SmithWatermanAlgoritam_CUDA.ipynb

{
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/LukaPavlovicGit/Concurrent-Algorithms/blob/main/SmithWatermanAlgoritam_CUDA.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "63H6wgfdD7Y4"
      },
      "source": [
        "# Paralelni algoritmi - Treći projekat\n",
        "\n",
        "- Projekat se radi individualno ili u paru\n",
        "\n",
        "- Rok za predaju projekata je **sreda 11.01.2023, do kraja dana (23:59)**. \n",
        "\n",
        "- Odbrane će biti organizovane u toku kolokvijumske nedelje.\n",
        "\n",
        "- Predaja projekata je putem e-maila, programski kod bi trebalo da\n",
        "bude prikacen uz sam e-mail. Subject e-maila treba da bude u formatu\n",
        "\"[RN3RA Projekat 1] Prezime (Prezime ukoliko se radi u paru)\". Tekst e-maila treba da bude u formi:\n",
        "\"Ime Prezime XX/GGRN\" (gde su XX i GG broje indeksa i godina upisa).\n",
        "Link ka Collab-u je opcion. Ukoliko dva\n",
        "studenta rade u paru, dovoljno je da jedan od njih preda, pri cemu\n",
        "drugog stavi u cc."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "RR7JjxGGrzys"
      },
      "source": [
        "## Smith–Waterman algoritam\n",
        "\n",
        "Smith–Waterman je algoritam za poravnavanje sekvenci baziran na principima dinamičkog programiranja. Za date dve sekvence A i B i parametre poravnanja `match`, `miss` i `gap` popunjava se matrica dimenzija $|A| \\times |B|$, tako što postavljamo $$M_{ij} = max \\begin{cases} {M_{i-1, j-1} + score(A_i, B_i)} \\\\ {M_{i-1, j} - gap}\\\\ {M_{i, j-1} - gap}\\\\ 0 \\end{cases}$$ pri čemu je: $$score(A_i, B_i) = \\begin{cases} {match \\quad \\text{ ako je } A_i = B_i }\\\\ {-miss \\quad \\text {ako je }A_i \\neq B_i } \\end{cases}$$\n",
        "\n",
        "Sve vrednosti u prvom redu i prvoj koloni su 0. Optimalna vrednost poravnanja je najveća vrednost u matrici.\n",
        "\n",
        "U klasičnoj implementaciji matrica se popunjava sekvencijalno  po kolonama ili po redovima (slika 1).   \n",
        "\n",
        "![1-s2.0-S0021999110000823-gr2.jpg](data:image/jpeg;base64,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)  \n",
        "Slika 1: Smith–Waterman matrica\n",
        "\n",
        "Zadatak ovog projekta je alternativna implementacija opisana u [Wozniak, 1997](https://academic.oup.com/bioinformatics/article/13/2/145/248735?login=false) kod koje se matrica popunjava dijagonalno (slika 2), što omogaćava da se veći broj vrednosti računa u paraleli.  \n",
        "\n",
        "![3-s2.0-B9780123849885000139-f13-04-9780123849885.jpg](data:image/jpeg;base64,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)  \n",
        "Slika 2: Popunavanja po dijagonali"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "jkh2OjIsr-OQ"
      },
      "source": [
        "### 1. CUDA paralelna implementacija izačuavanja matrice [7 bodova]\n",
        "\n",
        "Koristeći jezik CUDA (opciono kroz pyCUDA okruženje) implementriati dijagonalno popunavanje Smith-Watrman matrice poravnanja. Pretpostaviti da su dužine ulaznih sekvenci takve da će najduža dijagonala biti manja ili jednaka broju niti u jednom bloku."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "kN58bdt8BpGU"
      },
      "source": [
        "### 2. CUDA paralelna implementacija pronalaženja optimalne vrednosti poravanja [4 boda]\n",
        "\n",
        "Koristeći jezik CUDA (opciono kroz pyCUDA okruženje) implementirati paralelnu redukciju kojom se pronalazi maksimalna vrednost u matrici izračunatoj u prethodnoj tački."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "mD9PU8ZNsTiP"
      },
      "source": [
        "### 2. Upotreba deljene i konstante memorije [4 boda]\n",
        "Modifikovati rešenje iz tačke 1 tako su sekvence smeštene u:  \n",
        "a) Konstantnoj memoriji  \n",
        "b) Deljenoj memoriji, pri čemu je kopiranje u deljenu memoriju paralelno.  "
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "4oVyth1JsX5O"
      },
      "source": [
        "### 3. Podrška za dugačke sekvence [5 bodova]\n",
        "\n",
        "Proširiti rešenje iz tačke 1 tako da može da radi i sa sekvencama kod kojih je najduže dijagonala veća od broja niti u jednom bloku. Rešenje dizajnirati tako što se matrica poravanja deli na više pod-matrica takvih da se svaka pod-matrica može rešiti upotrebom rešenja iz tačke 1 (pri čemu će biti potrebna odgovrajuća inicijalizacija prvog reda i prve kolone). "
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "1dsczbD3DqJr",
        "outputId": "fab342df-215c-48dd-8503-cce341b91c59"
      },
      "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 pycuda\n",
            "  Downloading pycuda-2022.2.2.tar.gz (1.7 MB)\n",
            "\u001b[2K     \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m1.7/1.7 MB\u001b[0m \u001b[31m21.4 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
            "\u001b[?25h  Installing build dependencies ... \u001b[?25l\u001b[?25hdone\n",
            "  Getting requirements to build wheel ... \u001b[?25l\u001b[?25hdone\n",
            "  Preparing metadata (pyproject.toml) ... \u001b[?25l\u001b[?25hdone\n",
            "Collecting mako\n",
            "  Downloading Mako-1.2.4-py3-none-any.whl (78 kB)\n",
            "\u001b[2K     \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m78.7/78.7 KB\u001b[0m \u001b[31m7.1 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
            "\u001b[?25hCollecting pytools>=2011.2\n",
            "  Downloading pytools-2022.1.14.tar.gz (74 kB)\n",
