code/Case Study A.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# OpenCL Heterogeneous Mapping\n",
    "\n",
    "This notebook contains the experiments for Case Study A - using deep learning to predict the optimal device mapping for OpenCL kernels, without hand engineered features.\n",
    "\n",
    "## 1. Runtime Data\n",
    "\n",
    "We use the runtime data from our CGO'17 paper [1]. The code to reproduce those experiments can be found in this, artifact evaluated [GitHub repository](https://github.com/ChrisCummins/paper-synthesizing-benchmarks).\n",
    "\n",
    "> [1] Cummins, C., Petoumenos, P., Zang, W., & Leather, H. (2017). [Synthesizing Benchmarks for Predictive Modeling](http://chriscummins.cc/pub/2017-cgo.pdf). In CGO. IEEE.\n",
    "\n",
    "The runtime data comes from two plaforms. AMD:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Unnamed: 0</th>\n",
       "      <th>benchmark</th>\n",
       "      <th>dataset</th>\n",
       "      <th>comp</th>\n",
       "      <th>rational</th>\n",
       "      <th>mem</th>\n",
       "      <th>localmem</th>\n",
       "      <th>coalesced</th>\n",
       "      <th>atomic</th>\n",
       "      <th>transfer</th>\n",
       "      <th>wgsize</th>\n",
       "      <th>oracle</th>\n",
       "      <th>runtime_cpu</th>\n",
       "      <th>runtime_gpu</th>\n",
       "      <th>src</th>\n",
       "      <th>seq</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>amd-app-sdk-3.0-BinomialOption-binomial_options</td>\n",
       "      <td>default</td>\n",
       "      <td>98</td>\n",
       "      <td>8</td>\n",
       "      <td>13</td>\n",
       "      <td>11</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>2048</td>\n",
       "      <td>255</td>\n",
       "      <td>GPU</td>\n",
       "      <td>3.291073</td>\n",
       "      <td>1.443983</td>\n",
       "      <td>__kernel void A(int a, const __global float4* ...</td>\n",
       "      <td>[129 129 129 ...,   0 127   0]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>amd-app-sdk-3.0-BitonicSort-bitonicSort</td>\n",
       "      <td>default</td>\n",
       "      <td>12</td>\n",
       "      <td>2</td>\n",
       "      <td>6</td>\n",
       "      <td>0</td>\n",
       "      <td>6</td>\n",
       "      <td>0</td>\n",
       "      <td>131072</td>\n",
       "      <td>256</td>\n",
       "      <td>CPU</td>\n",
       "      <td>0.121940</td>\n",
       "      <td>0.279521</td>\n",
       "      <td>__kernel void A(__global uint* a, const uint b...</td>\n",
       "      <td>[129 129 129 ...,   0 127   0]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>678</th>\n",
       "      <td>678</td>\n",
       "      <td>shoc-1.1.5-Stencil2D-StencilKernel</td>\n",
       "      <td>default</td>\n",
       "      <td>144</td>\n",
       "      <td>7</td>\n",
       "      <td>16</td>\n",
       "      <td>12</td>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "      <td>71647488</td>\n",
       "      <td>256</td>\n",
       "      <td>GPU</td>\n",
       "      <td>17.866730</td>\n",
       "      <td>16.666380</td>\n",
       "      <td>__kernel void A(__global float* a, __global fl...</td>\n",
       "      <td>[  1  96   1 ..., 127   0 127]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>679</th>\n",
       "      <td>679</td>\n",
       "      <td>shoc-1.1.5-Triad-Triad</td>\n",
       "      <td>default</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "      <td>117440512</td>\n",
       "      <td>128</td>\n",
       "      <td>CPU</td>\n",
       "      <td>876.457180</td>\n",
       "      <td>968.494606</td>\n",
       "      <td>__kernel void A(__global const float* a, __glo...</td>\n",
       "      <td>[129 129 129 ...,   0 127   0]</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>680 rows × 16 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "     Unnamed: 0                                        benchmark  dataset  \\\n",
       "0             0  amd-app-sdk-3.0-BinomialOption-binomial_options  default   \n",
       "1             1          amd-app-sdk-3.0-BitonicSort-bitonicSort  default   \n",
       "..          ...                                              ...      ...   \n",
       "678         678               shoc-1.1.5-Stencil2D-StencilKernel  default   \n",
       "679         679                           shoc-1.1.5-Triad-Triad  default   \n",
       "\n",
       "     comp  rational  mem  localmem  coalesced  atomic   transfer  wgsize  \\\n",
       "0      98         8   13        11          2       0       2048     255   \n",
       "1      12         2    6         0          6       0     131072     256   \n",
       "..    ...       ...  ...       ...        ...     ...        ...     ...   \n",
       "678   144         7   16        12          3       0   71647488     256   \n",
       "679     2         0    3         0          3       0  117440512     128   \n",
       "\n",
       "    oracle  runtime_cpu  runtime_gpu  \\\n",
       "0      GPU     3.291073     1.443983   \n",
       "1      CPU     0.121940     0.279521   \n",
       "..     ...          ...          ...   \n",
       "678    GPU    17.866730    16.666380   \n",
       "679    CPU   876.457180   968.494606   \n",
       "\n",
       "                                                   src  \\\n",
       "0    __kernel void A(int a, const __global float4* ...   \n",
       "1    __kernel void A(__global uint* a, const uint b...   \n",
       "..                                                 ...   \n",
       "678  __kernel void A(__global float* a, __global fl...   \n",
       "679  __kernel void A(__global const float* a, __glo...   \n",
       "\n",
       "                                seq  \n",
       "0    [129 129 129 ...,   0 127   0]  \n",
       "1    [129 129 129 ...,   0 127   0]  \n",
       "..                              ...  \n",
       "678  [  1  96   1 ..., 127   0 127]  \n",
       "679  [129 129 129 ...,   0 127   0]  \n",
       "\n",
       "[680 rows x 16 columns]"
      ]
     },
     "execution_count": 1,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import pandas as pd\n",
    "pd.set_option('display.max_rows', 5)\n",
    "\n",
    "pd.read_csv(\"../data/case-study-a/cgo17-amd.csv\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "and NVIDIA:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Unnamed: 0</th>\n",
       "      <th>benchmark</th>\n",
       "      <th>dataset</th>\n",
       "      <th>comp</th>\n",
       "      <th>rational</th>\n",
       "      <th>mem</th>\n",
       "      <th>localmem</th>\n",
       "      <th>coalesced</th>\n",
       "      <th>atomic</th>\n",
       "      <th>transfer</th>\n",
       "      <th>wgsize</th>\n",
       "      <th>oracle</th>\n",
       "      <th>runtime_cpu</th>\n",
       "      <th>runtime_gpu</th>\n",
       "      <th>src</th>\n",
       "      <th>seq</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>amd-app-sdk-3.0-BinomialOption-binomial_options</td>\n",
       "      <td>default</td>\n",
       "      <td>98</td>\n",
       "      <td>8</td>\n",
       "      <td>13</td>\n",
       "      <td>11</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>2048</td>\n",
       "      <td>255</td>\n",
       "      <td>GPU</td>\n",
       "      <td>3.291073</td>\n",
       "      <td>0.152038</td>\n",
       "      <td>__kernel void A(int a, const __global float4* ...</td>\n",
       "      <td>[129 129 129 ...,   0 127   0]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>amd-app-sdk-3.0-BitonicSort-bitonicSort</td>\n",
       "      <td>default</td>\n",
       "      <td>12</td>\n",
       "      <td>2</td>\n",
       "      <td>6</td>\n",
       "      <td>0</td>\n",
       "      <td>6</td>\n",
       "      <td>0</td>\n",
       "      <td>131072</td>\n",
       "      <td>256</td>\n",
       "      <td>GPU</td>\n",
       "      <td>0.121940</td>\n",
       "      <td>0.057834</td>\n",
       "      <td>__kernel void A(__global uint* a, const uint b...</td>\n",
       "      <td>[129 129 129 ...,   0 127   0]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>678</th>\n",
       "      <td>678</td>\n",
       "      <td>shoc-1.1.5-Stencil2D-StencilKernel</td>\n",
       "      <td>default</td>\n",
       "      <td>144</td>\n",
       "      <td>7</td>\n",
       "      <td>16</td>\n",
       "      <td>12</td>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "      <td>71647488</td>\n",
       "      <td>256</td>\n",
       "      <td>CPU</td>\n",
       "      <td>17.866730</td>\n",
       "      <td>48.047807</td>\n",
       "      <td>__kernel void A(__global float* a, __global fl...</td>\n",
       "      <td>[  1  96   1 ..., 127   0 127]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>679</th>\n",
       "      <td>679</td>\n",
       "      <td>shoc-1.1.5-Triad-Triad</td>\n",
       "      <td>default</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "      <td>117440512</td>\n",
       "      <td>128</td>\n",
       "      <td>CPU</td>\n",
       "      <td>876.457180</td>\n",
       "      <td>3184.516221</td>\n",
       "      <td>__kernel void A(__global const float* a, __glo...</td>\n",
       "      <td>[129 129 129 ...,   0 127   0]</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>680 rows × 16 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "     Unnamed: 0                                        benchmark  dataset  \\\n",
       "0             0  amd-app-sdk-3.0-BinomialOption-binomial_options  default   \n",
       "1             1          amd-app-sdk-3.0-BitonicSort-bitonicSort  default   \n",
       "..          ...                                              ...      ...   \n",
       "678         678               shoc-1.1.5-Stencil2D-StencilKernel  default   \n",
       "679         679                           shoc-1.1.5-Triad-Triad  default   \n",
       "\n",
       "     comp  rational  mem  localmem  coalesced  atomic   transfer  wgsize  \\\n",
       "0      98         8   13        11          2       0       2048     255   \n",
       "1      12         2    6         0          6       0     131072     256   \n",
       "..    ...       ...  ...       ...        ...     ...        ...     ...   \n",
