User CPU time

classic Classic list List threaded Threaded
6 messages Options
Reply | Threaded
Open this post in threaded view
|

User CPU time

Febri Rochel
Dear all,

thank you very much for your previous reply. However I'm still quite confuse about the units of User CPU time in Experimenter. You said that User CpU unit is in nanoseconds but then if you transform it to seconds to would have to divide it by "10E6" then wouldn't that be microseconds and not nanoseconds?
if i have a value of 0.04 for the user cpu time what is it in seconds?

Could you please give me a concrete definition for the user CPU time? what are we measuring here exactly? and what is the difference with cpu user time and measure time?

I would like to compare different feature selection methods by their time consumption with respect to the time consumption of a classifier without feature selection. Could you please suggest on which values i should use? user CPU or measure time?

thank you

>> After opening the result of the experimenter in Excel (CSV format) I
>> noticed that in Measure Time (and SelectionTime) when using a classifier
>> without feature selection, it gives a '?' result. I'm guessing that
>> SelectionTime might be the time measured to do the feature selection and
>> measure time is perhaps the total of classifier + feature selection time. Is
>> this correct? if yes, what is the unit of this measurement (microseconds or
>> seconds)
>
> Yes, this is correct. The unit of measurement is milliseconds.
>
>>
>> Could you please tell me the units for 'UserCPU Time' as well?
>
> Nanoseconds. In the Javadoc for the method (getCurrentThreadUserTime) of the
> class (ThreadMXBean) that is used to get this value, it states "Returns the
> CPU time that the current thread has executed in user mode in nanoseconds.
> The returned value is of nanoseconds precison but not necessarily
> nanoseconds accuracy."

All of them should get transformed into seconds. "Elapsed_time_*" gets
divided by 1000 and the "CPU*" gets divided by 10E6.

Cheers, Peter
-- 


_______________________________________________
Wekalist mailing list
[hidden email]
https://list.scms.waikato.ac.nz/mailman/listinfo/wekalist
Reply | Threaded
Open this post in threaded view
|

Re: User CPU time

Peter Reutemann
> thank you very much for your previous reply. However I'm still quite confuse
> about the units of User CPU time in Experimenter. You said that User CpU
> unit is in nanoseconds but then if you transform it to seconds to would have
> to divide it by "10E6" then wouldn't that be microseconds and not
> nanoseconds?

Sorry, CPU time first gets divided by 1000 and then by 10E6. Wasn't
clear from what I've written. The most important part from my post
was, that the time is *always* in seconds.

> if i have a value of 0.04 for the user cpu time what is it in seconds?
> Could you please give me a concrete definition for the user CPU time? what
> are we measuring here exactly? and what is the difference with cpu user time
> and measure time?

UserCPU time is what the thread used in actual CPU cycles, hence most
accurate. Elapsed time is quite inaccurate, since it's only the
difference between the system's time (other processes/threads could
steal valuable processing time!).

> I would like to compare different feature selection methods by their
> time consumption with respect to the time consumption of a classifier
> without feature selection. Could you please suggest on which values i should
> use? user CPU or measure time?

UserCPU.

Cheers, Peter
--
Peter Reutemann, Dept. of Computer Science, University of Waikato, NZ
http://www.cs.waikato.ac.nz/~fracpete/ Ph. +64 (7) 858-5174

_______________________________________________
Wekalist mailing list
[hidden email]
https://list.scms.waikato.ac.nz/mailman/listinfo/wekalist
Reply | Threaded
Open this post in threaded view
|

Re: User CPU time

Febri Rochel
In reply to this post by Febri Rochel
Hi, 

"that the time is *always* in seconds"?
meaning that if I get a value of 0.4 in the user CPU this means that the time used for the CPU to process it is 0.4*10E-9?

thank you

Best Regards,

Febri


> thank you very much for your previous reply. However I'm still quite confuse
> about the units of User CPU time in Experimenter. You said that User CpU
> unit is in nanoseconds but then if you transform it to seconds to would have
> to divide it by "10E6" then wouldn't that be microseconds and not
> nanoseconds?

Sorry, CPU time first gets divided by 1000 and then by 10E6. Wasn't
clear from what I've written. The most important part from my post
was, that the time is *always* in seconds.

> if i have a value of 0.04 for the user cpu time what is it in seconds?
> Could you please give me a concrete definition for the user CPU time? what
> are we measuring here exactly? and what is the difference with cpu user time
> and measure time?

UserCPU time is what the thread used in actual CPU cycles, hence most
accurate. Elapsed time is quite inaccurate, since it's only the
difference between the system's time (other processes/threads could
steal valuable processing time!).

> I would like to compare different feature selection methods by their
> time consumption with respect to the time consumption of a classifier
> without feature selection. Could you please suggest on which values i should
> use? user CPU or measure time?

UserCPU.

Cheers, Peter


_______________________________________________
Wekalist mailing list
[hidden email]
https://list.scms.waikato.ac.nz/mailman/listinfo/wekalist
Reply | Threaded
Open this post in threaded view
|

Re: Re: User CPU time

Peter Reutemann
> "that the time is *always* in seconds"?
> meaning that if I get a value of 0.4 in the user CPU this means that the
> time used for the CPU to process it is 0.4*10E-9?

"0.4" as user CPU time means 0.4 seconds have passed. The unit is
seconds, not milliseconds or nanoseconds or anything else.

Cheers, Peter
--
Peter Reutemann, Dept. of Computer Science, University of Waikato, NZ
http://www.cs.waikato.ac.nz/~fracpete/ Ph. +64 (7) 858-5174

_______________________________________________
Wekalist mailing list
[hidden email]
https://list.scms.waikato.ac.nz/mailman/listinfo/wekalist
Reply | Threaded
Open this post in threaded view
|

Re: User CPU time

Zakir Afridi
Dear Sir, if the time has taken is 0 sec then how it is justifiable. how 0
sec is convertible to milliseconds?



--
Sent from: http://weka.8497.n7.nabble.com/
_______________________________________________
Wekalist mailing list
Send posts to: [hidden email]
To subscribe, unsubscribe, etc., visit https://list.waikato.ac.nz/mailman/listinfo/wekalist
List etiquette: http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html
Reply | Threaded
Open this post in threaded view
|

Re: User CPU time

Eibe Frank-2
Administrator

The output is rounded to two decimal places. The Experimenter will give you more precise numeric values if you need them.

 

Cheers,

Eibe

 

From: [hidden email]
Sent: Saturday, 9 February 2019 3:05 PM
To: [hidden email]
Subject: Re: [Wekalist] User CPU time

 

Dear Sir, if the time has taken is 0 sec then how it is justifiable. how 0

sec is convertible to milliseconds?

 

 

 

--

Sent from: http://weka.8497.n7.nabble.com/

_______________________________________________

Wekalist mailing list

Send posts to: [hidden email]

To subscribe, unsubscribe, etc., visit https://list.waikato.ac.nz/mailman/listinfo/wekalist

List etiquette: http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html

 


_______________________________________________
Wekalist mailing list
Send posts to: [hidden email]
To subscribe, unsubscribe, etc., visit https://list.waikato.ac.nz/mailman/listinfo/wekalist
List etiquette: http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html