            "\u001b[2K     \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m74.6/74.6 KB\u001b[0m \u001b[31m8.0 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
            "\u001b[?25h  Preparing metadata (setup.py) ... \u001b[?25l\u001b[?25hdone\n",
            "Requirement already satisfied: appdirs>=1.4.0 in /usr/local/lib/python3.8/dist-packages (from pycuda) (1.4.4)\n",
            "Requirement already satisfied: platformdirs>=2.2.0 in /usr/local/lib/python3.8/dist-packages (from pytools>=2011.2->pycuda) (2.6.2)\n",
            "Requirement already satisfied: typing_extensions>=4.0 in /usr/local/lib/python3.8/dist-packages (from pytools>=2011.2->pycuda) (4.4.0)\n",
            "Requirement already satisfied: MarkupSafe>=0.9.2 in /usr/local/lib/python3.8/dist-packages (from mako->pycuda) (2.0.1)\n",
            "Building wheels for collected packages: pycuda, pytools\n",
            "  Building wheel for pycuda (pyproject.toml) ... \u001b[?25l\u001b[?25hdone\n",
            "  Created wheel for pycuda: filename=pycuda-2022.2.2-cp38-cp38-linux_x86_64.whl size=661775 sha256=7ceff7994f071f01a7b948f2a80f0ae2b027ebc458c83e7d465ef68f75995a81\n",
            "  Stored in directory: /root/.cache/pip/wheels/7b/41/0d/7cecb04af969d283ebe4a69579a8b2baec0d010a1ac4159f7e\n",
            "  Building wheel for pytools (setup.py) ... \u001b[?25l\u001b[?25hdone\n",
            "  Created wheel for pytools: filename=pytools-2022.1.14-py2.py3-none-any.whl size=69870 sha256=7faf21f91b82acbed399b2b6b05ec60891aa7eb973954efda9a0d0aa1da4f315\n",
            "  Stored in directory: /root/.cache/pip/wheels/cb/fc/a9/1e7e56fe02d7f58eaff555f22e79d4fc2d817012291254bae2\n",
            "Successfully built pycuda pytools\n",
            "Installing collected packages: pytools, mako, pycuda\n",
            "Successfully installed mako-1.2.4 pycuda-2022.2.2 pytools-2022.1.14\n"
          ]
        }
      ],
      "source": [
        "!pip install pycuda"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "VdnYQBGg-1D-",
        "outputId": "e5257245-7a13-41cf-a44a-0e628e12d4e2"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Wed Jan 18 10:51:21 2023       \n",
            "+-----------------------------------------------------------------------------+\n",
            "| NVIDIA-SMI 460.32.03    Driver Version: 460.32.03    CUDA Version: 11.2     |\n",
            "|-------------------------------+----------------------+----------------------+\n",
            "| GPU  Name        Persistence-M| Bus-Id        Disp.A | Volatile Uncorr. ECC |\n",
            "| Fan  Temp  Perf  Pwr:Usage/Cap|         Memory-Usage | GPU-Util  Compute M. |\n",
            "|                               |                      |               MIG M. |\n",
            "|===============================+======================+======================|\n",
            "|   0  Tesla T4            Off  | 00000000:00:04.0 Off |                    0 |\n",
            "| N/A   54C    P0    26W /  70W |      0MiB / 15109MiB |      0%      Default |\n",
            "|                               |                      |                  N/A |\n",
            "+-------------------------------+----------------------+----------------------+\n",
            "                                                                               \n",
            "+-----------------------------------------------------------------------------+\n",
            "| Processes:                                                                  |\n",
            "|  GPU   GI   CI        PID   Type   Process name                  GPU Memory |\n",
            "|        ID   ID                                                   Usage      |\n",
            "|=============================================================================|\n",
            "|  No running processes found                                                 |\n",
            "+-----------------------------------------------------------------------------+\n"
          ]
        }
      ],
      "source": [
        "!nvidia-smi"
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Smith-Waterman algoritam "
      ],
      "metadata": {
        "id": "rRiyhcnB8YeD"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Prva tacka"
      ],
      "metadata": {
        "id": "MtgI19A1C2bT"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "import numpy as np\n",
        "import sys\n",
        "import math\n",
        "import pycuda.driver as cuda\n",
        "import pycuda.autoinit\n",
        "from pycuda.compiler import SourceModule\n",
        "\n",
        "mod = SourceModule(\"\"\"\n",
        "    __global__ void smith_waterman( int *mat, \n",
        "                                    int mat_width,\n",
        "                                    int mat_height,\n",
        "                                    int magic_number, \n",
        "                                    int miss, \n",
        "                                    int match, \n",
        "                                    int gap_penalty,\n",
        "                                    char arr1[], \n",
        "                                    char arr2[] ) {\n",
        "            \n",
        "            if( blockIdx.x == 0 && threadIdx.x == 0 ||\n",
        "                blockIdx.y == 0 && threadIdx.y == 0 ||\n",
        "                threadIdx.x >= mat_width ||\n",
        "                threadIdx.y >= mat_height ||\n",
        "                threadIdx.x + threadIdx.y != magic_number)\n",
        "                return;\n",
        "\n",