       "678   144         7   16        12          3       0   71647488     256   \n",
       "679     2         0    3         0          3       0  117440512     128   \n",
       "\n",
       "    oracle  runtime_cpu  runtime_gpu  \\\n",
       "0      GPU     3.291073     0.152038   \n",
       "1      GPU     0.121940     0.057834   \n",
       "..     ...          ...          ...   \n",
       "678    CPU    17.866730    48.047807   \n",
       "679    CPU   876.457180  3184.516221   \n",
       "\n",
       "                                                   src  \\\n",
       "0    __kernel void A(int a, const __global float4* ...   \n",
       "1    __kernel void A(__global uint* a, const uint b...   \n",
       "..                                                 ...   \n",
       "678  __kernel void A(__global float* a, __global fl...   \n",
       "679  __kernel void A(__global const float* a, __glo...   \n",
       "\n",
       "                                seq  \n",
       "0    [129 129 129 ...,   0 127   0]  \n",
       "1    [129 129 129 ...,   0 127   0]  \n",
       "..                              ...  \n",
       "678  [  1  96   1 ..., 127   0 127]  \n",
       "679  [129 129 129 ...,   0 127   0]  \n",
       "\n",
       "[680 rows x 16 columns]"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pd.read_csv(\"../data/case-study-a/cgo17-nvidia.csv\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 2. Predictive Models\n",
    "\n",
    "We define a base class for implementing predictive models for heterogeneous mapping:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "from clgen import atomizer as clgen\n",
    "\n",
    "class HeterogemeousMappingModel(object):\n",
    "    \"\"\"\n",
    "    A model for predicting OpenCL heterogeneous device mappings.\n",
    "    \n",
    "    Attributes\n",
    "    ----------\n",
    "    __name__ : str\n",
    "        Model name\n",
    "    __basename__ : str\n",
    "        Shortened name, used for files\n",
    "    \"\"\"\n",
    "    __name__ = None\n",
    "    __basename__ = None\n",
    "    \n",
    "    def init(self, seed: int) -> None:\n",
    "        \"\"\"\n",
    "        Initialize the model.\n",
    "        \n",
    "        Do whatever is required to setup a new heterogeneous model here.\n",
    "        This method is called prior to training and predicting.\n",
    "        This method may be omitted if no initial setup is required.\n",
    "        \n",
    "        Parameters\n",
    "        ----------\n",
    "        seed : int\n",
    "            The seed value used to reproducible results. May be 'None',\n",
    "            indicating that no seed is to be used.\n",
    "        \"\"\"\n",
    "        pass\n",
    "    \n",
    "    def save(self, outpath: str) -> None:\n",
    "        \"\"\"\n",
    "        Save model state.\n",
    "        \n",
    "        This must capture all of the relevant state of the model. It is up\n",
    "        to implementing classes to determine how best to save the model.\n",
    "        \n",
    "        Parameters\n",
    "        ----------\n",
    "        outpath : str\n",
    "            The path to save the model state to.\n",
    "        \"\"\"\n",
    "        raise NotImplementedError\n",
    "    \n",
    "    def restore(self, inpath: str) -> None:\n",
    "        \"\"\"\n",
    "        Load a trained model from file.\n",
    "        \n",
    "        This is called in place of init() if a saved model file exists. It\n",
    "        must restore all of the required model state.\n",
    "    \n",
    "        Parameters\n",
    "        ----------\n",
    "        inpath : str\n",
    "            The path to load the model from. This is the same path as\n",
    "            was passed to save() to create the file.\n",
    "        \"\"\"\n",
    "        raise NotImplementedError\n",
    "        \n",
    "    def train(self, df: pd.DataFrame, features: np.array, sequences: np.array,\n",
    "              y: np.array, y_1hot: np.array, verbose: bool=False) -> None:\n",
    "        \"\"\"\n",
    "        Train a model.\n",
    "        \n",
    "        Parameters\n",
    "        ----------\n",
    "        df : pd.DataFrame\n",
    "            The platform dataframe.\n",
    "        \n",
    "        features : np.array\n",
    "            An array of feature vectors of shape (n,4).\n",
    "        \n",
    "        sequences : np.array\n",
    "            An array of encoded source code sequences of shape (n,seq_length).\n",
    "\n",
    "        y : np.array\n",
    "            An array of optimal device mappings of shape (n,1).\n",
    "        \n",
    "        y_1hot : np.array\n",
    "            An array of optimal device mappings of shape (n,2), in 1-hot encoding.\n",
    "            \n",
    "        verbose: bool, optional\n",
    "            Whether to print verbose status messages during training.\n",
    "        \"\"\"\n",
    "        raise NotImplementedError\n",
    "\n",
    "    def predict(self, features: np.array, sequences: np.array, y: np.array,\n",
    "                y_1hot: np.array, verbose: bool=False) -> np.array:\n",
    "        \"\"\"\n",
    "        Make predictions for programs.\n",
    "        \n",
    "        Parameters\n",
    "        ----------\n",
    "        features : np.array\n",
    "            An array of feature vectors of shape (n,4).\n",
    "        \n",
    "        sequences : np.array\n",
    "            An array of encoded source code sequences of shape (n,seq_length).\n",
    "\n",
    "        y : np.array\n",
    "            An array of optimal device mappings of shape (n,1).\n",
    "        \n",
    "        y_1hot : np.array\n",
    "            An array of optimal device mappings of shape (n,2), in 1-hot encoding.\n",
    "            \n",
    "        verbose: bool, optional\n",
    "            Whether to print verbose status messages.\n",
    "            \n",
    "        Returns\n",
    "        -------\n",
    "        np.array\n",
    "            Predicted 'y' values (optimal device mappings) with shape (n,1).\n",
    "        \"\"\"\n",
    "        raise NotImplementedError"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We also define some utility code which we will use later:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from typing import List\n",
    "\n",
    "def platform2str(platform: str) -> str:\n",
    "    \"\"\" get full platform name \"\"\"\n",
    "    if platform == \"amd\":\n",
    "        return \"AMD Tahiti 7970\"\n",
    "    elif platform == \"nvidia\":\n",
    "        return \"NVIDIA GTX 970\"\n",
    "    else:\n",
    "        raise LookupException\n",
    "        \n",
    "def escape_suite_name(g: str) -> str:\n",
    "    \"\"\" format benchmark suite name for display \"\"\"\n",
    "    c = g.split('-')\n",
    "    if c[0] == \"amd\" or c[0] == \"nvidia\":\n",
    "        return c[0].upper() + \" SDK\"\n",
    "    if c[0] == \"npb\" or c[0] == \"shoc\":\n",
    "        return c[0].upper()\n",
    "    elif c[0] == \"parboil\" or c[0] == \"polybench\" or c[0] == \"rodinia\":\n",
    "        return c[0].capitalize()\n",
    "    else:\n",
    "        raise LookupError\n",
    "\n",
    "\n",
    "def escape_benchmark_name(g: str) -> str:\n",
    "    \"\"\"escape benchmark name for display\"\"\"\n",
    "    c = g.split('-')\n",
    "    return escape_suite_name(c[0]).split()[0] + \".\" + c[-2]\n",
    "\n",
    "    \n",
    "def grewe_features(df: pd.DataFrame) -> np.array:\n",
    "    \"\"\" extract Grewe et al. feature vector from runtime data \"\"\"\n",
    "    return np.array([\n",
    "            (df[\"transfer\"].values / (df[\"comp\"].values + df[\"mem\"].values)),  # F1\n",
    "            (df[\"coalesced\"].values / df[\"mem\"].values),  # F2\n",
    "            ((df[\"localmem\"].values / df[\"mem\"].values) * df[\"wgsize\"].values),  # F3\n",
    "            (df[\"comp\"].values / df[\"mem\"].values),  # F4\n",
    "        ]).T\n",
    "\n",
    "def auxiliary_inputs(df: pd.DataFrame) -> np.array:\n",
    "    \"\"\" get dsize and wgsize auxiliary inputs \"\"\"\n",
    "    return np.array([\n",
    "        df[\"transfer\"].values,\n",
    "        df[\"wgsize\"].values,\n",
    "    ]).T\n",
    "\n",
    "def encode_1hot(y: np.array) -> np.array:\n",
    "    \"\"\" 1-hot encode labels \"\"\"\n",
    "    labels = np.vstack([np.expand_dims(x, axis=0) for x in y])\n",
    "    l2 = [x[0] for x in labels]\n",
    "    l1 = [not x for x in l2]\n",
    "    return np.array(list(zip(l1, l2)), dtype=np.int32)\n",
    "\n",
    "def encode_srcs(srcs: List[str]) -> np.array:\n",
    "    \"\"\" encode and pad source code for learning \"\"\"\n",
    "    from keras.preprocessing.sequence import pad_sequences\n",
    "    \n",
    "    seqs = [atomizer.atomize(src) for src in srcs]\n",
    "    pad_val = atomizer.vocab_size\n",
    "    encoded = np.array(pad_sequences(seqs, maxlen=1024, value=pad_val))\n",
    "    return np.vstack([np.expand_dims(x, axis=0) for x in encoded])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 3. Experimental Methodology\n",
    "\n",
    "Random seed for reproducible results:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "seed = 204"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The source encoder (see the *'Language Model.ipynb'* notebook for details):"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "GreedyAtomizer[128 tokens]"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "srcs = '\\n'.join(pd.read_csv(\"../data/case-study-a/cgo17-amd.csv\")['src'].values)\n",
    "atomizer = clgen.GreedyAtomizer.from_text(srcs)\n",
    "atomizer"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Our method to evaluate a predictive model using 10-fold cross-validation:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from labm8 import fs\n",
    "\n",
    "def evaluate(model: HeterogemeousMappingModel) -> pd.DataFrame:\n",
    "    \"\"\"\n",
    "    Evaluate a model.\n",
    "    \n",
    "    Performs 10-fold cross-validation of the model's effectiveness at predicting\n",
    "    OpenCL device mappings. Results are cached.\n",
    "    \n",
    "    Parameters\n",
    "    ----------\n",
    "    model : HeterogeneousMappingModel\n",
