[R] Which system.time() component to use?

Ravi Varadhan rvaradhan at jhmi.edu
Sun Feb 28 20:29:04 CET 2010


Thank you Jim and Gabor.  It seems that for my purpose of comparing the computational effort (not I/O) the optimal thing is to use the sum of user and system CPU times to get the total CPU.  It also seems like when there is not much I/O involved, which is the case for my algorithms, the total CPU is not that different from the total elapsed time.

Best regards,
Ravi.
____________________________________________________________________

Ravi Varadhan, Ph.D.
Assistant Professor,
Division of Geriatric Medicine and Gerontology
School of Medicine
Johns Hopkins University

Ph. (410) 502-2619
email: rvaradhan at jhmi.edu


----- Original Message -----
From: jim holtman <jholtman at gmail.com>
Date: Sunday, February 28, 2010 0:03 am
Subject: Re: [R] Which system.time() component to use?
To: Ravi Varadhan <rvaradhan at jhmi.edu>
Cc: Gabor Grothendieck <ggrothendieck at gmail.com>, r-help at r-project.org


> A lot depends on what you are trying to measure.
> 
> You should add the system and user CPU times to get a better idea of
> the CPU utilization.  For some classes of problems it might be good to
> separate them if you were doing a lot of I/O or other system calls
> that might be using time, but for 99% of the cases adding them is the
> way to go.
> 
> You also want to look at elapsed times.  If the script is CPU bound
> the elapsed and total CPU times should be close.  In the case that
> Gabor gave of sleeping for 60 seconds, no CPU time was used, but it
> was 60 seconds of elapsed time.  If there is a big difference, it
> might be due to a lot of I/O or possible paging if you did not have
> enough memory.
> 
> On Sat, Feb 27, 2010 at 11:44 PM, Ravi Varadhan <rvaradhan at jhmi.edu> wrote:
> >
> > Thanks, Gabor.  Your reply is helpful, but it still doesn't answer 
> whether I should use the sum of the first two components of 
> system.time (user + system CPU) or only the first one (user CPU).
> >
> > Ravi.
> > ____________________________________________________________________
> >
> > Ravi Varadhan, Ph.D.
> > Assistant Professor,
> > Division of Geriatric Medicine and Gerontology
> > School of Medicine
> > Johns Hopkins University
> >
> > Ph. (410) 502-2619
> > email: rvaradhan at jhmi.edu
> >
> >
> > ----- Original Message -----
> > From: Gabor Grothendieck <ggrothendieck at gmail.com>
> > Date: Saturday, February 27, 2010 9:47 pm
> > Subject: Re: [R] Which system.time() component to use?
> > To: Ravi Varadhan <rvaradhan at jhmi.edu>
> > Cc: r-help at r-project.org
> >
> >
> >> Try this:
> >>
> >> > system.time(Sys.sleep(60))
> >>    user  system elapsed
> >>    0.00    0.00   60.05
> >> > pt <- proc.time(); Sys.sleep(60); proc.time() - pt
> >>    user  system elapsed
> >>    0.00    0.00   60.01
> >>
> >> On Sat, Feb 27, 2010 at 9:33 PM, Ravi Varadhan <rvaradhan at jhmi.edu> 
> wrote:
> >> >
> >> > Hi,
> >> >
> >> > The `system.time(expr)' command provide 3 different times for
> >> evaluating the expression `expr'; the first two are user and system
> >> CPUs and the third one is total elapsed time.  Suppose I want to
> >> compare two different computational procedures for performing the same
> >> task, which component of `system.time' is most meaningful in the sense
> >> that it most accurately reflects the computational effort of the
> >> algorithm, and does not depend upon the idiosyncrasies of the
> >> operating system.
> >> >
> >> > I have always been using the first component of `system.time', which
> >> is the user CPU.  Should I use the sum of user and system CPU or is
> >> the total elapsed time a better measure?  I would appreciate UseR's
> >> feedback on this.
> >> >
> >> > Thanks very much.
> >> >
> >> > Best,
> >> > Ravi.
> >> > ____________________________________________________________________
> >> >
> >> > Ravi Varadhan, Ph.D.
> >> > Assistant Professor,
> >> > Division of Geriatric Medicine and Gerontology
> >> > School of Medicine
> >> > Johns Hopkins University
> >> >
> >> > Ph. (410) 502-2619
> >> > email: rvaradhan at jhmi.edu
> >> >
> >> > ______________________________________________
> >> > R-help at r-project.org mailing list
> >> >
> >> > PLEASE do read the posting guide
> >> > and provide commented, minimal, self-contained, reproducible code.
> >> >
> >
> > ______________________________________________
> > R-help at r-project.org mailing list
> > 
> > PLEASE do read the posting guide 
> > and provide commented, minimal, self-contained, reproducible code.
> >
> 
> 
> 
> -- 
> Jim Holtman
> Cincinnati, OH
> +1 513 646 9390
> 
> What is the problem that you are trying to solve?



More information about the R-help mailing list