<div class="message-text"><span>

                Hi,

</span><br><span>I am trying to convert a simple CUDA program to LLVM IR using clang 3.0. The program is as follows,

</span><br><span>#include<stdio.h></span><br><span>#nclude<clang/test/SemaCUDA/cuda.h></span><br><br><span>__global__ void kernfunc(int *a)

</span><br><span> {

</span><br><span>*a=threadIdx.x+blockIdx.x*blockDim.x;

</span><br><span>}

</span><br><br><span>int main()

</span><br><span>{

</span><br><span>int *h_a,*d_a,n;

</span><br><span>                                        

</span><br><span>n=sizeof(int);

</span><br><span>h_a=(int*)malloc(n);

</span><br><span>*h_a=5;

</span><br><br><span>cudaMalloc((void*)&d_a,n);

</span><br><span>cudaMemcpy(d_a,h_a,n,cudaMemcpyHostToDevice);

</span><br><span>kernelfunc<<<1,1>>>(d_a);

</span><br><span>cudaMemcpy(h_a,d_a,n,cudaMemcpyDeviceToHost);

</span><br><br><span>printf("%d",*h_a);

</span><br><span>return 0;

</span><br><span>}

What additional header files should be included? What part of the code is currently not supported by clang 3.0?

</span><br><span>Thank you:)

</span><br><span>

        </span></div>

        <div class="ad shaded-bg-color rounded" style="display:block;margin:0 0 1em;padding:.5em">

                                        </div>