    "        The predictive model to evaluate.\n",
    "        \n",
    "    Returns\n",
    "    -------\n",
    "    pd.Dataframe\n",
    "        Evaluation results.\n",
    "    \"\"\"\n",
    "    from sklearn.model_selection import StratifiedKFold\n",
    "    from progressbar import ProgressBar\n",
    "    \n",
    "    progressbar = [0, ProgressBar(max_value=10*2)]\n",
    "\n",
    "    data = []\n",
    "    for i, platform in enumerate([\"amd\", \"nvidia\"]):\n",
    "        platform_name = platform2str(platform)\n",
    "\n",
    "        # load runtime data\n",
    "        df = pd.read_csv(f\"../data/case-study-a/cgo17-{platform}.csv\")\n",
    "        \n",
    "        sequences = None  # defer sequence encoding until needed (it's expensive)\n",
    "        \n",
    "        # values used for training & predictions\n",
    "        features = grewe_features(df)\n",
    "        aux_in = auxiliary_inputs(df)\n",
    "        \n",
    "        # optimal mappings\n",
    "        y = np.array([1 if x == \"GPU\" else 0 for x in df[\"oracle\"].values])\n",
    "        y_1hot = encode_1hot(y)\n",
    "        \n",
    "        # 10-fold cross-validation\n",
    "        kf = StratifiedKFold(n_splits=10, shuffle=True, random_state=seed)\n",
    "        for j, (train_index, test_index) in enumerate(kf.split(features, y)):\n",
    "            model_path = f\"../data/case-study-a/models/{model.__basename__}-{platform}-{j}.model\"\n",
    "            predictions_path = f\"../data/case-study-a/predictions/{model.__basename__}-{platform}-{j}.result\"\n",
    "            \n",
    "            if fs.exists(predictions_path):\n",
    "                # load result from cache\n",
    "                with open(predictions_path, 'rb') as infile:\n",
    "                    p = pickle.load(infile)\n",
    "            else:\n",
    "                if sequences is None:  # encode source codes if needed\n",
    "                    sequences = encode_srcs(df[\"src\"].values)\n",
    "\n",
    "                if fs.exists(model_path):\n",
    "                    # restore trained model from cache\n",
    "                    model.restore(model_path)\n",
    "                else:\n",
    "                    # train and cache a model\n",
    "                    model.init(seed=seed)                   \n",
    "                    model.train(df=df,\n",
    "                                features=features[train_index],\n",
    "                                aux_in=aux_in[train_index],\n",
    "                                sequences=sequences[train_index],\n",
    "                                y=y[train_index],\n",
    "                                y_1hot=y_1hot[train_index],\n",
    "                                verbose=False)\n",
    "                    fs.mkdir(fs.dirname(model_path))\n",
    "                    model.save(model_path)\n",
    "\n",
    "                # test model\n",
    "                p = model.predict(\n",
    "                    features=features[test_index],\n",
    "                    aux_in=aux_in[test_index],\n",
    "                    sequences=sequences[test_index],\n",
    "                    y=y[test_index],\n",
    "                    y_1hot=y_1hot[test_index],\n",
    "                    verbose=False)\n",
    "    \n",
    "                # cache results\n",
    "                fs.mkdir(fs.dirname(predictions_path))\n",
    "                with open(predictions_path, 'wb') as outfile:\n",
    "                    pickle.dump(p, outfile)\n",
    "\n",
    "            # benchmarks\n",
    "            benchmarks = df['benchmark'].values[test_index]\n",
    "            # oracle device mappings\n",
    "            o = y[test_index]\n",
    "            # whether predictions were correct or not\n",
    "            correct = p == o\n",
    "            # runtimes of baseline mapping (CPU on AMD, GPU on NVIDIA)\n",
    "            zero_r_dev = \"runtime_cpu\" if platform == \"amd\" else \"runtime_gpu\"\n",
    "            zer_r_runtimes = df[zero_r_dev][test_index]\n",
    "            # speedups of predictions\n",
    "            runtimes = df[['runtime_cpu', 'runtime_gpu']].values[test_index]\n",
    "            p_runtimes = [r[p_] for p_, r in zip(p, runtimes)]\n",
    "            p_speedup = zer_r_runtimes / p_runtimes\n",
    "    \n",
    "            # sanity check\n",
    "            assert(len(benchmarks) == len(o) == len(correct) == len(p) == len(p_speedup))\n",
    "\n",
    "            # record results\n",
    "            for benchmark_, o_, p_, correct_, p_speedup_ in zip(benchmarks, o, p, correct, p_speedup):\n",
    "                data.append({\n",
    "                    \"Model\": model.__name__,\n",
    "                    \"Platform\": platform_name,\n",
    "                    'Benchmark': escape_benchmark_name(benchmark_),\n",
    "                    'Benchmark Suite': escape_suite_name(benchmark_),\n",
    "                    \"Oracle Mapping\": o_,\n",
    "                    \"Predicted Mapping\": p_,\n",
    "                    \"Correct?\": correct_,\n",
    "                    \"Speedup\": p_speedup_,\n",
    "                })\n",
    "            \n",
    "            # update progress bar\n",
    "            progressbar[0] += 1\n",
    "            progressbar[1].update(progressbar[0])\n",
    "\n",
    "    return pd.DataFrame(\n",
    "        data, index=range(1, len(data)+1), columns=[\n",
    "            \"Model\",\n",
    "            \"Platform\",\n",
    "            \"Benchmark\",\n",
    "            \"Benchmark Suite\", \n",
    "            \"Oracle Mapping\", \n",
    "            \"Predicted Mapping\", \n",
    "            \"Correct?\", \n",
    "            \"Speedup\"\n",
    "        ])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 3.1. Static Mapping\n",
    "\n",
    "A static mapping selects the device which is most frequently optimal for that platform, i.e. it is the upper bound of performance which can be achieved without a heuristic. We use this as a baseline to compare predictive models against:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import pickle\n",
    "\n",
    "class StaticMapping(HeterogemeousMappingModel):\n",
    "    __name__ = \"Static mapping\"\n",
    "    __basename__ = \"static\"\n",
    "    \n",
    "    def init(self, seed: int): return self\n",
    "    \n",
    "    def save(self, outpath):\n",
    "        with open(outpath, \"wb\") as outfile:\n",
    "            pickle.dump(self.model, outfile)\n",
    "\n",
    "    def restore(self, inpath):\n",
    "        with open(inpath, \"rb\") as infile:\n",
    "            self.model = pickle.load(infile)\n",
    "\n",
    "    def train(self, df=None, **train):\n",
    "        from collections import Counter\n",
    "        \n",
    "        # select the Zero-R device: the most frequently optimal device\n",
    "        zero_r_device = Counter(df['oracle']).most_common(1)[0][0]\n",
    "        self.model = 1 if zero_r_device == \"GPU\" else 0\n",
    "\n",
    "    def predict(self, **test):\n",
    "        if self.model:\n",
    "            return np.ones(len(test[\"y\"])).astype(np.int32)\n",
    "        else:\n",
    "            return np.zeros(len(test[\"y\"])).astype(dtype=np.int32)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Evaluating static mapping ...\n",
      " 70% (14 of 20) |#####################         | Elapsed Time: 0:00:00 ETA:  0:00:00"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th>Correct?</th>\n",
       "      <th>Speedup</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Platform</th>\n",
       "      <th>Benchmark Suite</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th rowspan=\"7\" valign=\"top\">AMD Tahiti 7970</th>\n",
       "      <th>AMD SDK</th>\n",
       "      <td>0.625000</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>NPB</th>\n",
       "      <td>0.618596</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>NVIDIA SDK</th>\n",
       "      <td>0.083333</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Parboil</th>\n",
       "      <td>0.473684</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Polybench</th>\n",
       "      <td>0.074074</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Rodinia</th>\n",
       "      <td>0.548387</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>SHOC</th>\n",
       "      <td>0.729167</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"7\" valign=\"top\">NVIDIA GTX 970</th>\n",
       "      <th>AMD SDK</th>\n",
       "      <td>0.937500</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>NPB</th>\n",
       "      <td>0.622391</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>NVIDIA SDK</th>\n",
       "      <td>0.583333</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Parboil</th>\n",
       "      <td>0.315789</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Polybench</th>\n",
       "      <td>0.555556</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Rodinia</th>\n",
       "      <td>0.387097</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>SHOC</th>\n",
       "      <td>0.083333</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                 Correct?  Speedup\n",
       "Platform        Benchmark Suite                   \n",
       "AMD Tahiti 7970 AMD SDK          0.625000      1.0\n",
       "                NPB              0.618596      1.0\n",
       "                NVIDIA SDK       0.083333      1.0\n",
       "                Parboil          0.473684      1.0\n",
       "                Polybench        0.074074      1.0\n",
       "                Rodinia          0.548387      1.0\n",
       "                SHOC             0.729167      1.0\n",
       "NVIDIA GTX 970  AMD SDK          0.937500      1.0\n",
       "                NPB              0.622391      1.0\n",
       "                NVIDIA SDK       0.583333      1.0\n",
       "                Parboil          0.315789      1.0\n",
       "                Polybench        0.555556      1.0\n",
       "                Rodinia          0.387097      1.0\n",
       "                SHOC             0.083333      1.0"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import sys\n",
    "\n",