        "\n",
        "                int idx = threadIdx.x + threadIdx.y*mat_width;\n",
        "                int score = arr2[threadIdx.x - 1] == arr1[threadIdx.y - 1] ? match : miss;\n",
        "                int max = 0;\n",
        "                max = max > mat[idx - 1] - gap_penalty ? max : mat[idx - 1] - gap_penalty;                 // max > mat[i][j-1] - gap_penalty\n",
        "                max = max > mat[idx - mat_width] - gap_penalty ? max : mat[idx - mat_width] - gap_penalty; // max > mat[i-1][j] - gap_penalty\n",
        "                max = max > mat[idx - mat_width - 1] + score ? max : mat[idx - mat_width - 1] + score;     // max > mat[i-1][j-1] + score\n",
        "\n",
        "                mat[idx] = max; \n",
        "        }\n",
        "\"\"\")\n",
        "\n",
        "np.set_printoptions(threshold=sys.maxsize)\n",
        "\n",
        "gap_penalty = 2\n",
        "match = 3\n",
        "miss = -3\n",
        "\n",
        "arr1 = np.array(['c', 'a', 'g', 'c','c', 'a', 'g', 'c','c','c', 'a', 'r', 'c', 'c', 'c']) # height\n",
        "arr2 = np.array(['c', 'a', 'g', 'c','c', 'a', 'g', 'c','c','c', 'a', 'r', 'c', 'c', 'c', 'a', 'r']) # width\n",
        "\n",
        "mat = np.zeros([len(arr1) + 1, len(arr2) + 1], dtype=np.int32)\n",
        "\n",
        "arr1_gpu = cuda.mem_alloc(arr1.nbytes) # alociranje CUDA memorije\n",
        "arr2_gpu = cuda.mem_alloc(arr2.nbytes) # alociranje CUDA memorije\n",
        "mat_gpu = cuda.mem_alloc(mat.nbytes)   # alociranje CUDA memorije\n",
        "\n",
        "cuda.memcpy_htod(arr1_gpu, arr1)  # kopiranje Host TO Device\n",
        "cuda.memcpy_htod(arr2_gpu, arr2)  # kopiranje Host TO Device\n",
        "cuda.memcpy_htod(mat_gpu, mat)    # kopiranje Host TO Device\n",
        "\n",
        "func = mod.get_function(\"smith_waterman\")\n",
        "\n",
        "height = len(arr1) + 1\n",
        "width = len(arr2) + 1\n",
        "num_of_diagonals = (height + width - 1)\n",
        "\n",
        "for k in range(num_of_diagonals):\n",
        "  func( mat_gpu, \n",
        "        np.int32(width),\n",
        "        np.int32(height),\n",
        "        np.int32(k), \n",
        "        np.int32(miss),\n",
        "        np.int32(match), \n",
        "        np.int32(gap_penalty),\n",
        "        arr1_gpu, \n",
        "        arr2_gpu, \n",
        "        block=(32, 32, 1), \n",
        "        grid=(math.ceil(width/32), math.ceil(height/32), 1) ) \n",
        "\n",
        "cuda.memcpy_dtoh(arr1, arr1_gpu)  # kopiranje Device TO Host\n",
        "cuda.memcpy_dtoh(arr2, arr2_gpu)  # kopiranje Device TO Host\n",
        "cuda.memcpy_dtoh(mat, mat_gpu)    # kopiranje Device TO Host\n",
        "\n",
        "print(mat)"
      ],
      "metadata": {
        "id": "VYO_pau1CwQ6",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "124cdd0a-fc94-4bd4-8e06-cdeebdc77b81"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "[[ 0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            " [ 0  3  1  0  0  0  0  0  0  0  0  0  0  3  1  0  0  3]\n",
            " [ 0  1  6  4  3  1  3  3  3  1  3  3  3  1  6  4  3  1]\n",
            " [ 0  0  4  9  7  5  4  6  6  4  4  6  6  4  4  9  7  5]\n",
            " [ 0  0  3  7 12 10  8  7  9  7  7  7  9  7  7  7 12 10]\n",
            " [ 0  0  1  5 10 15 13 11  9  7  5  5  7  6  5  5 10  9]\n",
            " [ 0  0  3  4  8 13 18 16 14 12 10  8  8  6  9  8  8  7]\n",
            " [ 0  0  3  6  7 11 16 21 19 17 15 13 11  9  9 12 11  9]\n",
            " [ 0  0  3  6  9  9 14 19 24 22 20 18 16 14 12 12 15 13]\n",
            " [ 0  0  1  4  7  7 12 17 22 27 25 23 21 19 17 15 13 12]\n",
            " [ 0  0  3  4  7  5 10 15 20 25 30 28 26 24 22 20 18 16]\n",
            " [ 0  0  3  6  7  5  8 13 18 23 28 33 31 29 27 25 23 21]\n",
            " [ 0  0  3  6  9  7  8 11 16 21 26 31 36 34 32 30 28 26]\n",
            " [ 0  3  1  4  7  6  6  9 14 19 24 29 34 39 37 35 33 31]\n",
            " [ 0  1  6  4  7  5  9  9 12 17 22 27 32 37 42 40 38 36]\n",
            " [ 0  0  4  9  7  5  8 12 12 15 20 25 30 35 40 45 43 41]]\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Druga tacka"
      ],
      "metadata": {
        "id": "o0hLLOA5fDOO"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "import pycuda.driver as cuda\n",
        "import pycuda.autoinit\n",
        "from pycuda.compiler import SourceModule\n",
        "import math\n",
        "import numpy as np\n",
        "\n",
        "mod = SourceModule(\"\"\"\n",
        "     __global__ void max_value(int *a, int *result){\n",
        "     \n",
        "       extern __shared__ int data[];\n",
        "       int idx = threadIdx.x;\n",
        "       \n",
        "       data[idx] = a[idx];\n",
        "       __syncthreads();\n",
        "       \n",
        "       for(unsigned int s=blockDim.x/2; s >= 1; s = s/2) { \n",
        "          if (idx < s){\n",
        "            data[idx] = max(data[idx], data[idx+s]); \n",
        "          }   \n",
        "          __syncthreads(); \n",
        "       }\n",
        "       \n",
        "       // write result for this block to global mem\n",
        "       if (idx == 0) result[0] = data[0];\n",
        "\n",
        "    }\n",
        "   \"\"\")\n",
        "\n",
        "\n",
        "result = np.zeros(1, dtype=np.int32)\n",
        "\n",
        "mat_gpu = cuda.mem_alloc(mat.nbytes)\n",
        "result_gpu = cuda.mem_alloc(result.nbytes)\n",
        "\n",
        "cuda.memcpy_htod(mat_gpu, mat)        # kopiranje Host TO Device\n",
