如何正确地从传递的数组构造CUSP矩阵

我正在尝试将CUSP集成到现有的Fortran代码中。现在我只是想从Fortran中获得推力/CUSP的基本设置，并使用它们来构建CUSP矩阵(现在的coo格式)。由于这个线程，我已经能够获得一个包装器，如C例程编译成一个库，并将其与Fortran代码链接:unresolated -references-using-ifort-with-nvcc-and-cusp

和我可以验证Fortran是正确馈送数组指针感谢从前面的线程的帮助:从传入的Fortran数组生成CUSP coo_matrix

不幸的是，我仍然不能得到CUSP使用这些来生成和打印矩阵。代码和输出如下所示:

$ ./fort_cusp_test
 testing 1 2 3
n: 3, nnz: 9
     i,  row_i,  col_j,        val_v
     0,      1,      1,   1.0000e+00
     1,      1,      2,   2.0000e+00
     2,      1,      3,   3.0000e+00
     3,      2,      1,   4.0000e+00
     4,      2,      2,   5.0000e+00
     5,      2,      3,   6.0000e+00
     6,      3,      1,   7.0000e+00
     7,      3,      2,   8.0000e+00
     8,      3,      3,   9.0000e+00
initialized row_i into thrust
initialized col_j into thrust
initialized val_v into thrust
defined CUSP integer array view for row_i and col_j
defined CUSP float array view for val_v
loaded row_i into a CUSP integer array view
loaded col_j into a CUSP integer array view
loaded val_v into a CUSP float array view
defined CUSP coo_matrix view
Built matrix A from CUSP device views
sparse matrix <3, 3> with 9 entries
libc++abi.dylib: terminating with uncaught exception of type thrust::system::system_error: invalid argument
Program received signal SIGABRT: Process abort signal.
Backtrace for this error:
#0  0x10d0fdff6
#1  0x10d0fd593
#2  0x7fff8593ff19
Abort trap: 6

fort_cusp_test.f90

program fort_cuda_test
   implicit none
 ! interface
 !    subroutine test_coo_mat_print_(row_i,col_j,val_v,n,nnz) bind(C)
 !       use, intrinsic :: ISO_C_BINDING, ONLY: C_INT,C_FLOAT
 !       implicit none
 !       integer(C_INT),value :: n, nnz
 !       integer(C_INT) :: row_i(:), col_j(:)
 !       real(C_FLOAT) :: val_v(:)
 !    end subroutine test_coo_mat_print_
 ! end interface
   integer*4   n
   integer*4   nnz
   integer*4, target :: rowI(9),colJ(9)
   real*4, target :: valV(9)
   integer*4, pointer ::   row_i(:)
   integer*4, pointer ::   col_j(:)
   real*4, pointer ::   val_v(:)
   n     =  3
   nnz   =  9
   rowI =  (/ 1, 1, 1, 2, 2, 2, 3, 3, 3/)
   colJ =  (/ 1, 2, 3, 1, 2, 3, 1, 2, 3/)
   valV =  (/ 1, 2, 3, 4, 5, 6, 7, 8, 9/)
   row_i => rowI
   col_j => colJ
   val_v => valV
   write(*,*) "testing 1 2 3"
   call test_coo_mat_print (rowI,colJ,valV,n,nnz)
end program fort_cuda_test