    "pd.set_option('display.max_rows', 14)\n",
    "\n",
    "print(\"Evaluating static mapping ...\", file=sys.stderr)\n",
    "baseline = evaluate(StaticMapping())\n",
    "baseline.groupby(['Platform', 'Benchmark Suite'])['Platform', 'Correct?', 'Speedup'].mean()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Summary of results. The *Correct?* column shows model accuracy. In the paper we report that the static mapping is accurate for **58.8%** of cases on AMD and **56.9%** of cases on NVIDIA:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Correct?</th>\n",
       "      <th>Speedup</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Platform</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>AMD Tahiti 7970</th>\n",
       "      <td>0.588235</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>NVIDIA GTX 970</th>\n",
       "      <td>0.569118</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                 Correct?  Speedup\n",
       "Platform                          \n",
       "AMD Tahiti 7970  0.588235      1.0\n",
       "NVIDIA GTX 970   0.569118      1.0"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "baseline.groupby(['Platform'])['Platform', 'Correct?', 'Speedup'].mean()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 3.2. Grewe et al. Model"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The Grewe et al. predictive model uses decision trees and hand engineered features to predict optimal device mapping, described in publication:\n",
    "\n",
    "> [2] Grewe, D., Wang, Z., & O’Boyle, M. (2013). [Portable Mapping of Data Parallel Programs to OpenCL for Heterogeneous Systems](http://www.ece.neu.edu/groups/nucar/NUCARTALKS/cgo2013-grewe.pdf). In CGO. IEEE."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "class Grewe(HeterogemeousMappingModel):\n",
    "    __name__ = \"Grewe et al.\"\n",
    "    __basename__ = \"grewe\"\n",
    "    \n",
    "    def init(self, seed: int):\n",
    "        from sklearn.tree import DecisionTreeClassifier\n",
    "\n",
    "        self.model = DecisionTreeClassifier(\n",
    "            random_state=seed, splitter=\"best\",\n",
    "            criterion=\"entropy\", max_depth=5,\n",
    "            min_samples_leaf=5)\n",
    "        return self\n",
    "    \n",
    "    def save(self, outpath):\n",
    "        with open(outpath, \"wb\") as outfile:\n",
    "            pickle.dump(self.model, outfile)\n",
    "\n",
    "    def restore(self, inpath):\n",
    "        with open(inpath, \"rb\") as infile:\n",
    "            self.model = pickle.load(infile)\n",
    "        \n",
    "    def train(self, **train):\n",
    "        self.model.fit(train[\"features\"], train[\"y\"])\n",
    "\n",
    "    def predict(self, **test):\n",
    "        return self.model.predict(test[\"features\"])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Evaluating Grewe et al. ...\n",
      " 65% (13 of 20) |###################           | Elapsed Time: 0:00:00 ETA:  0:00:00"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th>Correct?</th>\n",
       "      <th>Speedup</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Platform</th>\n",
       "      <th>Benchmark Suite</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th rowspan=\"7\" valign=\"top\">AMD Tahiti 7970</th>\n",
       "      <th>AMD SDK</th>\n",
       "      <td>0.750000</td>\n",
       "      <td>1.088049</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>NPB</th>\n",
       "      <td>0.740038</td>\n",
       "      <td>2.447364</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>NVIDIA SDK</th>\n",
       "      <td>0.750000</td>\n",
       "      <td>3.694169</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Parboil</th>\n",
       "      <td>0.631579</td>\n",
       "      <td>5.927718</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Polybench</th>\n",
       "      <td>0.592593</td>\n",
       "      <td>13.283587</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Rodinia</th>\n",
       "      <td>0.645161</td>\n",
       "      <td>3.355448</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>SHOC</th>\n",
       "      <td>0.833333</td>\n",
       "      <td>1.024105</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"7\" valign=\"top\">NVIDIA GTX 970</th>\n",
       "      <th>AMD SDK</th>\n",
       "      <td>0.625000</td>\n",
       "      <td>0.783911</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>NPB</th>\n",
       "      <td>0.785579</td>\n",
       "      <td>1.270429</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>NVIDIA SDK</th>\n",
       "      <td>0.416667</td>\n",
       "      <td>0.922276</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Parboil</th>\n",
       "      <td>0.263158</td>\n",
       "      <td>1.014902</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Polybench</th>\n",
       "      <td>0.592593</td>\n",
       "      <td>1.036523</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Rodinia</th>\n",
       "      <td>0.387097</td>\n",
       "      <td>0.993468</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>SHOC</th>\n",
       "      <td>0.708333</td>\n",
       "      <td>1.851509</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                 Correct?    Speedup\n",
       "Platform        Benchmark Suite                     \n",
       "AMD Tahiti 7970 AMD SDK          0.750000   1.088049\n",
       "                NPB              0.740038   2.447364\n",
       "                NVIDIA SDK       0.750000   3.694169\n",
       "                Parboil          0.631579   5.927718\n",
       "                Polybench        0.592593  13.283587\n",
       "                Rodinia          0.645161   3.355448\n",
       "                SHOC             0.833333   1.024105\n",
       "NVIDIA GTX 970  AMD SDK          0.625000   0.783911\n",
       "                NPB              0.785579   1.270429\n",
       "                NVIDIA SDK       0.416667   0.922276\n",
       "                Parboil          0.263158   1.014902\n",
       "                Polybench        0.592593   1.036523\n",
       "                Rodinia          0.387097   0.993468\n",
       "                SHOC             0.708333   1.851509"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "print(\"Evaluating Grewe et al. ...\", file=sys.stderr)\n",
    "grewe = evaluate(Grewe())\n",
    "grewe.groupby(['Platform', 'Benchmark Suite'])['Platform', 'Correct?', 'Speedup'].mean()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Summary of results. The *Correct?* column shows model accuracy. In the paper we report that the Grewe et al. model is accurate **73%** of cases, achieving speedups of **2.91$\\times$** on AMD and **1.26$\\times$** on NVIDIA:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Correct?</th>\n",
       "      <th>Speedup</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Platform</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>AMD Tahiti 7970</th>\n",
       "      <td>0.733824</td>\n",
       "      <td>2.905822</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>NVIDIA GTX 970</th>\n",
       "      <td>0.729412</td>\n",
       "      <td>1.264801</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                 Correct?   Speedup\n",
       "Platform                           \n",
       "AMD Tahiti 7970  0.733824  2.905822\n",
       "NVIDIA GTX 970   0.729412  1.264801"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "grewe.groupby(['Platform'])['Platform', 'Correct?', 'Speedup'].mean()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 3.3. DeepTune\n",
    "\n",
    "We predict optimal device mapping from raw source code inputs, and the\n",
    "two auxiliary inputs wgsize and dsize (which cannot be obtained statically\n",
    "from the program source):"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "class DeepTune(HeterogemeousMappingModel):\n",
    "    __name__ = \"DeepTune\"\n",
    "    __basename__ = \"deeptune\"\n",
    "    \n",
    "    def init(self, seed: int):\n",
    "        from keras.layers import Input, Embedding, LSTM, Dense\n",
    "        from keras.layers.merge import Concatenate\n",
    "        from keras.layers.normalization import BatchNormalization\n",
    "        from keras.models import Model\n",
    "\n",
    "        np.random.seed(seed)\n",
    "        \n",
    "        # Language model. Takes as inputs source code sequences.\n",
    "        code_in = Input(shape=(1024,), dtype=\"int32\", name=\"code_in\")\n",
    "        x = Embedding(input_dim=atomizer.vocab_size + 1, input_length=1024, \n",
    "                      output_dim=64, name=\"embedding\")(code_in)\n",
    "        x = LSTM(64, implementation=1, return_sequences=True, name=\"lstm_1\")(x)\n",
    "        x = LSTM(64, implementation=1, name=\"lstm_2\")(x)\n",
    "        langmodel_out = Dense(2, activation=\"sigmoid\")(x)\n",
    "        \n",
    "        # Auxiliary inputs. wgsize and dsize.\n",
    "        auxiliary_inputs = Input(shape=(2,))\n",
    "        \n",
    "        # Heuristic model. Takes as inputs the language model,\n",
    "        #   outputs 1-hot encoded device mapping\n",
    "        x = Concatenate()([auxiliary_inputs, x])\n",
    "        x = BatchNormalization()(x)\n",
    "        x = Dense(32, activation=\"relu\")(x)\n",
    "        out = Dense(2, activation=\"sigmoid\")(x)\n",
    "\n",
    "        self.model = Model(inputs=[auxiliary_inputs, code_in], outputs=[out, langmodel_out])\n",
    "        self.model.compile(\n",
    "            optimizer=\"adam\", metrics=['accuracy'],\n",
    "            loss=[\"categorical_crossentropy\", \"categorical_crossentropy\"],\n",
    "            loss_weights=[1., .2])\n",
    "        \n",
    "        return self\n",
    "\n",
    "    def save(self, outpath):\n",
    "        self.model.save(outpath)\n",
    "\n",
    "    def restore(self, inpath):\n",
    "        from keras.models import load_model\n",
    "        self.model = load_model(inpath)\n",
    "        \n",
    "    def train(self, **train):\n",
    "        self.model.fit([train[\"aux_in\"], train[\"sequences\"]], [train[\"y_1hot\"], train[\"y_1hot\"]],\n",
    "                       epochs=50, batch_size=64, verbose=train[\"verbose\"], shuffle=True)\n",