        "cuda.memcpy_htod(result_gpu, result)  # kopiranje Host TO Device\n",
        "\n",
        "fun = mod.get_function(\"max_value\")\n",
        "\n",
        "length = mat.size\n",
        "fun(  mat_gpu, \n",
        "      result_gpu,\n",
        "      shared=mat.nbytes, \n",
        "      block=(length, 1, 1), \n",
        "      grid=(1, 1, 1)  )\n",
        "\n",
        "cuda.memcpy_dtoh(result, result_gpu)\n",
        "\n",
        "print(result)"
      ],
      "metadata": {
        "id": "vOXZFErpSFDW",
        "outputId": "8226e3c5-e2e4-481c-adfa-147f13b96d7e",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "[45]\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [],
      "metadata": {
        "id": "yTq1G5p8DDuU"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Treca tacka pod a)"
      ],
      "metadata": {
        "id": "9h3GiJTNDEwG"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "import numpy as np\n",
        "import sys\n",
        "import math\n",
        "import pycuda.driver as cuda\n",
        "import pycuda.autoinit\n",
        "from pycuda.compiler import SourceModule\n",
        "\n",
        "mod = SourceModule(\"\"\"\n",
        "\n",
        "    __constant__  int mat_const[1024];\n",
        "    __constant__  char arr1_const[128];\n",
        "    __constant__  char arr2_const[128];\n",
        "\n",
        "    __global__ void smith_waterman_const( int *mat,\n",
        "                                          int mat_width,\n",
        "                                          int mat_height,\n",
        "                                          int magic_number, \n",
        "                                          int miss, \n",
        "                                          int match, \n",
        "                                          int gap_penalty ) {\n",
        "            \n",
        "            if( blockIdx.x == 0 && threadIdx.x == 0 ||\n",
        "                blockIdx.y == 0 && threadIdx.y == 0 ||\n",
        "                threadIdx.x >= mat_width ||\n",
        "                threadIdx.y >= mat_height ||\n",
        "                threadIdx.x + threadIdx.y != magic_number)\n",
        "                return;\n",
        "\n",
        "                \n",
        "                int idx = threadIdx.x + threadIdx.y*mat_width;\n",
        "                int score = arr2_const[threadIdx.x - 1] == arr1_const[threadIdx.y - 1] ? match : miss;\n",
        "                int max = 0;\n",
        "                max = max > mat_const[idx - 1] - gap_penalty ? max : mat_const[idx - 1] - gap_penalty;                 // max > mat[i][j-1] - gap_penalty\n",
        "                max = max > mat_const[idx - mat_width] - gap_penalty ? max : mat_const[idx - mat_width] - gap_penalty; // max > mat[i-1][j] - gap_penalty\n",
        "                max = max > mat_const[idx - mat_width - 1] + score ? max : mat_const[idx - mat_width - 1] + score;     // max > mat[i-1][j-1] + score\n",
        "\n",
        "                mat[idx] = max;\n",
        "        }\n",
        "\"\"\")\n",
        "\n",
        "\n",
        "np.set_printoptions(threshold=sys.maxsize)\n",
        "\n",
        "gap_penalty = 2\n",
        "match = 3\n",
        "miss = -3\n",
        "\n",
        "arr1 = np.array(['c', 'a', 'g', 'c','c', 'a', 'g', 'c','c','c', 'a', 'r', 'c', 'c', 'c']) # height\n",
        "arr2 = np.array(['c', 'a', 'g', 'c','c', 'a', 'g', 'c','c','c', 'a', 'r', 'c', 'c', 'c', 'a', 'r']) # width\n",
        "\n",
        "mat = np.zeros([len(arr1) + 1, len(arr2) + 1], dtype=np.int32)\n",
        "\n",
        "\n",
        "arr1_gpu = mod.get_global('arr1_const')[0]   # alociranje CUDA memorije\n",
        "arr2_gpu = mod.get_global('arr2_const')[0]   # alociranje CUDA memorije\n",
        "\n",
        "cuda.memcpy_htod(arr1_gpu, arr1)  # kopiranje Host TO Device\n",
        "cuda.memcpy_htod(arr2_gpu, arr2)  # kopiranje Host TO Device\n",
        "\n",
        "func = mod.get_function(\"smith_waterman_const\")\n",
        "\n",
        "height = len(arr1) + 1\n",
        "width = len(arr2) + 1\n",
        "num_of_diagonals = (height + width - 1)\n",
        "\n",
        "for k in range(num_of_diagonals):\n",
        "\n",
        "  mat_gpu = cuda.mem_alloc(mat.nbytes)   # alociranje CUDA memorije\n",
        "  mat_gpu_c = mod.get_global('mat_const')[0]      # alociranje CUDA memorije\n",
        "\n",
        "  cuda.memcpy_htod(mat_gpu_c, mat)  # kopiranje Host TO Device\n",
        "  cuda.memcpy_htod(mat_gpu, mat)  # kopiranje Host TO Device\n",
        "\n",
        "  func( mat_gpu,\n",
        "        np.int32(width),\n",
        "        np.int32(height),\n",
        "        np.int32(k), \n",
        "        np.int32(miss),\n",
        "        np.int32(match), \n",
        "        np.int32(gap_penalty),\n",
        "        block=(32, 32, 1), \n",
        "        grid=(math.ceil(width/32), math.ceil(height/32), 1) )\n",
        "  \n",
        "  cuda.memcpy_dtoh(mat, mat_gpu)    # kopiranje Device TO Host\n",
        "  \n",
        "cuda.memcpy_dtoh(arr1, arr1_gpu)  # kopiranje Device TO Host\n",
        "cuda.memcpy_dtoh(arr2, arr2_gpu)  # kopiranje Device TO Host\n",
        "cuda.memcpy_dtoh(mat, mat_gpu)    # kopiranje Device TO Host\n",
        "\n",
        "print(mat)"
      ],
      "metadata": {
        "id": "LNroSslwDIHF",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "bc37864b-3aa3-495c-d11d-e0d9c10e2d98"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "[[ 0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            " [ 0  3  1  0  0  0  0  0  0  0  0  0  0  3  1  0  0  3]\n",
