Take a small subset of your program that would run through the
critical sections and use ?Rprof to see where some of the hot spot
are. How do you know it is not using the CPU? Are you using perfmon
to look what is being used? Are you paging? If you are not paging,
and not doing a lot of I/O, then you should tie up one CPU 100% if you
are CPU bound.
You probably need to put some output in your program to mark its
progress. At a minimum, do the following:
cat('I am here', proc.time(), '\n')
By hcaning the initial string, you can see where you are and this is
also reporting the user CPU, system CPU and elapsed time. This should
be a good indication of where time is being spent. So there are a
number of things you can do to "instrument" your code. If I had a
program that was running for hours, I would definitely have something
that tell me where I am at and how much time is being taken. If you
have some large loop, then you could put out this information every
'n'th time through. The tag on the message would indicate this.
There is also the progress bar that I use a lot to see if I amd
"making" progress.
After you have instrumented your code and have use Rprof, you might
have some data that people would help you with.
If you are using dataframe a lot, remember that indexing into them can
be costly. Converting them to matrices, if appropriate, can give a
big speed. Rprof will show you this.
On Fri, May 27, 2011 at 2:00 PM, Debs Majumdar <debs_st...@yahoo.com> wrote:
> Hello,
>
> Are there some basic things one can do to speed up a R code? I am new to R
> and currently going through the following situation.
>
> I have run a R code on two different machines. I have R 2.12 installed on
> both.
>
> Desktop 1 is slightly older and has a dual core processor with 4gigs of
> RAM. Desktop 2 is newer one and has a xeon processor W3505 with 12gigs of
> RAM. Both run on Windows 7.
>
> I don't really see any significant speed up in the newer computer (Desktop
> 2). In the older one the program took around 5hrs 15 mins and in the newer
> one it took almost 4hrs 30mins.
>
> In the newer dekstop, R gives me the following:
>
>> memory.limit()
> [1] 1024
>> memory.size()
> [1] 20.03
>
> Is something hampering me here? Do I need to increase the limit and size?
> Can this change be made permanent? Or am I looking at the wrong place?
>
> I have never seen my R programs using much CPU or RAM when it runs? If this
> is not something inherent to R, then I guess I need to write more effiecient
> codes.
>
> Suggestions/solutions are welcome.
>
> Thanks,
>
> -Debs
>
>
> ______________________________________________
> R-help@r-project.org mailing list
> 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.
>
--
Jim Holtman
Data Munger Guru
What is the problem that you are trying to solve?
______________________________________________
R-help@r-project.org mailing list
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.
