Databases
StackOverflow2013
Postsdbo
PORcpp and CULA: segmentation fault
 

Note that there are some explanatory texts on larger screens.

plurals
  1. PORcpp and CULA: segmentation fault
    primarykey
    data
    text
    <p>I extracted the relevant bits from the <em>gputools R</em>-package to run a QR decomposition on my GPU using <em>Rcpp</em> by dynamically loading a shared library that links to <em>culatools</em>. Everything runs smoothly in the terminal and <em>R.app</em> on my Mac. The results agree with <em>R</em>'s <em>qr()</em> function, but the problem is that a segmentation fault occurs on exiting <em>R.app</em> (the error does not occur when using the terminal):</p> <pre><code>*** caught segfault *** address 0x10911b050, cause 'memory not mapped' </code></pre> <p>I think I narrowed down the problem to the pointers 'a' and 'tau' in the .c file that links to <em>culatools</em>:</p> <pre><code>#include&lt;cula.h&gt; void gpuQR(const int *m, const int *n, float *a, const int *lda, float *tau) { culaInitialize(); culaSgeqrf(m[0], n[0], a, lda[0], tau); culaShutdown(); } </code></pre> <p>I compiled the .c file on my Mac using:</p> <pre><code>/usr/local/cuda/bin/nvcc -gencode arch=compute_10,code=sm_10 -gencode arch=compute_11,code=sm_11 -gencode arch=compute_12,code=sm_12 -gencode arch=compute_13,code=sm_13 -gencode arch=compute_20,code=sm_20 -c -I. -I/usr/local/cula/include -m64 -Xcompiler -fPIC gpuQR.c -o gpuQR.o /usr/local/cuda/bin/nvcc -gencode arch=compute_10,code=sm_10 -gencode arch=compute_11,code=sm_11 -gencode arch=compute_12,code=sm_12 -gencode arch=compute_13,code=sm_13 -gencode arch=compute_20,code=sm_20 -shared -m64 -Xlinker -rpath,/usr/local/cula/lib64 -L/usr/local/cula/lib64 -lcula_core -lcula_lapack -lcublas -o gpuQR.so gpuQR.o </code></pre> <p>I wrote a .cpp file that uses <em>Rcpp</em> and dynamically loads the shared library gpuQR.so:</p> <pre><code>#include &lt;Rcpp.h&gt; #include &lt;dlfcn.h&gt; using namespace Rcpp; using namespace std; typedef void (*func)(int*, int*, float*, int*, float*); RcppExport SEXP gpuQR_Rcpp(SEXP x_, SEXP n_rows_, SEXP n_cols_) { vector&lt;float&gt; x = as&lt;vector&lt;float&gt; &gt;(x_); int n_rows = as&lt;int&gt;(n_rows_); int n_cols = as&lt;int&gt;(n_cols_); vector&lt;float&gt; scale(n_cols); void* lib_handle = dlopen("path/gpuQR.so", RTLD_LAZY); if (!lib_handle) { Rcout &lt;&lt; dlerror() &lt;&lt; endl; } else { func gpuQR = (func) dlsym(lib_handle, "gpuQR"); gpuQR(&amp;n_rows, &amp;n_cols, &amp;(x[0]), &amp;n_rows, &amp;(scale[0])); } dlclose(lib_handle); for(int ii = 1; ii &lt; n_rows; ii++) { for(int jj = 0; jj &lt; n_cols; jj++) { if(ii &gt; jj) { y[ii + jj * n_rows] *= scale[jj]; } } } return wrap(x); } </code></pre> <p>I compiled the .cpp file in <em>R</em> using:</p> <pre><code>library(Rcpp) PKG_LIBS &lt;- sprintf('%s $(LAPACK_LIBS) $(BLAS_LIBS) $(FLIBS)', Rcpp:::RcppLdFlags()) PKG_CPPFLAGS &lt;- sprintf('%s', Rcpp:::RcppCxxFlags()) Sys.setenv(PKG_LIBS = PKG_LIBS , PKG_CPPFLAGS = PKG_CPPFLAGS) R &lt;- file.path(R.home(component = 'bin'), 'R') file &lt;- 'path/gpuQR_Rcpp.cpp' cmd &lt;- sprintf('%s CMD SHLIB %s', R, paste(file, collapse = ' ')) system(cmd) </code></pre> <p>and ran an example:</p> <pre><code>dyn.load('path/gpuQR_Rcpp.so') set.seed(100) A &lt;- matrix(rnorm(9), 3, 3) n_row &lt;- nrow(A) n_col &lt;- ncol(A) res &lt;- .Call('gpuQR_Rcpp', c(A), n_row, n_col) matrix(res, n_row, n_col) [,1] [,2] [,3] [1,] 0.5250958 -0.8666927 0.8594266 [2,] -0.2504899 -0.3878644 -0.1277837 [3,] 0.1502908 0.4742033 -0.8804248 qr(A)$qr [,1] [,2] [,3] [1,] 0.5250957 -0.8666925 0.8594266 [2,] -0.2504899 -0.3878643 -0.1277838 [3,] 0.1502909 0.4742033 -0.8804247 </code></pre> <p>Does anybody have an idea how to fix the segmentation fault?</p>
    singulars
    1. This table or related slice is empty.
    1. This table or related slice is empty.
    1. PTQuestion
    1. USchris
    1. USchris
    plurals
    1. This table or related slice is empty.
    1. PL
      singulars
      1. LTLinked
    1. This table or related slice is empty.
    1. PO
      singulars
      1. PTAnswer
    2. PO
      singulars
      1. PTAnswer
    1. VO
      singulars
      1. PORcpp and CULA: segmentation fault
      1. This table or related slice is empty.
      1. VTUpMod
    1. COWhy are you using nvcc to compile your code? If it is plain C code, you can (and probably should) use the host C compiler. nvcc uses a C++ host compilation route by default. That may not be what you want.
      singulars
      1. PORcpp and CULA: segmentation fault
      1. UStalonmies
    2. COI just used what the *gputools R* package uses. Actually, *Nvidia* [documentation](http://docs.nvidia.com/cuda/cuda-compiler-driver-nvcc) says the following about this: "... nvcc mimics the behavior of the GNU compiler gcc: it accepts a range of conventional compiler options, such as for defining macros and include/library paths, and for steering the compilation process. All non-CUDA compilation steps are forwarded to a general purpose C compiler that is supported by nvcc, ..."
      singulars
      1. PORcpp and CULA: segmentation fault
      1. USchris
    3. COYes, I fully understand that. But be aware that the "general purpose compiler" is a C++ compiler, not a C compiler. So you C code gets compiled with a C++ compiler. Most of the time it doesn't matter but sometimes it does. This might well be one of those times....
      singulars
      1. PORcpp and CULA: segmentation fault
      1. UStalonmies
 

Querying!

 
Guidance
A row detail

Detail views are divided into sections. All the information in the data section comes from columns in the selected row. The other sections display data from other, related rows.

Related data can be related in a to-one or a to-many fashion. Captions of data related in a to-many fashion link to a list view showing a filtered view of the table.

Try moving around until you find a non-empty to-many entry and click on the label to get to one. You can move back to the root by clicking on the database name in the header.

SQuiL has stopped working due to an internal error.

If you are curious you may find further information in the browser console, which is accessible through the devtools (F12).

Reload