lecture3.ipynb

lecture3.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "provenance": [],
      "authorship_tag": "ABX9TyMlUwUw6zev4Hfkr2kn01bb",
      "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/gist/mscroggs/1178fa4d64b44eaa231432a3b795be17/lecture3.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 9,
      "metadata": {
        "id": "mBAm68OWvxYa"
      },
      "outputs": [],
      "source": [
        "import numpy as np\n",
        "import numba\n",
        "from numba import cuda\n",
        "from time import time"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "cuda.detect()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "DkXhP_Xev1cY",
        "outputId": "dac59f60-3fb0-4a3f-8a87-1618a045f168"
      },
      "execution_count": 2,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Found 1 CUDA devices\n",
            "id 0             b'Tesla T4'                              [SUPPORTED]\n",
            "                      Compute Capability: 7.5\n",
            "                           PCI Device ID: 4\n",
            "                              PCI Bus ID: 0\n",
            "                                    UUID: GPU-34e4039e-ae51-fd83-6c61-865ddcc43b66\n",
            "                                Watchdog: Disabled\n",
            "             FP32/FP64 Performance Ratio: 32\n",
            "Summary:\n",
            "\t1/1 devices are supported\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "True"
            ]
          },
          "metadata": {},
          "execution_count": 2
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "a = np.random.rand(100000000).astype(\"float32\")\n",
        "b = np.random.rand(100000000).astype(\"float32\")\n",
        "\n",
        "result = np.empty(100000000, dtype=\"float32\")"
      ],
      "metadata": {
        "id": "_wkTAbyo1oy3"
      },
      "execution_count": 50,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "start = time()\n",
        "result = a + b\n",
        "print(time() - start)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "EjBsAwYD2Pkf",
        "outputId": "996237cf-cbe9-4ca6-d88e-e997c7f2a594"
      },
      "execution_count": 53,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "0.12229013442993164\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "@numba.njit\n",
        "def add(a, b, result):\n",
        "  for i in range(100000000):\n",
        "    result[i] = a[i] + b[i]"
      ],
      "metadata": {
        "id": "uGJFTBVX2bVC"
      },
      "execution_count": 54,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "result = np.empty(100000000, dtype=\"float32\")\n",
        "\n",
        "start = time()\n",
        "add(a, b, result)\n",
        "print(time() - start)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "v-YVera02pt0",
        "outputId": "fab415e3-d73f-4207-bda1-deb75c36c361"
      },
      "execution_count": 58,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "0.12056088447570801\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "@cuda.jit\n",
        "def add2(a, b, result):\n",
        "  for i in range(100000000):\n",
        "    result[i] = a[i] + b[i]"
      ],
      "metadata": {
        "id": "v6xJt4_A29bj"
      },
      "execution_count": 59,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "result = np.empty(100000000, dtype=\"float32\")\n",
        "\n",
        "griddim = (128,)\n",
        "blockdim = (64,)\n",
        "\n",
        "a2 = cuda.to_device(a)\n",
        "b2 = cuda.to_device(b)\n",
        "r2 = cuda.to_device(result)\n",
        "\n",
        "start = time()\n",
        "\n",
        "\n",
        "\n",
        "add2[griddim, blockdim](a2, b2, r2)\n",
        "print(time() - start)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "GlYJn_ZS3lz0",
        "outputId": "b9a99ee9-376d-41e5-956c-9db4577ccd06"
      },
      "execution_count": 66,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "0.0003771781921386719\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [],
      "metadata": {
        "id": "kNN96rcr3n73"
      },
      "execution_count": null,
      "outputs": []
    }
  ]
}