cusp_runner.cu

#include <stdio.h>
#include <cusp/coo_matrix.h>
#include <iostream>
// #include <cusp/krylov/cg.h>
#include <cusp/print.h>
#if defined(__cplusplus)
extern "C" {
#endif
void test_coo_mat_print_(int * row_i, int * col_j, float * val_v, int * N, int * NNZ ) {
   int n, nnz;
   n = *N;
   nnz = *NNZ;
   printf("n: %d, nnz: %dn",n,nnz);
   printf("%6s, %6s, %6s, %12s n","i","row_i","col_j","val_v");
   for(int i=0;i<n;i++) {
      printf("%6d, %6d, %6d, %12.4en",i,row_i[i],col_j[i],val_v[i]);
   }
   //if ( false ) {
   //wrap raw input pointers with thrust::device_ptr
   thrust::device_ptr<int> wrapped_device_I(row_i);
   printf("initialized row_i into thrustn");
   thrust::device_ptr<int> wrapped_device_J(col_j);
   printf("initialized col_j into thrustn");
   thrust::device_ptr<float> wrapped_device_V(val_v);
   printf("initialized val_v into thrustn");
   //use array1d_view to wrap individual arrays
   typedef typename cusp::array1d_view< thrust::device_ptr<int> > DeviceIndexArrayView;
   printf("defined CUSP integer array view for row_i and col_jn");
   typedef typename cusp::array1d_view< thrust::device_ptr<float> > DeviceValueArrayView;
   printf("defined CUSP float array view for val_vn");
   DeviceIndexArrayView row_indices(wrapped_device_I, wrapped_device_I + nnz);
   printf("loaded row_i into a CUSP integer array viewn");
   DeviceIndexArrayView column_indices(wrapped_device_J, wrapped_device_J + nnz);
   printf("loaded col_j into a CUSP integer array viewn");
   DeviceValueArrayView values(wrapped_device_V, wrapped_device_V + nnz);
   printf("loaded val_v into a CUSP float array viewn");
   //combine array1d_views into coo_matrix_view
   typedef cusp::coo_matrix_view<DeviceIndexArrayView,DeviceIndexArrayView,DeviceValueArrayView> DeviceView;
   printf("defined CUSP coo_matrix viewn");
   //construct coo_matrix_view from array1d_views
   DeviceView A(n,n,nnz,row_indices,column_indices,values);
   printf("Built matrix A from CUSP device viewsn");
   cusp::print(A);
   printf("Printed matrix An");
 //}
}
#if defined(__cplusplus)
}
#endif

Makefile

Test:
   nvcc -Xcompiler="-fPIC" -shared cusp_runner.cu -o cusp_runner.so -I/Developer/NVIDIA/CUDA-6.5/include/cusp
   gfortran -c fort_cusp_test.f90
   gfortran fort_cusp_test.o cusp_runner.so -L/Developer/NVIDIA/CUDA-6.5/lib -lcudart -o fort_cusp_test
clean:
   rm *.o *.so fort_cusp_test

cusp_runner.cu:

#include <stdio.h>
#include <cusp/coo_matrix.h>
#include <iostream>
// #include <cusp/krylov/cg.h>
#include <cusp/print.h>
#if defined(__cplusplus)
extern "C" {
#endif
void test_coo_mat_print_(int * row_i, int * col_j, float * val_v, int * N, int * NNZ ) {
   int n, nnz;
   n = *N;
   nnz = *NNZ;
   printf("n: %d, nnz: %dn",n,nnz);
   printf("printing input (row_i, col_j, val_v)n");
   printf("%6s, %6s, %6s, %12s n","i","row_i","col_j","val_v");
   for(int i=0;i<nnz;i++) {
      printf("%6d, %6d, %6d, %12.4en",i,row_i[i],col_j[i],val_v[i]);
   }
   printf("initializing thrust device vectorsn");
   thrust::device_vector<int> device_I(row_i,row_i+nnz);
   printf("device_I initializedn");
   thrust::device_vector<int> device_J(col_j,col_j+nnz);
   printf("device_J initializedn");
   thrust::device_vector<float> device_V(val_v,val_v+nnz);
   printf("device_V initializedn");
   cusp::coo_matrix<int, float, cusp::device_memory> A(n,n,nnz);
   printf("initialized empty CUSP coo_matrix on devicen");
   A.row_indices = device_I;
   printf("loaded device_I into A.row_indicesn");
   A.column_indices = device_J;
   printf("loaded device_J into A.column_indicesn");
   A.values = device_V;
   printf("loaded device_V into A.valuesn");
   cusp::print(A);
   printf("Printed matrix An");
 //}
}
#if defined(__cplusplus)
}
#endif