CudaFFT.cu

Go to the documentation of this file.

#include <lal/LALDatatypes.h>
#include "CudaFunctions.h"
 
void XLALCudaError(cudaError_t error, const char *file, int line)
{
    if(error != cudaSuccess)    
    {      
        fprintf( stderr, "%s:%d %s\n", file, line, cudaGetErrorString(error));
        exit(1);
    }
}
 
void XLALCudaFFTError(cufftResult_t error, const char *file, int line)
{
    if(error != CUFFT_SUCCESS) 
    {
        /* As there are no GetErrorString function available for CUDA FFT, 
         * the error messages had to be hard-coded, 
         * and needs to be updated with new CUDA releases. 
         */
        switch( error )
        {
            case CUFFT_INVALID_PLAN:
              fprintf( stderr, "%s:%d The plan handle is invalid\n", file, line );
              break;
        
            case CUFFT_INVALID_VALUE:
              fprintf( stderr, "%s:%d The input data and/or output data is not valid\n", file, line );
              break;
              
            case CUFFT_INTERNAL_ERROR:
              fprintf( stderr, "%s:%d Internal driver error is detected\n", file, line );
              break;
 
            case CUFFT_EXEC_FAILED:
              fprintf( stderr, "%s:%d CUFFT failed to execute the transform on GPU\n", file, line );
              break;
 
            case CUFFT_SETUP_FAILED:
              fprintf( stderr, "%s:%d CUFFT library failed to initialize\n", file, line );
              break;
 
            default:
                fprintf( stderr, "%s:%d Cuda FFT Error: %d\n", file, line, error);
        }
        exit(1);
    }
}
 
int cudafft_execute_r2c(cufftHandle plan,
    cufftComplex *output, const cufftReal *input,
    cufftComplex *d_output, cufftReal *d_input,UINT4 size)
{
    UINT4 inputBytes = size * sizeof(cufftReal);
    UINT4 outputBytes = (size/2 + 1) * sizeof(cufftComplex);
 
    XLALCUDACHECK(cudaMemcpy( d_input, input, inputBytes, cudaMemcpyHostToDevice ));
 
    XLALCUDAFFTCHECK(cufftExecR2C(plan, d_input, d_output));
 
    XLALCUDACHECK(cudaMemcpy( output, d_output, outputBytes, cudaMemcpyDeviceToHost ));
 
    return 0;
}
 
int cudafft_execute_c2r(cufftHandle plan,
    cufftReal *output, const cufftComplex *input,
    cufftReal *d_output, cufftComplex *d_input, UINT4 size)
{
    UINT4 inputBytes = (size/2 + 1) * sizeof(cufftComplex);
    UINT4 outputBytes = size * sizeof(cufftReal);
 
    XLALCUDACHECK(cudaMemcpy( d_input, input, inputBytes, cudaMemcpyHostToDevice ));
 
    XLALCUDAFFTCHECK(cufftExecC2R(plan, d_input, d_output));
 
    XLALCUDACHECK(cudaMemcpy( output, d_output, outputBytes, cudaMemcpyDeviceToHost ));
 
    return 0;
}
 
int cudafft_execute_c2c(cufftHandle plan,
    cufftComplex *output, const cufftComplex *input,
    cufftComplex *d_output, cufftComplex *d_input,
    INT4 direction, UINT4 size)
{
    UINT4 nBytes = size * sizeof(cufftComplex);
 
    XLALCUDACHECK(cudaMemcpy( d_input, input, nBytes, cudaMemcpyHostToDevice ));
 
    XLALCUDAFFTCHECK(cufftExecC2C(plan, d_input, d_output, direction));
 
    XLALCUDACHECK(cudaMemcpy( output, d_output, nBytes, cudaMemcpyDeviceToHost ));
 
    return 0;
}