            " [ 0  1  6  4  3  1  3  3  3  1  3  3  3  1  6  4  3  1]\n",
            " [ 0  0  4  9  7  5  4  6  6  4  4  6  6  4  4  9  7  5]\n",
            " [ 0  0  3  7 12 10  8  7  9  7  7  7  9  7  7  7 12 10]\n",
            " [ 0  0  1  5 10 15 13 11  9  7  5  5  7  6  5  5 10  9]\n",
            " [ 0  0  3  4  8 13 18 16 14 12 10  8  8  6  9  8  8  7]\n",
            " [ 0  0  3  6  7 11 16 21 19 17 15 13 11  9  9 12 11  9]\n",
            " [ 0  0  3  6  9  9 14 19 24 22 20 18 16 14 12 12 15 13]\n",
            " [ 0  0  1  4  7  7 12 17 22 27 25 23 21 19 17 15 13 12]\n",
            " [ 0  0  3  4  7  5 10 15 20 25 30 28 26 24 22 20 18 16]\n",
            " [ 0  0  3  6  7  5  8 13 18 23 28 33 31 29 27 25 23 21]\n",
            " [ 0  0  3  6  9  7  8 11 16 21 26 31 36 34 32 30 28 26]\n",
            " [ 0  3  1  4  7  6  6  9 14 19 24 29 34 39 37 35 33 31]\n",
            " [ 0  1  6  4  7  5  9  9 12 17 22 27 32 37 42 40 38 36]\n",
            " [ 0  0  4  9  7  5  8 12 12 15 20 25 30 35 40 45 43 41]]\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Treca tacka pod b)"
      ],
      "metadata": {
        "id": "GdoM8_Y60PiU"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "import numpy as np\n",
        "import sys\n",
        "import math\n",
        "import pycuda.driver as cuda\n",
        "import pycuda.autoinit\n",
        "from pycuda.compiler import SourceModule\n",
        "\n",
        "mod = SourceModule(\"\"\"\n",
        "    __global__ void smith_waterman_shared( int *mat, \n",
        "                                            int mat_width,\n",
        "                                            int mat_height,\n",
        "                                            int magic_number, \n",
        "                                            int miss, \n",
        "                                            int match, \n",
        "                                            int gap_penalty,\n",
        "                                            char arr1[], \n",
        "                                            char arr2[] ) {\n",
        "              \n",
        "            if( threadIdx.x >= mat_width ||\n",
        "                threadIdx.y >= mat_height )\n",
        "              return;\n",
        "\n",
        "            extern __shared__ int mat_shared[1024];\n",
        "            extern __shared__ int arr1_shared[32];\n",
        "            extern __shared__ int arr2_shared[32];\n",
        "            extern __shared__ int miss_match[32][32];\n",
        "\n",
        "            int idx = threadIdx.x + threadIdx.y*mat_width;\n",
        "\n",
        "            mat_shared[idx] = mat[idx];\n",
        "\n",
        "            if(threadIdx.y == 0)\n",
        "              arr2_shared[threadIdx.x] = arr2[threadIdx.x];\n",
        "            if(threadIdx.x == 0)\n",
        "              arr1_shared[threadIdx.y] = arr1[threadIdx.y];\n",
        "\n",
        "            __syncthreads();\n",
        "\n",
        "            if( blockIdx.x == 0 && threadIdx.x == 0 ||\n",
        "                blockIdx.y == 0 && threadIdx.y == 0 ||\n",
        "                threadIdx.x + threadIdx.y != magic_number)\n",
        "                return;\n",
        "                \n",
        "                int score = arr2_shared[threadIdx.x - 1] == arr1_shared[threadIdx.y - 1] ? match : miss;\n",
        "                int max = 0;\n",
        "                max = max > mat_shared[idx - 1] - gap_penalty ? max : mat_shared[idx - 1] - gap_penalty;                 // max > mat[i][j-1] - gap_penalty\n",
        "                max = max > mat_shared[idx - mat_width] - gap_penalty ? max : mat_shared[idx - mat_width] - gap_penalty; // max > mat[i-1][j] - gap_penalty\n",
        "                max = max > mat_shared[idx - mat_width - 1] + score ? max : mat_shared[idx - mat_width - 1] + score;     // max > mat[i-1][j-1] + score\n",
        "\n",
        "                mat[idx] = max; \n",
        "        }\n",
        "\"\"\")\n",
        "\n",
        "np.set_printoptions(threshold=sys.maxsize)\n",
        "\n",
        "gap_penalty = 2\n",
        "match = 3\n",
        "miss = -3\n",
        "\n",
        "arr1 = np.array(['c', 'a', 'g', 'c','c', 'a', 'g', 'c','c','c', 'a', 'r', 'c', 'c', 'c']) # height\n",
        "arr2 = np.array(['c', 'a', 'g', 'c','c', 'a', 'g', 'c','c','c', 'a', 'r', 'c', 'c', 'c', 'a', 'r']) # width\n",
        "\n",
        "mat = np.zeros([len(arr1) + 1, len(arr2) + 1], dtype=np.int32)\n",
        "\n",
        "arr1_gpu = cuda.mem_alloc(arr1.nbytes) # alociranje CUDA memorije\n",
        "arr2_gpu = cuda.mem_alloc(arr2.nbytes) # alociranje CUDA memorije\n",
        "mat_gpu = cuda.mem_alloc(mat.nbytes)   # alociranje CUDA memorije\n",
        "\n",
        "cuda.memcpy_htod(arr1_gpu, arr1)  # kopiranje Host TO Device\n",
        "cuda.memcpy_htod(arr2_gpu, arr2)  # kopiranje Host TO Device\n",
        "cuda.memcpy_htod(mat_gpu, mat)    # kopiranje Host TO Device\n",
        "\n",
        "func = mod.get_function(\"smith_waterman_shared\")\n",
        "\n",
        "height = len(arr1) + 1\n",
        "width = len(arr2) + 1\n",
        "num_of_diagonals = (height + width - 1)\n",
        "\n",
        "for k in range(num_of_diagonals):\n",
        "  func( mat_gpu, \n",
        "        np.int32(width),\n",
        "        np.int32(height),\n",
        "        np.int32(k), \n",
        "        np.int32(miss),\n",
        "        np.int32(match), \n",
