Week8/Conv1D_Modified.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "provenance": [],
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/mmmovania/CUDA_Spring2023/blob/main/Week8/Conv1D_Modified.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 3,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "OB2MY3st-VUm",
        "outputId": "759f3bc9-59c3-4e4b-d9c7-39e4cf20111e"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Looking in indexes: https://pypi.org/simple, https://us-python.pkg.dev/colab-wheels/public/simple/\n",
            "Collecting git+https://github.com/andreinechaev/nvcc4jupyter.git\n",
            "  Cloning https://github.com/andreinechaev/nvcc4jupyter.git to /tmp/pip-req-build-y_7eregx\n",
            "  Running command git clone --filter=blob:none --quiet https://github.com/andreinechaev/nvcc4jupyter.git /tmp/pip-req-build-y_7eregx\n",
            "  Resolved https://github.com/andreinechaev/nvcc4jupyter.git to commit aac710a35f52bb78ab34d2e52517237941399eff\n",
            "  Preparing metadata (setup.py) ... \u001b[?25l\u001b[?25hdone\n",
            "The nvcc_plugin extension is already loaded. To reload it, use:\n",
            "  %reload_ext nvcc_plugin\n"
          ]
        }
      ],
      "source": [
        "!pip install git+https://github.com/andreinechaev/nvcc4jupyter.git\n",
        "%load_ext nvcc_plugin"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "%%cu\n",
        "#include <stdio.h>\n",
        "\n",
        "\n",
        "inline cudaError_t checkCudaErr(cudaError_t err, const char* msg) {\n",
        "  if (err != cudaSuccess) {\n",
        "    fprintf(stderr, \"CUDA Runtime error at %s: %s\\n\", msg, cudaGetErrorString(err));\n",
        "  }\n",
        "  return err;\n",
        "}\n",
        "\n",
        "#define MASK_WIDTH 5\n",
        "__constant__ float M[MASK_WIDTH]; \n",
        "\n",
        "__global__ void convolution_1D_basic_kernel(float* N, float* P, int Mask_Width, int Width)\n",
        "{\n",
        "  int i = blockIdx.x*blockDim.x+threadIdx.x;  \n",
        "  float Pvalue=0; \n",
        "  int N_start_point = i-(Mask_Width/2); \n",
        " \n",
        "  for (int j=0; j<Mask_Width; j++) \n",
        "  {\n",
        "      if(N_start_point+j>=0 && N_start_point+j< Width) \n",
        "      { \n",
        "        Pvalue += N[N_start_point+j]*M[j]; \n",
        "      } \n",
        "  }\n",
        "  P[i]=Pvalue; \n",
        "} \n",
        "\n",
        " \n",
        "int main() { \n",
        "\t\tfloat   *a,  *c=0; \n",
        "    const int N = 16; \n",
        "    const int threadsPerBlock = 4;\n",
        "\n",
        "    float h_M[MASK_WIDTH]={3,4,5,4,3};\n",
        "\n",
        "    // Allocate Unified Memory -- accessible from CPU or GPU\n",
        "    checkCudaErr(cudaMallocManaged(&a, N*sizeof(float)), \"cudaMallocManaged1\");\n",
        "    checkCudaErr(cudaMallocManaged(&c, N*sizeof(float)), \"cudaMallocManaged1\");\n",
        "     \n",
        "    // fill in the memory with data\n",
        "    for (int i=0; i<N; i++) {\n",
        "        a[i] = i+1;\n",
        "        c[i] = 0; \n",
        "    }  \n",
        " \n",
        "    cudaMemcpyToSymbol(M,h_M,MASK_WIDTH*sizeof(float));\n",
        "    const int blocksPerGrid =  (N / threadsPerBlock);\n",
        " \n",
        "    // Prefetch the data to the GPU\n",
        "    int device = -1;\n",
        "    cudaGetDevice(&device);\n",
        "    cudaMemPrefetchAsync(a, N*sizeof(float), device, NULL);\n",
        "    cudaMemPrefetchAsync(M, MASK_WIDTH*sizeof(float), device, NULL);\n",
        "\n",
        "    //lets time the conv1D kernel\n",
        "    cudaEvent_t start, stop;\n",
        "    float gpu_elapsed_time_ms=0;\n",
        "    cudaEventCreate(&start);\n",
        "    cudaEventCreate(&stop);\n",
        "\n",
        "     // start to count execution time \n",
        "    cudaEventRecord(start, 0);\n",
        "    convolution_1D_basic_kernel<<<blocksPerGrid,threadsPerBlock>>>(a, c, MASK_WIDTH, N);\n",
        "    \n",
        "    cudaDeviceSynchronize();\n",
        " \n",
        "    // time counting terminate\n",
        "    cudaEventRecord(stop, 0);\n",
        "    cudaEventSynchronize(stop);\n",
        "\n",
        "    // compute time elapse on GPU computing\n",
        "    cudaEventElapsedTime(&gpu_elapsed_time_ms, start, stop);\n",
        "\n",
        "    //output the result\n",
        "    printf(\"Conv1D Result calculated in %3.3f msecs: \\n\", gpu_elapsed_time_ms);\n",
        "    for(int i=0; i<N; ++i)\n",
        "      printf(\" p[%d] = %3.3f\\n\", i, c[i]);\n",
        "\n",
        "    // free memory on the gpu side\n",
        "    checkCudaErr( cudaFree( a ) , \"cudaFree1\"); \n",
        "    checkCudaErr( cudaFree( c ) , \"cudaFree3\");\n",
        "\t\tcheckCudaErr( cudaDeviceReset(), \"cudaDeviceReset\");\n",
        "\n",
        "\t\treturn 0;\n",
        "}"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "GYZ627HO-XKd",
        "outputId": "591e2922-4662-4257-9c7b-3086703b307b"
      },
      "execution_count": 16,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Conv1D Result calculated in 0.033 msecs: \n",
            " p[0] = 22.000\n",
            " p[1] = 38.000\n",
            " p[2] = 57.000\n",
            " p[3] = 76.000\n",
            " p[4] = 95.000\n",
            " p[5] = 114.000\n",
            " p[6] = 133.000\n",
            " p[7] = 152.000\n",
            " p[8] = 171.000\n",
            " p[9] = 190.000\n",
            " p[10] = 209.000\n",
            " p[11] = 228.000\n",
            " p[12] = 247.000\n",
            " p[13] = 266.000\n",
            " p[14] = 234.000\n",
            " p[15] = 182.000\n",
            "\n"
          ]
        }
      ]
    }
  ]
}