    "\n",
    "    def predict(self, **test):\n",
    "        p = np.array(self.model.predict(\n",
    "            [test[\"aux_in\"], test[\"sequences\"]], batch_size=64, verbose=test[\"verbose\"]))\n",
    "        indices = [np.argmax(x) for x in p[0]]\n",
    "        return indices"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "An overview of the DeepTune model, showing the number of parameters in each layer, used to construct Table 5 in the paper:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "____________________________________________________________________________________________________\n",
      "Layer (type)                     Output Shape          Param #     Connected to                     \n",
      "====================================================================================================\n",
      "code_in (InputLayer)             (None, 1024)          0                                            \n",
      "____________________________________________________________________________________________________\n",
      "embedding (Embedding)            (None, 1024, 64)      8256                                         \n",
      "____________________________________________________________________________________________________\n",
      "lstm_1 (LSTM)                    (None, 1024, 64)      33024                                        \n",
      "____________________________________________________________________________________________________\n",
      "input_1 (InputLayer)             (None, 2)             0                                            \n",
      "____________________________________________________________________________________________________\n",
      "lstm_2 (LSTM)                    (None, 64)            33024                                        \n",
      "____________________________________________________________________________________________________\n",
      "concatenate_1 (Concatenate)      (None, 66)            0                                            \n",
      "____________________________________________________________________________________________________\n",
      "batch_normalization_1 (BatchNorm (None, 66)            264                                          \n",
      "____________________________________________________________________________________________________\n",
      "dense_2 (Dense)                  (None, 32)            2144                                         \n",
      "____________________________________________________________________________________________________\n",
      "dense_3 (Dense)                  (None, 2)             66                                           \n",
      "____________________________________________________________________________________________________\n",
      "dense_1 (Dense)                  (None, 2)             130                                          \n",
      "====================================================================================================\n",
      "Total params: 76,908.0\n",
      "Trainable params: 76,776.0\n",
      "Non-trainable params: 132.0\n",
      "____________________________________________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "deeptune_model = DeepTune()\n",
    "deeptune_model.init(seed)\n",
    "deeptune_model.model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/svg+xml": [
       "<svg height=\"636pt\" viewBox=\"0.00 0.00 656.00 636.00\" width=\"656pt\" xmlns=\"http://www.w3.org/2000/svg\" xmlns:xlink=\"http://www.w3.org/1999/xlink\">\n",
       "<g class=\"graph\" id=\"graph0\" transform=\"scale(1 1) rotate(0) translate(4 632)\">\n",
       "<title>G</title>\n",
       "<polygon fill=\"white\" points=\"-4,4 -4,-632 652,-632 652,4 -4,4\" stroke=\"none\"/>\n",
       "<!-- 140695028991648 -->\n",
       "<g class=\"node\" id=\"node1\"><title>140695028991648</title>\n",
       "<polygon fill=\"none\" points=\"356,-581.5 356,-627.5 627,-627.5 627,-581.5 356,-581.5\" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"419.5\" y=\"-600.8\">code_in: InputLayer</text>\n",
       "<polyline fill=\"none\" points=\"483,-581.5 483,-627.5 \" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"510.5\" y=\"-612.3\">input:</text>\n",
       "<polyline fill=\"none\" points=\"483,-604.5 538,-604.5 \" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"510.5\" y=\"-589.3\">output:</text>\n",
       "<polyline fill=\"none\" points=\"538,-581.5 538,-627.5 \" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"582.5\" y=\"-612.3\">(None, 1024)</text>\n",
       "<polyline fill=\"none\" points=\"538,-604.5 627,-604.5 \" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"582.5\" y=\"-589.3\">(None, 1024)</text>\n",
       "</g>\n",
       "<!-- 140695028991592 -->\n",
       "<g class=\"node\" id=\"node2\"><title>140695028991592</title>\n",
       "<polygon fill=\"none\" points=\"335,-498.5 335,-544.5 648,-544.5 648,-498.5 335,-498.5\" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"409\" y=\"-517.8\">embedding: Embedding</text>\n",
       "<polyline fill=\"none\" points=\"483,-498.5 483,-544.5 \" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"510.5\" y=\"-529.3\">input:</text>\n",
       "<polyline fill=\"none\" points=\"483,-521.5 538,-521.5 \" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"510.5\" y=\"-506.3\">output:</text>\n",
       "<polyline fill=\"none\" points=\"538,-498.5 538,-544.5 \" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"593\" y=\"-529.3\">(None, 1024)</text>\n",
       "<polyline fill=\"none\" points=\"538,-521.5 648,-521.5 \" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"593\" y=\"-506.3\">(None, 1024, 64)</text>\n",
       "</g>\n",
       "<!-- 140695028991648&#45;&gt;140695028991592 -->\n",
       "<g class=\"edge\" id=\"edge1\"><title>140695028991648-&gt;140695028991592</title>\n",
       "<path d=\"M491.5,-581.366C491.5,-573.152 491.5,-563.658 491.5,-554.725\" fill=\"none\" stroke=\"black\"/>\n",
       "<polygon fill=\"black\" points=\"495,-554.607 491.5,-544.607 488,-554.607 495,-554.607\" stroke=\"black\"/>\n",
       "</g>\n",
       "<!-- 140695028992656 -->\n",
       "<g class=\"node\" id=\"node3\"><title>140695028992656</title>\n",
       "<polygon fill=\"none\" points=\"360,-415.5 360,-461.5 623,-461.5 623,-415.5 360,-415.5\" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"409\" y=\"-434.8\">lstm_1: LSTM</text>\n",
       "<polyline fill=\"none\" points=\"458,-415.5 458,-461.5 \" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"485.5\" y=\"-446.3\">input:</text>\n",
       "<polyline fill=\"none\" points=\"458,-438.5 513,-438.5 \" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"485.5\" y=\"-423.3\">output:</text>\n",
       "<polyline fill=\"none\" points=\"513,-415.5 513,-461.5 \" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"568\" y=\"-446.3\">(None, 1024, 64)</text>\n",
       "<polyline fill=\"none\" points=\"513,-438.5 623,-438.5 \" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"568\" y=\"-423.3\">(None, 1024, 64)</text>\n",
       "</g>\n",
       "<!-- 140695028991592&#45;&gt;140695028992656 -->\n",
       "<g class=\"edge\" id=\"edge2\"><title>140695028991592-&gt;140695028992656</title>\n",
       "<path d=\"M491.5,-498.366C491.5,-490.152 491.5,-480.658 491.5,-471.725\" fill=\"none\" stroke=\"black\"/>\n",
       "<polygon fill=\"black\" points=\"495,-471.607 491.5,-461.607 488,-471.607 495,-471.607\" stroke=\"black\"/>\n",
       "</g>\n",
       "<!-- 140695795427200 -->\n",
       "<g class=\"node\" id=\"node5\"><title>140695795427200</title>\n",
       "<polygon fill=\"none\" points=\"360,-332.5 360,-378.5 623,-378.5 623,-332.5 360,-332.5\" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"409\" y=\"-351.8\">lstm_2: LSTM</text>\n",
       "<polyline fill=\"none\" points=\"458,-332.5 458,-378.5 \" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"485.5\" y=\"-363.3\">input:</text>\n",
       "<polyline fill=\"none\" points=\"458,-355.5 513,-355.5 \" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"485.5\" y=\"-340.3\">output:</text>\n",
       "<polyline fill=\"none\" points=\"513,-332.5 513,-378.5 \" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"568\" y=\"-363.3\">(None, 1024, 64)</text>\n",
       "<polyline fill=\"none\" points=\"513,-355.5 623,-355.5 \" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"568\" y=\"-340.3\">(None, 64)</text>\n",
       "</g>\n",
       "<!-- 140695028992656&#45;&gt;140695795427200 -->\n",
       "<g class=\"edge\" id=\"edge3\"><title>140695028992656-&gt;140695795427200</title>\n",
       "<path d=\"M491.5,-415.366C491.5,-407.152 491.5,-397.658 491.5,-388.725\" fill=\"none\" stroke=\"black\"/>\n",
       "<polygon fill=\"black\" points=\"495,-388.607 491.5,-378.607 488,-388.607 495,-388.607\" stroke=\"black\"/>\n",
       "</g>\n",
       "<!-- 140694262565744 -->\n",
       "<g class=\"node\" id=\"node4\"><title>140694262565744</title>\n",
       "<polygon fill=\"none\" points=\"71,-332.5 71,-378.5 320,-378.5 320,-332.5 71,-332.5\" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"133.5\" y=\"-351.8\">input_1: InputLayer</text>\n",
       "<polyline fill=\"none\" points=\"196,-332.5 196,-378.5 \" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"223.5\" y=\"-363.3\">input:</text>\n",
       "<polyline fill=\"none\" points=\"196,-355.5 251,-355.5 \" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"223.5\" y=\"-340.3\">output:</text>\n",
       "<polyline fill=\"none\" points=\"251,-332.5 251,-378.5 \" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"285.5\" y=\"-363.3\">(None, 2)</text>\n",
       "<polyline fill=\"none\" points=\"251,-355.5 320,-355.5 \" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"285.5\" y=\"-340.3\">(None, 2)</text>\n",
       "</g>\n",
       "<!-- 140694204621040 -->\n",
       "<g class=\"node\" id=\"node6\"><title>140694204621040</title>\n",
       "<polygon fill=\"none\" points=\"11.5,-249.5 11.5,-295.5 379.5,-295.5 379.5,-249.5 11.5,-249.5\" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"95.5\" y=\"-268.8\">concatenate_1: Concatenate</text>\n",
       "<polyline fill=\"none\" points=\"179.5,-249.5 179.5,-295.5 \" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"207\" y=\"-280.3\">input:</text>\n",