        "        np.int32(gap_penalty),\n",
        "        arr1_gpu, \n",
        "        arr2_gpu,\n",
        "        shared=(mat.nbytes + arr1.nbytes + arr2.nbytes + mat.nbytes),\n",
        "        block=(32, 32, 1), \n",
        "        grid=(math.ceil(width/32), math.ceil(height/32), 1) ) \n",
        "\n",
        "cuda.memcpy_dtoh(arr1, arr1_gpu)  # kopiranje Device TO Host\n",
        "cuda.memcpy_dtoh(arr2, arr2_gpu)  # kopiranje Device TO Host\n",
        "cuda.memcpy_dtoh(mat, mat_gpu)    # kopiranje Device TO Host\n",
        "\n",
        "print(mat)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "fkTi-qrC0L4q",
        "outputId": "139adc00-62a0-4ce6-fd54-843b72d90e01"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "[[ 0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0]\n",
            " [ 0  3  1  0  0  0  0  0  0  0  0  0  0  3  1  0  0  3]\n",
            " [ 0  1  6  4  3  1  3  3  3  1  3  3  3  1  6  4  3  1]\n",
            " [ 0  0  4  9  7  5  4  6  6  4  4  6  6  4  4  9  7  5]\n",
            " [ 0  0  3  7 12 10  8  7  9  7  7  7  9  7  7  7 12 10]\n",
            " [ 0  0  1  5 10 15 13 11  9  7  5  5  7  6  5  5 10  9]\n",
            " [ 0  0  3  4  8 13 18 16 14 12 10  8  8  6  9  8  8  7]\n",
            " [ 0  0  3  6  7 11 16 21 19 17 15 13 11  9  9 12 11  9]\n",
            " [ 0  0  3  6  9  9 14 19 24 22 20 18 16 14 12 12 15 13]\n",
            " [ 0  0  1  4  7  7 12 17 22 27 25 23 21 19 17 15 13 12]\n",
            " [ 0  0  3  4  7  5 10 15 20 25 30 28 26 24 22 20 18 16]\n",
            " [ 0  0  3  6  7  5  8 13 18 23 28 33 31 29 27 25 23 21]\n",
            " [ 0  0  3  6  9  7  8 11 16 21 26 31 36 34 32 30 28 26]\n",
            " [ 0  3  1  4  7  6  6  9 14 19 24 29 34 39 37 35 33 31]\n",
            " [ 0  1  6  4  7  5  9  9 12 17 22 27 32 37 42 40 38 36]\n",
            " [ 0  0  4  9  7  5  8 12 12 15 20 25 30 35 40 45 43 41]]\n"
          ]
        },
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "<ipython-input-7-0e1f0db17f9d>:8: UserWarning: The CUDA compiler succeeded, but said the following:\n",
            "kernel.cu(17): warning: extern declaration of the entity mat_shared is treated as a static definition\n",
            "\n",
            "kernel.cu(18): warning: extern declaration of the entity arr1_shared is treated as a static definition\n",
            "\n",
            "kernel.cu(19): warning: extern declaration of the entity arr2_shared is treated as a static definition\n",
            "\n",
            "\n",
            "  mod = SourceModule(\"\"\"\n",
            "/usr/local/lib/python3.8/dist-packages/google/colab/_variable_inspector.py:27: UserWarning: device_allocation in out-of-thread context could not be cleaned up\n",
            "  globals().clear()\n",
            "/usr/local/lib/python3.8/dist-packages/google/colab/_variable_inspector.py:27: UserWarning: module in out-of-thread context could not be cleaned up\n",
            "  globals().clear()\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Cetvrta tacka"
      ],
      "metadata": {
        "id": "i_2ZwsMSDIfU"
      }
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "rdA-voXcIcMv",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "acb677c4-b8f5-47e1-c233-480a1a91e87b"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "[[  0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0\n",
            "    0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0]\n",
            " [  0   3   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0\n",
            "    0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0]\n",
            " [  0   1   6   3   3   0   3   3   3   0   3   3   3   0   3   3   3   0\n",
            "    3   3   3   0   3   3   3   0   3   3   3   0   3   3   3   0   3   3]\n",
            " [  0   0   4   9   6   0   3   6   6   0   3   6   6   0   3   6   6   0\n",
            "    3   6   6   0   3   6   6   0   3   6   6   0   3   6   6   0   3   6]\n",
            " [  0   0   3   7  12   3   3   6   9   3   3   6   9   3   3   6   9   3\n",
            "    3   6   9   3   3   6   9   3   3   6   9   3   3   6   9   3   3   6]\n",
            " [  0   0   1   5  10  15   1   4   7   6   1   4   7   6   1   4   7   6\n",
            "    1   4   7   6   1   4   7   6   1   4   7   6   1   4   7   6   1   4]\n",
            " [  0   0   3   4   8  13  18   4   7   4   9   4   7   4   9   4   7   4\n",
            "    9   4   7   4   9   4   7   4   9   4   7   4   9   4   7   4   9   4]\n",
            " [  0   0   3   6   7  11  16  21   7   4   7  12   7   4   7  12   7   4\n",
            "    7  12   7   4   7  12   7   4   7  12   7   4   7  12   7   4   7  12]\n",
            " [  0   0   3   6   9   9  14  19  24   4   7  10  15   4   7  10  15   4\n",
            "    7  10  15   4   7  10  15   4   7  10  15   4   7  10  15   4   7  10]\n",
            " [  0   0   1   4   7   7  12  17  22  27   5   8  13  12   5   8  13  12\n",
            "    5   8  13  12   5   8  13  12   5   8  13  12   5   8  13  12   5   8]\n",
            " [  0   0   3   4   7   5  10  15  20  25  30   8  11  10  15   8  11  10\n",
            "   15   8  11  10  15   8  11  10  15   8  11  10  15   8  11  10  15   8]\n",
            " [  0   0   3   6   7   4   8  13  18  23  28  33  11   8  13  18  11   8\n",
            "   13  18  11   8  13  18  11   8  13  18  11   8  13  18  11   8  13  18]\n",
            " [  0   0   3   6   9   4   7  11  16  21  26  31  36   8  11  16  21   8\n",
