Dear ViennaCL users and developers,
I was playing with the tutorial problem custom-cuda.cu. I found a very
strange result.
|#include <iostream>
#include <string>
#ifndef VIENNACL_WITH_CUDA
#define VIENNACL_WITH_CUDA
#endif
// ViennaCL headers
#include "viennacl/vector.hpp"
#include "viennacl/matrix.hpp"
#include "viennacl/linalg/matrix_operations.hpp"
#include "viennacl/linalg/norm_2.hpp"
#include "viennacl/linalg/prod.hpp
// Some helper functions for this tutorial:
#include "Random.hpp"
template<typename NumericT>
__global__ void elementwise_multiplication(const NumericT * vec1,
const
NumericT * vec2,
NumericT * result,
unsigned int size)
{
for (unsigned int i = blockDim.x * blockIdx.x + threadIdx.x;
i < size;
i += gridDim.x * blockDim.x)
result[i] = vec1[i] * vec2[i];
}
int main()
{
typedef double NumericType;
std::size_t N = 5;
NumericType * device_vec1;
NumericType * device_vec2;
NumericType * device_result;
cudaMalloc(&device_vec1, N * sizeof(NumericType));
cudaMalloc(&device_vec2, N * sizeof(NumericType));
cudaMalloc(&device_result, N * sizeof(NumericType));
std::vector<NumericType> temp(N);
for (std::size_t i=0; i<temp.size(); ++i)
temp[i] = NumericType(i);
cudaMemcpy(device_vec1, &(temp[0]), temp.size() * sizeof(NumericType),
cudaMemcpyHostToDevice );
for (std::size_t i=0; i<temp.size(); ++i)
temp[i] = NumericType(2*i);
cudaMemcpy(device_vec2, &(temp[0]), temp.size() * sizeof(NumericType),
cudaMemcpyHostToDevice );
viennacl::vector<NumericType> vcl_vec1(device_vec1, viennacl::CUDA_MEMORY,
N);
viennacl::vector<NumericType> vcl_vec2(device_vec2, viennacl::CUDA_MEMORY,
N);
viennacl::vector<NumericType> vcl_result(device_result,
viennacl::CUDA_MEMORY, N);
std::cout << "Using existing kernel within ViennaCL:" << std::endl;
elementwise_multiplication<<<128,
128>>>(vcl_vec1.handle().cuda_handle().get(),
vcl_vec2.handle().cuda_handle().get(),
vcl_result.handle().cuda_handle().get(),
N);
std::cout << "vec1 : " << vcl_vec1 << std::endl;
std::cout << "vec2 : " << vcl_vec2 << std::endl;
std::cout << "result: " << vcl_result << std::endl;
std::cout << "!!!! TUTORIAL COMPLETED SUCCESSFULLY !!!!" << std::endl;
return EXIT_SUCCESS;|
}
And after compiling the result I am getting the following result
| Using existing kernel within ViennaCL:
vec1 : [5](0,1,2,3,4)
vec2 : [5](0,2,4,6,8)
result: [5](60,160,260,360,460)
!!!! TUTORIAL COMPLETED SUCCESSFULLY !!!!|
|
|Should not kernel produce the following result?
| result: [5](0,2,8,18,32)|
Am I missing something ? Thank you in advance.
With Best Regards,
Arijit Hazra
------------------------------------------------------------------------------
One dashboard for servers and applications across Physical-Virtual-Cloud
Widest out-of-the-box monitoring support with 50+ applications
Performance metrics, stats and reports that give you Actionable Insights
Deep dive visibility with transaction tracing using APM Insight.
http://ad.doubleclick.net/ddm/clk/290420510;117567292;y
_______________________________________________
ViennaCL-support mailing list
ViennaCL-support@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/viennacl-support