       "<polyline fill=\"none\" points=\"179.5,-272.5 234.5,-272.5 \" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"207\" y=\"-257.3\">output:</text>\n",
       "<polyline fill=\"none\" points=\"234.5,-249.5 234.5,-295.5 \" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"307\" y=\"-280.3\">[(None, 2), (None, 64)]</text>\n",
       "<polyline fill=\"none\" points=\"234.5,-272.5 379.5,-272.5 \" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"307\" y=\"-257.3\">(None, 66)</text>\n",
       "</g>\n",
       "<!-- 140694262565744&#45;&gt;140694204621040 -->\n",
       "<g class=\"edge\" id=\"edge4\"><title>140694262565744-&gt;140694204621040</title>\n",
       "<path d=\"M195.5,-332.366C195.5,-324.152 195.5,-314.658 195.5,-305.725\" fill=\"none\" stroke=\"black\"/>\n",
       "<polygon fill=\"black\" points=\"199,-305.607 195.5,-295.607 192,-305.607 199,-305.607\" stroke=\"black\"/>\n",
       "</g>\n",
       "<!-- 140695795427200&#45;&gt;140694204621040 -->\n",
       "<g class=\"edge\" id=\"edge5\"><title>140695795427200-&gt;140694204621040</title>\n",
       "<path d=\"M411.008,-332.473C372.26,-321.87 325.655,-309.117 285.712,-298.187\" fill=\"none\" stroke=\"black\"/>\n",
       "<polygon fill=\"black\" points=\"286.546,-294.786 275.976,-295.522 284.698,-301.538 286.546,-294.786\" stroke=\"black\"/>\n",
       "</g>\n",
       "<!-- 140695741463520 -->\n",
       "<g class=\"node\" id=\"node10\"><title>140695741463520</title>\n",
       "<polygon fill=\"none\" points=\"398,-249.5 398,-295.5 631,-295.5 631,-249.5 398,-249.5\" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"449\" y=\"-268.8\">dense_1: Dense</text>\n",
       "<polyline fill=\"none\" points=\"500,-249.5 500,-295.5 \" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"527.5\" y=\"-280.3\">input:</text>\n",
       "<polyline fill=\"none\" points=\"500,-272.5 555,-272.5 \" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"527.5\" y=\"-257.3\">output:</text>\n",
       "<polyline fill=\"none\" points=\"555,-249.5 555,-295.5 \" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"593\" y=\"-280.3\">(None, 64)</text>\n",
       "<polyline fill=\"none\" points=\"555,-272.5 631,-272.5 \" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"593\" y=\"-257.3\">(None, 2)</text>\n",
       "</g>\n",
       "<!-- 140695795427200&#45;&gt;140695741463520 -->\n",
       "<g class=\"edge\" id=\"edge9\"><title>140695795427200-&gt;140695741463520</title>\n",
       "<path d=\"M497.785,-332.366C500.143,-324.062 502.872,-314.451 505.432,-305.434\" fill=\"none\" stroke=\"black\"/>\n",
       "<polygon fill=\"black\" points=\"508.858,-306.183 508.223,-295.607 502.124,-304.27 508.858,-306.183\" stroke=\"black\"/>\n",
       "</g>\n",
       "<!-- 140694204624232 -->\n",
       "<g class=\"node\" id=\"node7\"><title>140694204624232</title>\n",
       "<polygon fill=\"none\" points=\"0,-166.5 0,-212.5 391,-212.5 391,-166.5 0,-166.5\" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"130\" y=\"-185.8\">batch_normalization_1: BatchNormalization</text>\n",
       "<polyline fill=\"none\" points=\"260,-166.5 260,-212.5 \" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"287.5\" y=\"-197.3\">input:</text>\n",
       "<polyline fill=\"none\" points=\"260,-189.5 315,-189.5 \" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"287.5\" y=\"-174.3\">output:</text>\n",
       "<polyline fill=\"none\" points=\"315,-166.5 315,-212.5 \" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"353\" y=\"-197.3\">(None, 66)</text>\n",
       "<polyline fill=\"none\" points=\"315,-189.5 391,-189.5 \" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"353\" y=\"-174.3\">(None, 66)</text>\n",
       "</g>\n",
       "<!-- 140694204621040&#45;&gt;140694204624232 -->\n",
       "<g class=\"edge\" id=\"edge6\"><title>140694204621040-&gt;140694204624232</title>\n",
       "<path d=\"M195.5,-249.366C195.5,-241.152 195.5,-231.658 195.5,-222.725\" fill=\"none\" stroke=\"black\"/>\n",
       "<polygon fill=\"black\" points=\"199,-222.607 195.5,-212.607 192,-222.607 199,-222.607\" stroke=\"black\"/>\n",
       "</g>\n",
       "<!-- 140694204711600 -->\n",
       "<g class=\"node\" id=\"node8\"><title>140694204711600</title>\n",
       "<polygon fill=\"none\" points=\"79,-83.5 79,-129.5 312,-129.5 312,-83.5 79,-83.5\" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"130\" y=\"-102.8\">dense_2: Dense</text>\n",
       "<polyline fill=\"none\" points=\"181,-83.5 181,-129.5 \" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"208.5\" y=\"-114.3\">input:</text>\n",
       "<polyline fill=\"none\" points=\"181,-106.5 236,-106.5 \" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"208.5\" y=\"-91.3\">output:</text>\n",
       "<polyline fill=\"none\" points=\"236,-83.5 236,-129.5 \" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"274\" y=\"-114.3\">(None, 66)</text>\n",
       "<polyline fill=\"none\" points=\"236,-106.5 312,-106.5 \" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"274\" y=\"-91.3\">(None, 32)</text>\n",
       "</g>\n",
       "<!-- 140694204624232&#45;&gt;140694204711600 -->\n",
       "<g class=\"edge\" id=\"edge7\"><title>140694204624232-&gt;140694204711600</title>\n",
       "<path d=\"M195.5,-166.366C195.5,-158.152 195.5,-148.658 195.5,-139.725\" fill=\"none\" stroke=\"black\"/>\n",
       "<polygon fill=\"black\" points=\"199,-139.607 195.5,-129.607 192,-139.607 199,-139.607\" stroke=\"black\"/>\n",
       "</g>\n",
       "<!-- 140694204095568 -->\n",
       "<g class=\"node\" id=\"node9\"><title>140694204095568</title>\n",
       "<polygon fill=\"none\" points=\"79,-0.5 79,-46.5 312,-46.5 312,-0.5 79,-0.5\" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"130\" y=\"-19.8\">dense_3: Dense</text>\n",
       "<polyline fill=\"none\" points=\"181,-0.5 181,-46.5 \" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"208.5\" y=\"-31.3\">input:</text>\n",
       "<polyline fill=\"none\" points=\"181,-23.5 236,-23.5 \" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"208.5\" y=\"-8.3\">output:</text>\n",
       "<polyline fill=\"none\" points=\"236,-0.5 236,-46.5 \" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"274\" y=\"-31.3\">(None, 32)</text>\n",
       "<polyline fill=\"none\" points=\"236,-23.5 312,-23.5 \" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"274\" y=\"-8.3\">(None, 2)</text>\n",
       "</g>\n",
       "<!-- 140694204711600&#45;&gt;140694204095568 -->\n",
       "<g class=\"edge\" id=\"edge8\"><title>140694204711600-&gt;140694204095568</title>\n",
       "<path d=\"M195.5,-83.3664C195.5,-75.1516 195.5,-65.6579 195.5,-56.7252\" fill=\"none\" stroke=\"black\"/>\n",
       "<polygon fill=\"black\" points=\"199,-56.6068 195.5,-46.6068 192,-56.6069 199,-56.6068\" stroke=\"black\"/>\n",
       "</g>\n",
       "</g>\n",
       "</svg>"
      ],
      "text/plain": [
       "<IPython.core.display.SVG object>"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "def isnotebook():\n",
    "    # written by @mtd http://stackoverflow.com/a/39662359\n",
    "    try:\n",
    "        shell = get_ipython().__class__.__name__\n",
    "        if shell == 'ZMQInteractiveShell':  # Jupyter notebook or qtconsole?\n",
    "            return True\n",
    "        elif shell == 'TerminalInteractiveShell':  # Terminal running IPython?\n",
    "            return False\n",
    "        else:\n",
    "            return False  # Other type (?)\n",
    "    except NameError:\n",
    "        return False      # Probably standard Python interpreter\n",
    "\n",
    "\n",
    "img = None\n",
    "if isnotebook():\n",
    "    from keras.utils.vis_utils import model_to_dot\n",
    "    from IPython.display import SVG\n",
    "    img = SVG(model_to_dot(deeptune_model.model, show_shapes=True).create(prog='dot', format='svg'))\n",
    "img"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Evaluating DeepTune ...\n",
      " 80% (16 of 20) |########################      | Elapsed Time: 0:00:00 ETA:  0:00:00"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th>Correct?</th>\n",
       "      <th>Speedup</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Platform</th>\n",
       "      <th>Benchmark Suite</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th rowspan=\"7\" valign=\"top\">AMD Tahiti 7970</th>\n",
       "      <th>AMD SDK</th>\n",
       "      <td>0.562500</td>\n",
       "      <td>0.846116</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>NPB</th>\n",
       "      <td>0.851992</td>\n",
       "      <td>2.851700</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>NVIDIA SDK</th>\n",
       "      <td>0.666667</td>\n",
       "      <td>3.669208</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Parboil</th>\n",
       "      <td>0.947368</td>\n",
       "      <td>6.403654</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Polybench</th>\n",
       "      <td>0.814815</td>\n",
       "      <td>14.065773</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Rodinia</th>\n",
       "      <td>0.741935</td>\n",
       "      <td>4.752888</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>SHOC</th>\n",
       "      <td>0.833333</td>\n",
       "      <td>1.229528</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"7\" valign=\"top\">NVIDIA GTX 970</th>\n",
       "      <th>AMD SDK</th>\n",
       "      <td>0.500000</td>\n",
       "      <td>0.716424</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>NPB</th>\n",
       "      <td>0.836812</td>\n",
       "      <td>1.407999</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>NVIDIA SDK</th>\n",
       "      <td>0.666667</td>\n",
       "      <td>1.069700</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Parboil</th>\n",
       "      <td>0.789474</td>\n",
       "      <td>1.459396</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Polybench</th>\n",
       "      <td>0.555556</td>\n",
       "      <td>1.038921</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Rodinia</th>\n",
       "      <td>0.677419</td>\n",
       "      <td>1.566724</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>SHOC</th>\n",
       "      <td>0.791667</td>\n",
       "      <td>1.867671</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                 Correct?    Speedup\n",
