В этом домашнем задании мне нужно завершить код для умножения двух прямоугольных матриц, используя CUDA C. После того, как я выполнил код, я отправил, и решение было правильным для набора данных, когда матрицы были квадратными, в то время как результат не соответствовал ожидаемому значение, когда матрицы не были квадратными.

Вот код после того, как я добавил недостающие части:

#include    <wb.h>

#define wbCheck(stmt) do {                             \
    cudaError_t err = stmt;                            \
    if (err != cudaSuccess) {                          \
        wbLog(ERROR, "Failed to run stmt ", #stmt);    \
        return -1;                                     \
    }                                                  \
} while(0)

// Compute C = A * B
__global__ void matrixMultiply(float * A, float * B, float * C,
               int numARows, int numAColumns,
               int numBRows, int numBColumns,
               int numCRows, int numCColumns) {
   //@@ Insert code to implement matrix multiplication here
   int Row = blockIdx.y * blockDim.y + threadIdx.y;
   int Col = blockIdx.x * blockDim.x + threadIdx.x;
   if (numAColumns != numBRows) return ;
   if ((Row < numARows) && (Col < numBColumns)){
       float Cvalue = 0;
       for (int k = 0 ; k < numAColumns ; ++k )
       Cvalue += A[Row*numAColumns + k] * B[k * numBRows + Col];
       C[Row*numAColumns + Col] = Cvalue;
     }

    }



int main(int argc, char ** argv) {
   wbArg_t args;
   float * hostA; // The A matrix
   float * hostB; // The B matrix
   float * hostC; // The output C matrix
   float * deviceA;
   float * deviceB;
   float * deviceC;
   int numARows; // number of rows in the matrix A
   int numAColumns; // number of columns in the matrix A
   int numBRows; // number of rows in the matrix B
   int numBColumns; // number of columns in the matrix B
   int numCRows; // number of rows in the matrix C (you have to set this)
   int numCColumns; // number of columns in the matrix C (you have to set this)

   args = wbArg_read(argc, argv);

   wbTime_start(Generic, "Importing data and creating memory on host");
   hostA = (float *) wbImport(wbArg_getInputFile(args, 0), &numARows, &numAColumns);
   hostB = (float *) wbImport(wbArg_getInputFile(args, 1), &numBRows, &numBColumns);
   //@@ Set numCRows and numCColumns  
   numCRows = 0;
   numCColumns = 0;
   numCRows = numARows;
   numCColumns = numBColumns;  
   //@@ Allocate the hostC matrix
   hostC = (float*) malloc(sizeof(float)*numCRows*numCColumns);  
   wbTime_stop(Generic, "Importing data and creating memory on host");

   wbLog(TRACE, "The dimensions of A are ", numARows, " x ", numAColumns);
   wbLog(TRACE, "The dimensions of B are ", numBRows, " x ", numBColumns);

   wbTime_start(GPU, "Allocating GPU memory.");
   //@@ Allocate GPU memory here
   cudaMalloc((void**)&deviceA ,sizeof(float)*numARows*numAColumns );
   cudaMalloc((void**)&deviceB , sizeof(float)*numBRows*numBColumns);
   cudaMalloc((void**)&deviceC , sizeof(float)*numCRows*numCColumns);  

   wbTime_stop(GPU, "Allocating GPU memory.");

   wbTime_start(GPU, "Copying input memory to the GPU.");
   //@@ Copy memory to the GPU here

   cudaMemcpy(deviceA, hostA, sizeof(float)*numARows*numAColumns, cudaMemcpyHostToDevice);
   cudaMemcpy(deviceB, hostB, sizeof(float)*numBRows*numBColumns, cudaMemcpyHostToDevice);
   wbTime_stop(GPU, "Copying input memory to the GPU.");

   //@@ Initialize the grid and block dimensions here

   dim3 DimGrid(numARows / 8 , numBColumns / 8, 1);
   dim3 DimBlock(8 , 8, 1);

   wbTime_start(Compute, "Performing CUDA computation");

   //@@ Launch the GPU Kernel here
   matrixMultiply<<<DimGrid , DimBlock>>>(deviceA , deviceB , deviceC , numARows , numAColumns, numBRows ,numBColumns , numCRows , numCColumns);  

   cudaThreadSynchronize();
   wbTime_stop(Compute, "Performing CUDA computation");

   wbTime_start(Copy, "Copying output memory to the CPU");
   //@@ Copy the GPU memory back to the CPU here
   cudaMemcpy(hostC, deviceC, sizeof(float)*numCRows*numCColumns , cudaMemcpyDeviceToHost);  

   wbTime_stop(Copy, "Copying output memory to the CPU");

   wbTime_start(GPU, "Freeing GPU Memory");
   //@@ Free the GPU memory here

   cudaFree(deviceA);
   cudaFree(deviceB);
   cudaFree(deviceC);
   wbTime_stop(GPU, "Freeing GPU Memory");

   wbSolution(args, hostC, numCRows, numCColumns);

   free(hostA);
   free(hostB);
   free(hostC);

   return 0;
}

Я надеюсь, что вы можете помочь мне найти, какая часть неверна.