            "   11  16  21   8  11  16  21   8  11  16  21   8  11  16  21   8  11  16]\n",
            " [  0   0   1   4   7   6   5   9  14  19  24  29  34  39   9  14  19  18\n",
            "    9  14  19  18   9  14  19  24   9  14  19  24   9  14  19  18   9  14]\n",
            " [  0   0   3   4   7   4   9   8  12  17  22  27  32  37  42  12  17  16\n",
            "   21  12  17  16  21  12  17  22  27  12  17  22  27  12  17  16  21  12]\n",
            " [  0   0   3   6   7   4   7  12  11  15  20  25  30  35  40  45  15  14\n",
            "   19  24  15  14  19  24  15  20  25  30  15  20  25  30  15  14  19  24]\n",
            " [  0   0   3   6   9   4   7  10  15  13  18  23  28  33  38  43  48  12\n",
            "   17  22  27  12  17  22  27  18  23  28  33  18  23  28  33  12  17  22]\n",
            " [  0   0   1   4   7   6   5   8  13  12  16  21  26  31  36  41  46  51\n",
            "   15  20  25  24  15  20  25  24  21  26  31  30  21  26  31  36  15  20]\n",
            " [  0   0   3   4   7   4   9   8  11  10  15  19  24  29  34  39  44  49\n",
            "   54  18  23  22  27  18  23  22  27  24  29  28  33  24  29  34  39  18]\n",
            " [  0   0   3   6   7   4   7  12  11   8  13  18  22  27  32  37  42  47\n",
            "   52  57  21  20  25  30  21  20  25  30  27  26  31  36  27  32  37  42]\n",
            " [  0   0   3   6   9   4   7  10  15   8  11  16  21  25  30  35  40  45\n",
            "   50  55  60  18  23  28  33  18  23  28  33  24  29  34  39  30  35  40]\n",
            " [  0   0   1   4   7   6   5   8  13  12   9  14  19  23  28  33  38  43\n",
            "   48  53  58  63  21  26  31  30  21  26  31  30  27  32  37  36  33  38]\n",
            " [  0   0   3   4   7   4   9   8  11  10  15  12  17  21  26  31  36  41\n",
            "   46  51  56  61  66  24  29  28  33  24  29  28  33  30  35  34  39  36]\n",
            " [  0   0   3   6   7   4   7  12  11   8  13  18  15  19  24  29  34  39\n",
            "   44  49  54  59  64  69  27  26  31  36  27  26  31  36  33  32  37  42]\n",
            " [  0   0   3   6   9   4   7  10  15   8  11  16  21  17  22  27  32  37\n",
            "   42  47  52  57  62  67  72  24  29  34  39  24  29  34  39  30  35  40]\n",
            " [  0   0   1   4   7   6   5   8  13  12   9  14  19  24  20  25  30  35\n",
            "   40  45  50  55  60  65  70  75  27  32  37  42  27  32  37  36  33  38]\n",
            " [  0   0   3   4   7   4   9   8  11  10  15  12  17  22  27  23  28  33\n",
            "   38  43  48  53  58  63  68  73  78  30  35  40  45  30  35  34  39  36]\n",
            " [  0   0   3   6   7   4   7  12  11   8  13  18  15  20  25  30  26  31\n",
            "   36  41  46  51  56  61  66  71  76  81  33  38  43  48  33  32  37  42]\n",
            " [  0   0   3   6   9   4   7  10  15   8  11  16  21  18  23  28  33  29\n",
            "   34  39  44  49  54  59  64  69  74  79  84  36  41  46  51  30  35  40]\n",
            " [  0   0   1   4   7   6   5   8  13  12   9  14  19  24  21  26  31  30\n",
            "   32  37  42  47  52  57  62  67  72  77  82  87  39  44  49  48  33  38]\n",
            " [  0   0   3   4   7   4   9   8  11  10  15  12  17  22  27  24  29  28\n",
            "   33  35  40  45  50  55  60  65  70  75  80  85  90  42  47  46  51  36]\n",
            " [  0   0   3   6   7   4   7  12  11   8  13  18  15  20  25  30  27  26\n",
            "   31  36  38  43  48  53  58  63  68  73  78  83  88  93  45  44  49  54]\n",
            " [  0   0   3   6   9   4   7  10  15   8  11  16  21  18  23  28  33  24\n",
            "   29  34  39  41  46  51  56  61  66  71  76  81  86  91  96  42  47  52]\n",
            " [  0   0   1   4   7   6   5   8  13  12   9  14  19  18  21  26  31  36\n",
            "   27  32  37  39  44  49  54  59  64  69  74  79  84  89  94  99  45  50]\n",
            " [  0   0   3   4   7   4   9   8  11  10  15  12  17  16  21  24  29  34\n",
            "   39  30  35  37  42  47  52  57  62  67  72  77  82  87  92  97 102  48]\n",
            " [  0   0   3   6   7   4   7  12  11   8  13  18  15  14  19  24  27  32\n",
            "   37  42  33  35  40  45  50  55  60  65  70  75  80  85  90  95 100 105]]\n"
          ]
        }
      ],
      "source": [
        "import numpy as np\n",
        "import sys\n",
        "import math\n",
        "import pycuda.driver as cuda\n",
        "import pycuda.autoinit\n",
        "from pycuda.compiler import SourceModule\n",
        "\n",
        "mod = SourceModule(\"\"\"\n",
        "      __global__ void smith_waterman_advanced(  int *mat, \n",
        "                                        int mat_width,\n",
        "                                        int mat_height,\n",
        "                                        int magic_number, \n",
        "                                        int miss, \n",
        "                                        int match, \n",
        "                                        int gap_penalty,\n",
        "                                        char arr1[], \n",
        "                                        char arr2[] ) {\n",
        "\n",
        "          __shared__ int sub_mat[32 + 1][32 + 1];\n",
        "          __shared__ char sub_arr1[32];\n",
        "          __shared__ char sub_arr2[32];\n",
        "          \n",
        "          int y = threadIdx.y + blockDim.y*blockIdx.y;\n",
        "          int x = threadIdx.x + blockDim.x*blockIdx.x;\n",
        "\n",
        "          if( x >= mat_width ||\n",
        "              y >= mat_height )\n",