       "Platform        Benchmark Suite                     \n",
       "AMD Tahiti 7970 AMD SDK          0.562500   0.846116\n",
       "                NPB              0.851992   2.851700\n",
       "                NVIDIA SDK       0.666667   3.669208\n",
       "                Parboil          0.947368   6.403654\n",
       "                Polybench        0.814815  14.065773\n",
       "                Rodinia          0.741935   4.752888\n",
       "                SHOC             0.833333   1.229528\n",
       "NVIDIA GTX 970  AMD SDK          0.500000   0.716424\n",
       "                NPB              0.836812   1.407999\n",
       "                NVIDIA SDK       0.666667   1.069700\n",
       "                Parboil          0.789474   1.459396\n",
       "                Polybench        0.555556   1.038921\n",
       "                Rodinia          0.677419   1.566724\n",
       "                SHOC             0.791667   1.867671"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "print(\"Evaluating DeepTune ...\", file=sys.stderr)\n",
    "deeptune = evaluate(deeptune_model)\n",
    "deeptune.groupby(['Platform', 'Benchmark Suite'])['Platform', 'Correct?', 'Speedup'].mean()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Summary of results. The *Correct?* column shows model accuracy. In the paper we report that DeepTune is accurate for **82%** of cases, achieving speedups of **3.34$\\times$** on AMD and **1.41$\\times$** on NVIDIA:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Correct?</th>\n",
       "      <th>Speedup</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Platform</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>AMD Tahiti 7970</th>\n",
       "      <td>0.836765</td>\n",
       "      <td>3.335612</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>NVIDIA GTX 970</th>\n",
       "      <td>0.802941</td>\n",
       "      <td>1.412222</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                 Correct?   Speedup\n",
       "Platform                           \n",
       "AMD Tahiti 7970  0.836765  3.335612\n",
       "NVIDIA GTX 970   0.802941  1.412222"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "deeptune.groupby(['Platform'])['Platform', 'Correct?', 'Speedup'].mean()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 4. Evaluation\n",
    "\n",
    "Here we evaluate the quality of the models using two metrics: prediction accuracy, and performance relative to the static mapping:\n",
    "\n",
    "### 4.1. Prediction Accuracy"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Static mapping</th>\n",
       "      <th>Grewe et al.</th>\n",
       "      <th>DeepTune</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>AMD Tahiti 7970</th>\n",
       "      <td>58.823529</td>\n",
       "      <td>73.382353</td>\n",
       "      <td>83.676471</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>NVIDIA GTX 970</th>\n",
       "      <td>56.911765</td>\n",
       "      <td>72.941176</td>\n",
       "      <td>80.294118</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Average</th>\n",
       "      <td>57.867647</td>\n",
       "      <td>73.161765</td>\n",
       "      <td>81.985294</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                 Static mapping  Grewe et al.   DeepTune\n",
       "AMD Tahiti 7970       58.823529     73.382353  83.676471\n",
       "NVIDIA GTX 970        56.911765     72.941176  80.294118\n",
       "Average               57.867647     73.161765  81.985294"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "d = []\n",
    "d.append(np.append(baseline.groupby(['Platform'])['Correct?'].mean().values * 100,\n",
    "                   baseline['Correct?'].mean() * 100))\n",
    "d.append(np.append(grewe.groupby(['Platform'])['Correct?'].mean().values * 100,\n",
    "                   grewe['Correct?'].mean() * 100))\n",
    "d.append(np.append(deeptune.groupby(['Platform'])['Correct?'].mean().values * 100,\n",
    "                   deeptune['Correct?'].mean() * 100))\n",
    "d = np.array(d).T.reshape(3, 3)\n",
    "\n",
    "pd.DataFrame(d, columns=['Static mapping', 'Grewe et al.', 'DeepTune'],\n",
    "             index=['AMD Tahiti 7970', 'NVIDIA GTX 970', 'Average'])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Figure 6 of the paper:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAeoAAAJICAYAAACuUadCAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n1IQQQkgAo0BNCCGEBLD/APepIi49vm2xAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7ff649137860>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "if isnotebook():\n",
    "    import matplotlib\n",
    "    import matplotlib.pyplot as plt\n",
    "    import seaborn as sns\n",
    "\n",
    "    from matplotlib.ticker import FormatStrFormatter\n",
    "    from labm8 import viz\n",
    "\n",
    "    # Plotting configuration\n",
    "    %matplotlib inline\n",
    "    sns.set(style=\"ticks\", color_codes=True)\n",
    "    plt.style.use([\"seaborn-white\", \"seaborn-paper\"])\n",
    "    matplotlib.rcParams['figure.dpi'] = 120\n",
    "    \n",
    "    for i, platform in enumerate([\"AMD Tahiti 7970\", \"NVIDIA GTX 970\"]):\n",
    "        \n",
    "        def get_accuracies(df, platform):\n",
    "            \"\"\" get % accuracies for platform, aggregated by benchmark suite \"\"\"\n",
    "            # average by benchmark suites\n",
    "            series = df[df[\"Platform\"] == platform]\\\n",
    "                        .groupby(['Benchmark Suite'])['Correct?'].mean() * 100  # %\n",
    "            # append average across all benchamrks\n",
    "            average = pd.Series(df[df[\"Platform\"] == platform]['Correct?'].mean() * 100,\n",
    "                                index=[\"Average\"])\n",
    "            return series.append(average)\n",
    "\n",
    "        # Aggregate data\n",
    "        models, results = (StaticMapping, Grewe, DeepTune), (baseline, grewe, deeptune)\n",
    "        dfs = [get_accuracies(r, platform) for r in results]\n",
    "\n",
    "        accuracies = np.concatenate([dfs[i].values for i in range(len(dfs))])\n",
    "        benchmarks = np.concatenate([dfs[i].keys().values for i in range(len(dfs))])\n",
    "        models = np.concatenate([np.array([m().__name__] * len(dfs[0])) for m in models])\n",
    "\n",
    "        benchmarks = np.append(benchmarks, \"Average\")\n",
    "\n",
    "        df = pd.DataFrame([{\"Benchmark\": b, \"Accuracy\": a, \"Model\": m}\n",
    "                           for b, a, m in zip(benchmarks, accuracies, models)])\n",
    "\n",
    "        # Plot\n",
    "        plt.subplot(2, 1, i + 1)\n",
    "        palette = sns.cubehelix_palette(3, rot=.2, light=.85, dark=.35)\n",
    "        ax = sns.barplot(x=\"Benchmark\", y=\"Accuracy\", hue=\"Model\", palette=palette, data=df)\n",
    "\n",
    "        # x axis\n",
    "        c = chr(ord('a') + i)\n",
    "        plt.xlabel(f\"({c}) {platform}\", fontsize=10)\n",
    "        plt.setp(ax.get_xticklabels(), rotation=25)  # rotate x ticks\n",
    "        plt.axvline(x=plt.xlim()[1] - 1, color=\"k\", lw=1, linestyle=\"--\")  # average line\n",
    "\n",
    "        # y axis\n",
    "        plt.ylim((0, 100))    \n",
    "        plt.gca().yaxis.set_major_formatter(FormatStrFormatter(\"%d%%\"))\n",
    "        plt.ylabel(\"Accuracy\")\n",
    "\n",
    "        # legend\n",
    "        if i == 0:\n",
    "            plt.legend(loc=(0.003, 1.05), ncol=3)\n",
    "            ax.get_legend().set_title(\"\")  # no legend title\n",
    "            ax.get_legend().draw_frame(True)\n",
    "        elif i == 1:\n",
    "            ax.legend().set_visible(False)\n",
    "\n",
    "    viz.finalise(\n",
    "        # TODO: fs.path(\"~/Inbox/cgo-acc.pdf\"),\n",
    "        figsize=(4.25, 4.8))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 4.2. Speedups\n",
    "\n",
    "Mean speedup:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Static mapping</th>\n",
       "      <th>Grewe et al.</th>\n",
       "      <th>DeepTune</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>AMD Tahiti 7970</th>\n",
       "      <td>1.0</td>\n",
       "      <td>2.905822</td>\n",
       "      <td>3.335612</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>NVIDIA GTX 970</th>\n",
       "      <td>1.0</td>\n",
       "      <td>1.264801</td>\n",
       "      <td>1.412222</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Average</th>\n",
       "      <td>1.0</td>\n",
       "      <td>2.085312</td>\n",
       "      <td>2.373917</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                 Static mapping  Grewe et al.  DeepTune\n",
       "AMD Tahiti 7970             1.0      2.905822  3.335612\n",
       "NVIDIA GTX 970              1.0      1.264801  1.412222\n",
       "Average                     1.0      2.085312  2.373917"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "d = []\n",
    "d.append(np.append(baseline.groupby(['Platform'])['Speedup'].mean().values,\n",
    "                   baseline['Speedup'].mean()))\n",
    "d.append(np.append(grewe.groupby(['Platform'])['Speedup'].mean().values,\n",
    "                   grewe['Speedup'].mean()))\n",
    "d.append(np.append(deeptune.groupby(['Platform'])['Speedup'].mean().values,\n",
    "                   deeptune['Speedup'].mean()))\n",
    "d = np.array(d).T.reshape(3, 3)\n",
    "\n",
    "pd.DataFrame(d, columns=['Static mapping', 'Grewe et al.', 'DeepTune'],\n",
    "             index=['AMD Tahiti 7970', 'NVIDIA GTX 970', 'Average'])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In the paper we report geometric means of **1.18$\\times$** for Grewe et al., and **1.31$\\times$** for DeepTune:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Geometric mean of Grewe et al. 1.18x\n",
      "Geometric mean of DeepTune 1.31x\n"
     ]
    }
   ],
   "source": [
    "from labm8 import math as labmath\n",
    "\n",
    "grewe_geomean = labmath.geomean(grewe['Speedup'].values)\n",
    "deeptune_geomean = labmath.geomean(deeptune['Speedup'].values)\n",
    "\n",
    "print(f\"Geometric mean of Grewe et al. {grewe_geomean:.2f}x\")\n",
