Program Listing for File NuMagSANSlib_StructureData.h

↰ Return to documentation for file (src/InputData/StructData/NuMagSANSlib_StructureData.h)

// File         : NuMagSANSlib_StructureData.h
// Author       : Michael Philipp ADAMS, M.Sc.
// Company      : University of Luxembourg
// Department   : Department of Physics and Materials Sciences
// Group        : NanoMagnetism Group
// Group Leader : Prof. Andreas Michels
// Version      : 23 November 2024
// OS           : Linux Ubuntu
// Language     : CUDA C++

#include <iostream>
#include <fstream>
#include <sstream>
#include <sys/stat.h>
#include <sys/types.h>
#include <math.h>
#include <string>
#include <vector>
#include <stdlib.h>
#include <time.h>
#include <cuda_runtime.h>
#include <cublas_v2.h>
#include <stdexcept>
#include <math.h>
#include <chrono>
#include <dirent.h>
#include <unistd.h>

using namespace std;

struct StructureData {

    float* RotMat;        // rotation matrix R = [R0, R3, R6; R1, R4, R7; R2, R5, R8]
    float* x;             // position data x
    float* y;             // position data y
    float* z;             // position data z
    unsigned long int* K; // number of objects
};

void allocate_StructureDataRAM(StructureData* StructData, StructDataProperties* StructDataProp,
                               InputFileData* InputData) {

    unsigned long int K = StructDataProp->Number_Of_Elements;

    StructData->x = (float*)malloc(K * sizeof(float));
    StructData->y = (float*)malloc(K * sizeof(float));
    StructData->z = (float*)malloc(K * sizeof(float));
    StructData->K = (unsigned long int*)malloc(sizeof(unsigned long int));
    StructData->RotMat = (float*)malloc(9 * sizeof(float));

    if (!StructData->x || !StructData->y || !StructData->z || !StructData->K || !StructData->RotMat) {
        perror("Memory allocation failed");
        exit(EXIT_FAILURE);
    }

    *StructData->K = K;

    for (unsigned long int i = 0; i < K; i++) {
        StructData->x[i] = 0.0;
        StructData->y[i] = 0.0;
        StructData->z[i] = 0.0;
    }
}

void allocate_StructureDataGPU(StructureData* StructData, StructureData* StructData_gpu) {

    unsigned long int K = *StructData->K;

    cudaMalloc(&StructData_gpu->x, K * sizeof(float));
    cudaMalloc(&StructData_gpu->y, K * sizeof(float));
    cudaMalloc(&StructData_gpu->z, K * sizeof(float));
    cudaMalloc(&StructData_gpu->K, sizeof(unsigned long int));
    cudaMalloc(&StructData_gpu->RotMat, 9 * sizeof(float));

    cudaMemcpy(StructData_gpu->K, StructData->K, sizeof(unsigned long int), cudaMemcpyHostToDevice);
    cudaMemcpy(StructData_gpu->RotMat, StructData->RotMat, 9 * sizeof(float), cudaMemcpyHostToDevice);

    LogSystem::write("");
    // copy data from Host to Device
    LogSystem::write("copy data from RAM to GPU...");
    cudaMemcpy(StructData_gpu->x, StructData->x, K * sizeof(float), cudaMemcpyHostToDevice);
    LogSystem::write("   x done...");
    cudaMemcpy(StructData_gpu->y, StructData->y, K * sizeof(float), cudaMemcpyHostToDevice);
    LogSystem::write("   y done...");
    cudaMemcpy(StructData_gpu->z, StructData->z, K * sizeof(float), cudaMemcpyHostToDevice);
    LogSystem::write("   z done...");
    LogSystem::write("");
    LogSystem::write("data transfer finished...");
    LogSystem::write("");

    cudaMalloc(&StructData_gpu, sizeof(StructureData));
    cudaMemcpy(StructData_gpu, StructData, sizeof(StructureData), cudaMemcpyHostToDevice);
}

void copy_StructureDataRAM2GPU(StructureData* StructData, StructureData* StructData_gpu) {

    unsigned long int K = *StructData->K;

    cudaMemcpy(StructData_gpu->K, StructData->K, sizeof(unsigned long int), cudaMemcpyHostToDevice);
    cudaMemcpy(StructData_gpu->RotMat, StructData->RotMat, 9 * sizeof(float), cudaMemcpyHostToDevice);

