{
 "metadata": {
  "name": "",
  "signature": "sha256:c3a802f3d9358b3877e4556b6a1145c0ceec9752507b1a6cd5a7fb5ebbe09146"
 },
 "nbformat": 3,
 "nbformat_minor": 0,
 "worksheets": [
  {
   "cells": [
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "n=500\n",
      "mat=np.random.rand(n,n)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 1
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "mat.shape"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "metadata": {},
       "output_type": "pyout",
       "prompt_number": 2,
       "text": [
        "(500, 500)"
       ]
      }
     ],
     "prompt_number": 2
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "# Python, Numpy"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "def matmul(mat1,mat2):\n",
      "    out=np.zeros((mat1.shape[0],mat2.shape[1]))\n",
      "    for i in xrange(mat1.shape[0]):\n",
      "        for j in xrange(mat2.shape[1]):\n",
      "            for k in xrange(mat1.shape[1]):\n",
      "                out[i,j]+=mat1[i,k]*mat2[k,j]\n",
      "    return out\n",
      "\n",
      "def matmul_dot(mat1,mat2):\n",
      "    return np.dot(mat1,mat2)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 3
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "%timeit -r 3 -n 1 matmul(mat,mat)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "1 loops, best of 3: 2min 19s per loop\n"
       ]
      }
     ],
     "prompt_number": 4
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "%timeit -r 3 -n 1 matmul_dot(mat,mat)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "1 loops, best of 3: 6.17 ms per loop\n"
       ]
      }
     ],
     "prompt_number": 6
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "# Numba"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "import numba\n",
      "\n",
      "matmul_nb=numba.autojit(matmul)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 7
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "%timeit -r 3 -n 1 matmul_nb(mat,mat)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "1 loops, best of 3: 329 ms per loop\n"
       ]
      }
     ],
     "prompt_number": 8
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "# Cython"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "%load_ext cythonmagic"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 10
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "%%cython\n",
      "import cython\n",
      "import numpy as np\n",
      "cimport numpy as np\n",
      "@cython.boundscheck(False)\n",
      "@cython.wraparound(False)\n",
      "@cython.nonecheck(False)\n",
      "def matmul_cy(np.ndarray[double,ndim=2] mat1,np.ndarray[double,ndim=2] mat2):\n",
      "    cdef np.ndarray[double,ndim=2] out=np.zeros((mat1.shape[0],mat2.shape[1]))\n",
      "    cdef int i,j,k\n",
      "    for i in xrange(mat1.shape[0]):\n",
      "        for j in xrange(mat2.shape[1]):\n",
      "            for k in xrange(mat1.shape[1]):\n",
      "                out[i,j]+=mat1[i,k]*mat2[k,j]\n",
      "    return out"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 11
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "%timeit -r 3 -n 1 matmul_cy(mat,mat)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "1 loops, best of 3: 334 ms per loop\n"
       ]
      }
     ],
     "prompt_number": 12
    },
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "# Fortran"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "%%file matmul.f90\n",
      "subroutine matmul_s(n1,n2,n3,mat1,mat2,mat3)\n",
      "implicit none\n",
      "integer,intent(in):: n1,n2,n3\n",
      "real(kind=8),intent(in):: mat1(n1,n2),mat2(n2,n3)\n",
      "real(kind=8),intent(out):: mat3(n1,n3)\n",
      "integer i,j,k\n",
      "do i=1,n1\n",
      "do j=1,n3\n",
      "    mat3(i,j)=0.\n",
      "    do k=1,n2\n",
      "        mat3(i,j)=mat3(i,j)+mat1(i,k)*mat2(k,j)\n",
      "    enddo\n",
      "enddo\n",
      "enddo\n",
      "end subroutine\n",
      "\n",
      "subroutine matmul_fi(n1,n2,n3,mat1,mat2,mat3)\n",
      "implicit none\n",
      "integer,intent(in):: n1,n2,n3\n",
      "real(kind=8),intent(in):: mat1(n1,n2),mat2(n2,n3)\n",
      "real(kind=8),intent(out):: mat3(n1,n3)\n",
      "mat3=matmul(mat1,mat2)\n",
      "end subroutine"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "Overwriting matmul.f90\n"
       ]
      }
     ],
     "prompt_number": 14
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "!f2py -c -m matmul_fortran matmul.f90 --f90exec=/opt/local/bin/gfortran-mp-4.9 > log.txt"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 15
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "import matmul_fortran\n",
      "print matmul_fortran.__doc__"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "This module 'matmul_fortran' is auto-generated with f2py (version:2).\n",
        "Functions:\n",
        "  mat3 = matmul_s(mat1,mat2,n1=shape(mat1,0),n2=shape(mat1,1),n3=shape(mat2,1))\n",
        "  mat3 = matmul_f(mat1,mat2,n1=shape(mat1,0),n2=shape(mat1,1),n3=shape(mat2,1))\n",
        "  mat3 = matmul_fi(mat1,mat2,n1=shape(mat1,0),n2=shape(mat1,1),n3=shape(mat2,1))\n",
        ".\n"
       ]
      }
     ],
     "prompt_number": 16
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "%timeit -r 3 -n 1 matmul_fortran.matmul_s(mat,mat)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "1 loops, best of 3: 108 ms per loop\n"
       ]
      }
     ],
     "prompt_number": 17
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "%timeit -r 3 -n 1 matmul_fortran.matmul_fi(mat,mat)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "1 loops, best of 3: 70.7 ms per loop\n"
       ]
      }
     ],
     "prompt_number": 19
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 21
    }
   ],
   "metadata": {}
  }
 ]
}