[Rcpp-devel] Segfault error during simulation in Rcpp

[Matteo Fasiolo]

Sorry yesterday evening by mistake I sent a message only to Dirk:

Thanks, you are right I'll try cxxfunction and/or I'll see whether I can
move avoid doing so many calls to C++.

For the sake of minimality: this code is enough to have a segfault or to
make R stall
(at least on my computer):

library(Rcpp)
myFun <- cppFunction('NumericMatrix myFun(NumericMatrix input){ return
input; }')

n <- 10
x <- 1:n^2
N <- 1e6
b <- 0
A <- matrix(x, n, n)
for (j in 1:N) {
    res <- myFun(A)
    a <- res[1,1]
    b <- b + a
}

cat(sprintf("Done, b is %d\n", b))

Thanks a lot to everybody for the help!


On Fri, May 17, 2013 at 4:17 AM, Ivan Popivanov <ivan.popivanov at gmail.com>wrote:

> Ignore my previous mail ...
>
>
> On Thu, May 16, 2013 at 11:12 PM, Ivan Popivanov <ivan.popivanov at gmail.com
> > wrote:
>
>> How is this code supposed to work? If n=1e6, then the matrix has 1e12
>> elements, right? That's in the terabyte range - the memory manager is going
>> to blow up or overflow and the results would be unpredictable. Or am I
>> missing something?
>>
>> Regards,
>> Ivan
>>
>>
>> On Thu, May 16, 2013 at 7:14 PM, Jonathan Olmsted <jpolmsted at gmail.com>wrote:
>>
>>> Matteo,
>>>
>>>
>>>
>>>> The other obvious of course is that you are not forced to control a
>>>> loop over
>>>> 10^6 elements from R either:  pass N=10^6 down to C++ code, and run
>>>> your N
>>>> loops there.  You will also get a considerable speed boost.
>>>>
>>>>
>>> ^ I could have been more explicit. This is what I meant. You might may
>>> or may not be able to do this in the sourceCpp() framework, though. How you
>>> do this is much more obvious if you go the route of making an Rcpp package.
>>> For as much as the convenience functions are a blessing, I've always found
>>> it hard to think about how I'd do more complex things with them. It could
>>> be possible, I just don't see it. I find the package approach to be a nice
>>> balance between abstracting away from some details but keeping some in the
>>> forefront of your mind. If you email me off the list and I can send you a
>>> very small package that using Rcpp. Multiple C++ functions are created,
>>> only some are exposed to R.
>>>
>>> It seems like you'd probably do something similar.
>>>
>>> -Jonathan
>>>
>>>
>>>
>>>
>>>
>>>>  | On Thu, May 16, 2013 at 8:11 PM, Dirk Eddelbuettel <edd at debian.org>
>>>> wrote:
>>>> |
>>>> |
>>>> |     On 16 May 2013 at 14:49, Jonathan Olmsted wrote:
>>>> |     | Several things.
>>>> |     |
>>>> |     | Xiao, Dirk's code gives me a segfault immediately and reliably.
>>>> |     |
>>>> |     | All, when I do this whole song and dance using the "old"
>>>> Rcpp/inline/
>>>> |     | cxxfunction approach, I don't have any issues. One obvious
>>>> difference you
>>>> |     can
>>>> |     | see (like was mentioned) in the generated code (visibile using
>>>> verbose=
>>>> |     TRUE) is
>>>> |     | the declaration of an RNGScope object.
>>>> |     |
>>>> |     | But, if the memory issue crops up when you call a C++
>>>> function (syncing
>>>> |     with
>>>> |     | R's RNG state) 1e6 times AND you are already writing C++ maybe
>>>> this is
>>>> |     an
>>>> |     | opportunity to just put one more layer of the code into C++ and
>>>> create
>>>> |     only one
>>>> |     | such RNGScope object?
>>>> |
>>>> |     Beautiful. So we get to blame R Core after all?  ;-)
>>>> |
>>>> |     Dirk
>>>> |
>>>> |     --
>>>> |     Dirk Eddelbuettel | edd at debian.org | http://dirk.eddelbuettel.com
>>>> |     _______________________________________________
>>>> |     Rcpp-devel mailing list
>>>> |     Rcpp-devel at lists.r-forge.r-project.org
>>>> |
>>>> https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/rcpp-devel
>>>> |
>>>> |
>>>>
>>>> --
>>>> Dirk Eddelbuettel | edd at debian.org | http://dirk.eddelbuettel.com
>>>>
>>>
>>>
>>> _______________________________________________
>>> Rcpp-devel mailing list
>>> Rcpp-devel at lists.r-forge.r-project.org
>>> https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/rcpp-devel
>>>
>>
>>
>
> _______________________________________________
> Rcpp-devel mailing list
> Rcpp-devel at lists.r-forge.r-project.org
> https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/rcpp-devel
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.r-forge.r-project.org/pipermail/rcpp-devel/attachments/20130517/cf32ae23/attachment.html>