    LogSystem::write("");
    LogSystem::write("copy data from RAM to GPU...");
    cudaMemcpy(StructData_gpu->x, StructData->x, K * sizeof(float), cudaMemcpyHostToDevice);
    LogSystem::write("   x done...");
    cudaMemcpy(StructData_gpu->y, StructData->y, K * sizeof(float), cudaMemcpyHostToDevice);
    LogSystem::write("   y done...");
    cudaMemcpy(StructData_gpu->z, StructData->z, K * sizeof(float), cudaMemcpyHostToDevice);
    LogSystem::write("   z done...");
    LogSystem::write("");
    LogSystem::write("data transfer finished...");
    LogSystem::write("");
}

void read_StructureData(StructureData* StructData, StructDataProperties* StructDataProp, InputFileData* InputData) {

    LogSystem::write("");
    LogSystem::write("read StructData...");

    // unsigned long int K = *StructData->K;

    for (int i = 0; i < 9; i++) {
        StructData->RotMat[i] = InputData->RotMat[i];
    }

    string filename;
    unsigned long int n = 0;
    float x_buf, y_buf, z_buf;
    ifstream fin;

    filename = StructDataProp->GlobalFilePath;
    LogSystem::write(filename);

    fin.open(filename);
    n = 0;
    // read in the data
    while (fin >> x_buf >> y_buf >> z_buf) {
        StructData->x[n] = x_buf * InputData->XYZ_Unit_Factor;
        StructData->y[n] = y_buf * InputData->XYZ_Unit_Factor;
        StructData->z[n] = z_buf * InputData->XYZ_Unit_Factor;
        n += 1;
    }

    fin.close();
    LogSystem::write("read (x,y,z) StructData finished...");
}

void init_StructureData(StructureData* StructData, StructureData* StructData_gpu, StructDataProperties* StructDataProp,
                        InputFileData* InputData) {

    allocate_StructureDataRAM(StructData, StructDataProp, InputData);
    read_StructureData(StructData, StructDataProp, InputData);
    allocate_StructureDataGPU(StructData, StructData_gpu);
}

void init_StructureDataMemory(StructureData* StructData, StructureData* StructData_gpu,
                              StructDataProperties* StructDataProp, InputFileData* InputData) {

    allocate_StructureDataRAM(StructData, StructDataProp, InputData);
    allocate_StructureDataGPU(StructData, StructData_gpu);
}

void new_read_StructureData(StructureData* StructData, StructureData* StructData_gpu,
                            StructDataProperties* StructDataProp, InputFileData* InputData) {

    read_StructureData(StructData, StructDataProp, InputData);
    copy_StructureDataRAM2GPU(StructData, StructData_gpu);
}

void free_StructureData(StructureData* StructData, StructureData* StructData_gpu) {

    LogSystem::write("free StructData...");
    free(StructData->x);
    free(StructData->y);
    free(StructData->z);
    free(StructData->K);
    free(StructData->RotMat);

    cudaDeviceSynchronize();
    cudaFree(StructData_gpu->x);
    cudaDeviceSynchronize();
    cudaFree(StructData_gpu->y);
    cudaDeviceSynchronize();
    cudaFree(StructData_gpu->z);
    cudaDeviceSynchronize();
    cudaFree(StructData_gpu->K);
    cudaDeviceSynchronize();
    cudaFree(StructData_gpu->RotMat);
    cudaDeviceSynchronize();

    LogSystem::write("free StructData finished.");
}

void disp_StructureData(StructureData* StructData) {
    for (int k = 0; k < *StructData->K; k++) {
        cout << StructData->x[k] << " " << StructData->y[k] << " " << StructData->z[k] << "\n";
    }
}