[R] on execution time of a function...

[Jeff Newmiller]

it is not just the list that is OS-agnostic... R itself is quite OS-agnostic. However, depending on where you get your packages from they may have something unusual going on in their compiled C or Fortran code. It is just that for a computationally-oriented application like R code that messed with hardware interrupts seem out of place.

I suppose that a bug in a CUDA-related package could screw up interrupts... but the whole point of having an OS as hardware drivers/APIs is to insulate user-level programs like R from those issues, and at best you would post the details of your code and packages and someone might pick up on an odd package and point you toward a more appropriate forum for debugging it.

On June 12, 2018 11:39:40 PM HST, akshay kulkarni <akshay_e4 using hotmail.com> wrote:
>I do felt a little unsettled by your exhortions on the incongruity of
>posting this question on an OS-agnostic mailing list...I thought that
>there might be some issues on how R communicates with the OS...and also
>that some R packages might rectify the issue(in my experience, I have
>had a R package for every R issue that I had!)
>
>Anyway, I ran the  same function again and it is executing within
>limits. Also, I would be using AWS EC2 servers to run my R functions(I
>am a day trader in india and input some 250 stocks to R functions
>daily), and I don't think that the issue would persist on Intel Xeon
>processors and dedicated VMs....
>
>Anyway,thanks for all your concerns tolerating my query on an
>OS-agnostic mailing list!
>
>AKSHAY M KULKARNI
>________________________________
>From: Loris Bennett <loris.bennett using fu-berlin.de>
>Sent: Wednesday, June 13, 2018 1:37 PM
>To: Jeff Newmiller
>Cc: r-help using r-project.org; akshay kulkarni
>Subject: Re: [R] on execution time of a function...
>
>Hi Akshay,
>
>In addition to all the things Jeff rightly points out, contention for
>IO
>resources can be an issue.  So if another process was hogging the
>bandwidth while your program was attempting to read or write to disk,
>that could also have slowed things down.
>
>HTH
>
>Loris
>
>Jeff Newmiller <jdnewmil using dcn.davis.ca.us> writes:
>
>> Wow, you can find almost any explanation on the Internets. That
>doesn't mean you
>> should believe all of them. R does not do anything likely to tweak
>> interrupts... if that is your problem then you need to be on an
>> operating-system/computer-model-specific forum rather than this
>OS-agnostic
>> mailing list.
>>
>> It is far more likely that your overall memory usage conditions have
>changed
>> since the last time you ran it... or that you didn't actually record
>all of the
>> things you did last time in your script. (Newbie R users often do
>things at the
>> console without putting them in their scripts.)
>>
>> I suggest that you run your script one statement at a time and see
>where your
>> problem is. You might also want to make sure that other programs are
>not using
>> up a lot of your memory (which could involve some OS-specific tools
>or just
>> shutting down some other programs.)
>>
>> On June 12, 2018 7:36:40 PM HST, akshay kulkarni
><akshay_e4 using hotmail.com> wrote:
>>>I ran  a function in R three days ago and the execution time was
>about
>>>4 minutes. I ran the same function yesterday and the execution time
>was
>>>more than 6:50 minutes(I aborted the function after that time).
>>>
>>>I read in the Internet that this is possible. I also came to know
>that
>>>software or hardware interrupts are the main reasons.
>>>
>>>How do you know whether the delay was caused by interrupts? Which
>>>hardware or software triggered the interrupts? In general, how to
>know
>>>the exact cause of the delay in execution in R? Are there any
>packages
>>>for these analyses?
>>>
>>>Very many thanks for your time and effort....
>>>
>>>yours sincerely,
>>>AKSHAY M KULKARNI
>>>
>>>      [[alternative HTML version deleted]]
>>>
>>>______________________________________________
>>>R-help using r-project.org mailing list -- To UNSUBSCRIBE and more, see
>>>https://stat.ethz.ch/mailman/listinfo/r-help
>>>PLEASE do read the posting guide
>>>http://www.R-project.org/posting-guide.html
>>>and provide commented, minimal, self-contained, reproducible code.
>--
>Dr. Loris Bennett (Mr.)
>ZEDAT, Freie Universität Berlin         Email
>loris.bennett using fu-berlin.de

-- 
Sent from my phone. Please excuse my brevity.