sysrap/tests/tcomplexTest.cu

0001 /**
0002 name=tcomplexTest ; nvcc $name.cu -I.. -I/usr/local/cuda -o /tmp/$name && /tmp/$name
0003
0004 **/
0005
0006 #include <iostream>
0007 #include <iomanip>
0008
0009 #include "scuda.h"
0010 #include "tcomplex.h"
0011
0012
0013 /*
0014  we test the follow calculation:
0015
0016    a = 0 + 4i;
0017    sqrt of a = 2^0.5 + i* 2^0.5
0018
0019 */
0020
0021 __global__ void test_complex()
0022 {
0023
0024    unsigned ix = blockIdx.x * blockDim.x + threadIdx.x;
0025    unsigned iy = blockIdx.y * blockDim.y + threadIdx.y;
0026
0027    cuFloatComplex a = make_cuFloatComplex(ix, iy);
0028    cuFloatComplex b = tcomplex::cuSqrtf(a);
0029
0030    float  a_rho = tcomplex::cuRhof(a);
0031    float  a_the = tcomplex::cuThetaf(a);
0032
0033    float  b_rho = tcomplex::cuRhof(b);
0034    float  b_the = tcomplex::cuThetaf(b);
0035
0036    printf("//test_complex  (ix,iy) (%2i, %2i) (x,y) (%10.4f, %10.4f) a(rho,the)(%10.4f, %10.4f)  b(rho,the)(%10.4f, %10.4f ) \n",
0037          ix, iy, a.x, a.y, a_rho, a_the, b_rho, b_the );
0038 }
0039
0040
0041 void test_device_complex_api()
0042 {
0043     dim3 block(16,16);
0044     dim3 grid(1,1);
0045
0046     test_complex<<< grid , block >>>();
0047     cudaDeviceSynchronize();
0048 }
0049
0050 int main()
0051 {
0052     test_device_complex_api();
0053     return 0 ;
0054 }
0055
0056