic mode, no improvement, moving the mouse in the desktop still yielded an average of 10% CPU usage.

kevindd992002 said:

H2SO4 said:

Fire up MSCONFIG, choose a "diagnostic" startup, reboot... any improvement?

I'll try that now. So, this is just software problem right? Never with hardware?

Depends on how you look at it. The only way that different safe mode behaviour would be compatible with a "broken hardware" diagnosis is if the different (simpler) software environment in safe mode somehow failed to expose a hardware flaw. In your case, the hardware may not be outright broken though...

My gut feeling is that your hardware is somehow getting throttled in normal mode, but not in safe mode. Another mechanism - just like RMclock which you've now removed - is getting invoked on startup in normal mode and causing the processor to go a little more comatose. If that wild guess is even remotely accurate, running the same benchmark in safe and normal modes should produce very different results.

kevindd992002 said:

EDIT: Tried Diagnostic mode, no improvement, moving the mouse in the desktop still yielded an average of 10% CPU usage.

I'd like to see just how your machine performs during simple tasks that have nothing to do with the mouse. Try this please...

Paste the following into a batch file, or just use the one I uploaded here:

Code:

@echo off
echo Starting...
echo |time |find "time is"
for /L %%i IN (1,1,100000) DO @echo %i >nul
echo Finished!
echo |time |find "time is"

Run that batch file in both safe and normal modes and paste back the results please. They'll look like this:

N:\>stressbat.bat
Starting...
The current time is: 20:55:15.93
Finished!
The current time is: 20:55:49.24

ive noticed the same...i can get one of my cores to 100% just by simply movin my mouse cursor rly fast over my desktop....the process havin 100% cpu usage is...n the winner is: explorer.exe

the same thing is bein discussed here as well: http://forums.logitech.com/t5/G-Seri...age/m-p/278283

Did you modify your registry key labeled mousehovertime?

kevindd992002 said:

Did you modify your registry key labeled mousehovertime?

no i havent...is that thing gonna fix it???