        "              return;\n",
        "\n",
        "          sub_arr1[threadIdx.y] = arr1[y];\n",
        "          sub_arr2[threadIdx.x] = arr2[x];\n",
        "\n",
        "          sub_mat[threadIdx.y + 1][threadIdx.x + 1] = mat[x + y*mat_width];\n",
        "          if(threadIdx.x == 0){\n",
        "            sub_mat[threadIdx.y + 1][threadIdx.x] = mat[x + y*mat_width - 1];\n",
        "            sub_mat[threadIdx.y][threadIdx.x] = mat[x + y*mat_width - mat_width - 1];\n",
        "          }\n",
        "          if(threadIdx.y == 0){\n",
        "            sub_mat[threadIdx.y][threadIdx.x + 1] = mat[x + y*mat_width - mat_width];\n",
        "            sub_mat[threadIdx.y][threadIdx.x] = mat[x + y*mat_width - mat_width - 1];\n",
        "          }\n",
        "\n",
        "          __syncthreads();\n",
        "          \n",
        "          if( blockIdx.x == 0 && threadIdx.x == 0 ||\n",
        "              blockIdx.y == 0 && threadIdx.y == 0 ||\n",
        "              x + y != magic_number )\n",
        "            return;\n",
        " \n",
        "            int score = sub_arr2[threadIdx.x - 1] == sub_arr1[threadIdx.y - 1] ? match : miss;\n",
        "            int max = 0;\n",
        "            max = max > sub_mat[threadIdx.y + 1][threadIdx.x + 1] - gap_penalty ? max : sub_mat[threadIdx.y + 1][threadIdx.x] - gap_penalty;      // max > sub_mat[i][j-1] - gap_penalty\n",
        "            max = max > sub_mat[threadIdx.y][threadIdx.x + 1] - gap_penalty ? max : sub_mat[threadIdx.y][threadIdx.x + 1] - gap_penalty;          // max > sub_mat[i-1][j] - gap_penalty\n",
        "            max = max > sub_mat[threadIdx.y][threadIdx.x] + score ? max : sub_mat[threadIdx.y][threadIdx.x] + score;                              // max > sub_mat[i-1][j-1] + score\n",
        "\n",
        "            mat[x + y*mat_width] = max;\n",
        "      }\n",
        "\"\"\")\n",
        "\n",
        "np.set_printoptions(threshold=sys.maxsize)\n",
        "\n",
        "gap_penalty = 2\n",
        "match = 3\n",
        "miss = -3\n",
        "\n",
        "aarr1 = np.array(['x' , 'y', 'p','c', 'a', 'g', 'c','c', 'a', 'g', 'c','c','c', 'a', 'r', 'q', 'x' , 'y', 'p','c', 'a', 'g', 'c','c', 'a', 'g', 'c','c','c', 'a', 'r', 'q', 'x' , 'y', 'p']) # height\n",
        "aarr2 = np.array(['x' , 'y', 'p','c', 'a', 'g', 'c','c', 'a', 'g', 'c','c','c', 'a', 'r', 'q', 'x' , 'y', 'p','c', 'a', 'g', 'c','c', 'a', 'g', 'c','c','c', 'a', 'r', 'q', 'x' , 'y', 'p']) # width\n",
        "\n",
        "mmat = np.zeros([len(aarr1) + 1, len(aarr2) + 1], dtype=np.int32)\n",
        "\n",
        "aarr1_gpu = cuda.mem_alloc(aarr1.nbytes) # alociranje CUDA memorije\n",
        "aarr2_gpu = cuda.mem_alloc(aarr2.nbytes) # alociranje CUDA memorije\n",
        "mmat_gpu = cuda.mem_alloc(mmat.nbytes)   # alociranje CUDA memorije\n",
        "\n",
        "cuda.memcpy_htod(aarr1_gpu, aarr1)  # kopiranje Host TO Device\n",
        "cuda.memcpy_htod(aarr2_gpu, aarr2)  # kopiranje Host TO Device\n",
        "cuda.memcpy_htod(mmat_gpu, mmat)    # kopiranje Host TO Device\n",
        "\n",
        "ffunc = mod.get_function(\"smith_waterman_advanced\")\n",
        "\n",
        "hheight = len(aarr1) + 1\n",
        "wwidth = len(aarr2) + 1\n",
        "nnum_of_diagonals = (hheight + wwidth - 1)\n",
        "\n",
        "for k in range(nnum_of_diagonals):\n",
        "  ffunc( mmat_gpu, \n",
        "        np.int32(wwidth),\n",
        "        np.int32(hheight),\n",
        "        np.int32(k), \n",
        "        np.int32(miss),\n",
        "        np.int32(match), \n",
        "        np.int32(gap_penalty),\n",
        "        aarr1_gpu, \n",
        "        aarr2_gpu, \n",
        "        block=(32, 32, 1), \n",
        "        grid=(math.ceil(wwidth/32), math.ceil(hheight/32), 1) ) \n",
        "\n",
        "cuda.memcpy_dtoh(aarr1, aarr1_gpu)  # kopiranje Device TO Host\n",
        "cuda.memcpy_dtoh(aarr2, aarr2_gpu)  # kopiranje Device TO Host\n",
        "cuda.memcpy_dtoh(mmat, mmat_gpu)    # kopiranje Device TO Host\n",
        "\n",
        "print(mmat) \n"
      ]
    },
    {
      "cell_type": "markdown",
      "source": [],
      "metadata": {
        "id": "CT3nDTRkp94C"
      }
    },
    {
      "cell_type": "markdown",
      "source": [],
      "metadata": {
        "id": "n-fME9qLp8n-"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "import numpy as np\n",
        "\n",
        "def smith_waterman(arr1, arr2, miss, match, gap_penalty):\n",
        "  mat = np.zeros([len(arr1) + 1, len(arr2) + 1], dtype=np.int32)\n",
        "\n",
        "  for i in range(len(arr1)):\n",
        "    for j in range(len(arr2)):\n",
        "      score = match if arr1[i] == arr2[j] else miss\n",
        "      max = mat[i + 1][j] - gap_penalty if mat[i + 1][j] - gap_penalty > 0 else 0\n",
        "      max = mat[i][j + 1] - gap_penalty if mat[i][j + 1] - gap_penalty > max else max\n",
        "      max = mat[i][j] + score if mat[i][j] + score > max else max\n",
        "      mat[i + 1][j + 1] = max\n",
        "  \n",
        "  return mat"
      ],
      "metadata": {
        "id": "Lbsi-ErW8X7D"
      },
      "execution_count": null,
      "outputs": []
    }
  ],
  "metadata": {
    "accelerator": "GPU",
    "colab": {
      "provenance": [],
      "include_colab_link": true
    },
    "gpuClass": "standard",
    "kernelspec": {
      "display_name": "Python 3",
      "name": "python3"
    },
    "language_info": {
      "name": "python"
    }
  },
  "nbformat": 4,
  "nbformat_minor": 0
}