    "print(f\"Geometric mean of DeepTune {deeptune_geomean:.2f}x\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Figure 7 of the paper:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA+gAAAL4CAYAAADoAt3LAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nar8AAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7ff64943d4a8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "if isnotebook():\n",
    "    for i, platform in enumerate([\"AMD Tahiti 7970\", \"NVIDIA GTX 970\"]):\n",
    "\n",
    "        def get_speedups(df, platform):\n",
    "            \"\"\" get speedups for platform, aggregated by benchmark \"\"\"\n",
    "            # average by benchmark\n",
    "            series = df[df[\"Platform\"] == platform]\\\n",
    "                .groupby(['Benchmark'])['Speedup'].mean()\n",
    "            # average across all benchmarks\n",
    "            average = pd.Series(df[df['Platform'] == platform]['Speedup'].mean(), index=['Average'])\n",
    "            return series.append(average)\n",
    "\n",
    "        # Aggregate data\n",
    "        models, results = (Grewe, DeepTune), (grewe, deeptune)\n",
    "        dfs = [get_speedups(r, platform) for r in results]\n",
    "\n",
    "        speedups    = np.concatenate([dfs[i].values for i in range(len(dfs))])\n",
    "        benchmarks  = np.concatenate([dfs[i].keys().values for i in range(len(dfs))])\n",
    "        model_names = np.concatenate([np.array([m().__name__] * len(dfs[0])) for m in models])\n",
    "\n",
    "        df = pd.DataFrame([{\"Benchmark\": b, \"Speedup\": s - 1, \"Model\": m}\n",
    "                           for b, s, m in zip(benchmarks, speedups, model_names)])\n",
    "\n",
    "        # Plot\n",
    "        plt.subplot(2, 1, i + 1)\n",
    "        palette = sns.cubehelix_palette(2, rot=1.8, light=.85, dark=.35)\n",
    "        ax = sns.barplot(x=\"Benchmark\", y=\"Speedup\", hue=\"Model\", palette=palette, data=df)\n",
    "        c = chr(ord('a') + i)\n",
    "        plt.xlabel(f\"({c}) {platform}\", fontsize=10)\n",
    "        \n",
    "        # x axis\n",
    "        plt.setp(ax.get_xticklabels(), rotation=90)\n",
    "        plt.axvline(x=plt.xlim()[1] - 1, color=\"k\", lw=1, linestyle=\"--\")  # average line\n",
    "\n",
    "        # y axis\n",
    "        plt.ylim((-1, 3) if i else (-1, 9))\n",
    "        plt.ylabel(\"Speedup\", fontsize=10)\n",
    "\n",
    "        # legend\n",
    "        plt.legend(loc='upper left')\n",
    "        ax.get_legend().set_title(\"\")\n",
    "        ax.get_legend().draw_frame(True)\n",
    "\n",
    "        plt.axhline(y=0, color=\"k\", lw=.5)  # speedup line\n",
    "        ax.set_yticklabels([\"{:.1f}x\".format(i + 1) for i in ax.get_yticks()])\n",
    "    \n",
    "    viz.finalise(figsize=(8.5, 6.5))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 4.3. Comparison to State-of-the-art\n",
    "\n",
    "Here we directly compare performance of DeepTune against the Grewe et al. state-of-the-art model:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DeepTune outperforms Grewe et al. by 14%\n"
     ]
    }
   ],
   "source": [
    "performance_improvement = (deeptune[\"Speedup\"].mean() / grewe[\"Speedup\"].mean()) - 1\n",
    "\n",
    "print(f\"DeepTune outperforms Grewe et al. by {performance_improvement:.0%}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DeepTune matched or improved over state-of-the-art in 1214 of 1360 cases (89.3%)\n"
     ]
    }
   ],
   "source": [
    "num_cases = len(deeptune)\n",
    "num_better = sum(1 for d, g in zip(deeptune[\"Speedup\"], grewe[\"Speedup\"]) if d >= g)\n",
    "ratio_better = num_better / num_cases\n",
    "print(\"DeepTune matched or improved over state-of-the-art in \"\n",
    "      f\"{num_better} of {num_cases} cases ({ratio_better:.1%})\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 4.4. Training without Auxiliary Inputs\n",
    "\n",
    "Finally, we evaluate the role of auxiliary inputs on the quality of the learned model. We configure a DeepTune model without the two auxiliary inputs for workgroup and dataset size, and repeat the evaluation."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "class DeepTuneNoAux(HeterogemeousMappingModel):\n",
    "    __name__ = \"DeepTune wo. Auxiliary Inputs\"\n",
    "    __basename__ = \"deeptune-no-aux\"\n",
    "    \n",
    "    def init(self, seed: int):\n",
    "        from keras.layers import Input, Embedding, LSTM, Dense\n",
    "        from keras.layers.merge import Concatenate\n",
    "        from keras.layers.normalization import BatchNormalization\n",
    "        from keras.models import Model\n",
    "\n",
    "        np.random.seed(seed)\n",
    "        \n",
    "        # Language model. Takes as inputs source code sequences.\n",
    "        code_in = Input(shape=(1024,), dtype=\"int32\", name=\"code_in\")\n",
    "        x = Embedding(input_dim=atomizer.vocab_size + 1, input_length=1024, \n",
    "                      output_dim=64, name=\"embedding\")(code_in)\n",
    "        x = LSTM(64, implementation=1, return_sequences=True, name=\"lstm_1\")(x)\n",
    "        x = LSTM(64, implementation=1, name=\"lstm_2\")(x)\n",
    "            \n",
    "        # Heuristic model. Takes as inputs the language model,\n",
    "        #   outputs 1-hot encoded device mapping\n",
    "        x = BatchNormalization()(x)\n",
    "        x = Dense(32, activation=\"relu\")(x)\n",
    "        out = Dense(2, activation=\"sigmoid\")(x)\n",
    "\n",
    "        self.model = Model(inputs=[code_in], outputs=[out])\n",
    "        self.model.compile(\n",
    "            optimizer=\"adam\", metrics=['accuracy'],\n",
    "            loss=\"categorical_crossentropy\")\n",
    "        \n",
    "        return self\n",
    "\n",
    "    def save(self, outpath):\n",
    "        self.model.save(outpath)\n",
    "\n",
    "    def restore(self, inpath):\n",
    "        from keras.models import load_model\n",
    "        self.model = load_model(inpath)\n",
    "        \n",
    "    def train(self, **train):\n",
    "        self.model.fit(train[\"sequences\"], train[\"y_1hot\"],\n",
    "                       epochs=50, batch_size=64, verbose=train[\"verbose\"], shuffle=True)\n",
    "\n",
    "    def predict(self, **test):\n",
    "        p = np.array(self.model.predict(test[\"sequences\"], batch_size=64, verbose=test[\"verbose\"]))\n",
    "        indices = [np.argmax(x) for x in p]\n",
    "        return indices"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "code_in (InputLayer)         (None, 1024)              0         \n",
      "_________________________________________________________________\n",
      "embedding (Embedding)        (None, 1024, 64)          8256      \n",
      "_________________________________________________________________\n",
      "lstm_1 (LSTM)                (None, 1024, 64)          33024     \n",
      "_________________________________________________________________\n",
      "lstm_2 (LSTM)                (None, 64)                33024     \n",
      "_________________________________________________________________\n",
      "batch_normalization_2 (Batch (None, 64)                256       \n",
      "_________________________________________________________________\n",
      "dense_4 (Dense)              (None, 32)                2080      \n",
      "_________________________________________________________________\n",
      "dense_5 (Dense)              (None, 2)                 66        \n",
      "=================================================================\n",
      "Total params: 76,706.0\n",
      "Trainable params: 76,578.0\n",
      "Non-trainable params: 128.0\n",
      "_________________________________________________________________\n"
     ]
    },
    {
     "data": {
      "image/svg+xml": [
       "<svg height=\"553pt\" viewBox=\"0.00 0.00 399.00 553.00\" width=\"399pt\" xmlns=\"http://www.w3.org/2000/svg\" xmlns:xlink=\"http://www.w3.org/1999/xlink\">\n",
       "<g class=\"graph\" id=\"graph0\" transform=\"scale(1 1) rotate(0) translate(4 549)\">\n",
       "<title>G</title>\n",
       "<polygon fill=\"white\" points=\"-4,4 -4,-549 395,-549 395,4 -4,4\" stroke=\"none\"/>\n",
       "<!-- 140694076643760 -->\n",
       "<g class=\"node\" id=\"node1\"><title>140694076643760</title>\n",
       "<polygon fill=\"none\" points=\"60,-498.5 60,-544.5 331,-544.5 331,-498.5 60,-498.5\" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"123.5\" y=\"-517.8\">code_in: InputLayer</text>\n",
       "<polyline fill=\"none\" points=\"187,-498.5 187,-544.5 \" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"214.5\" y=\"-529.3\">input:</text>\n",
       "<polyline fill=\"none\" points=\"187,-521.5 242,-521.5 \" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"214.5\" y=\"-506.3\">output:</text>\n",
       "<polyline fill=\"none\" points=\"242,-498.5 242,-544.5 \" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"286.5\" y=\"-529.3\">(None, 1024)</text>\n",
       "<polyline fill=\"none\" points=\"242,-521.5 331,-521.5 \" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"286.5\" y=\"-506.3\">(None, 1024)</text>\n",
       "</g>\n",
       "<!-- 140694076641688 -->\n",
       "<g class=\"node\" id=\"node2\"><title>140694076641688</title>\n",
       "<polygon fill=\"none\" points=\"39,-415.5 39,-461.5 352,-461.5 352,-415.5 39,-415.5\" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"113\" y=\"-434.8\">embedding: Embedding</text>\n",
       "<polyline fill=\"none\" points=\"187,-415.5 187,-461.5 \" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"214.5\" y=\"-446.3\">input:</text>\n",
       "<polyline fill=\"none\" points=\"187,-438.5 242,-438.5 \" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"214.5\" y=\"-423.3\">output:</text>\n",
       "<polyline fill=\"none\" points=\"242,-415.5 242,-461.5 \" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"297\" y=\"-446.3\">(None, 1024)</text>\n",
       "<polyline fill=\"none\" points=\"242,-438.5 352,-438.5 \" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"297\" y=\"-423.3\">(None, 1024, 64)</text>\n",
       "</g>\n",
       "<!-- 140694076643760&#45;&gt;140694076641688 -->\n",
       "<g class=\"edge\" id=\"edge1\"><title>140694076643760-&gt;140694076641688</title>\n",
       "<path d=\"M195.5,-498.366C195.5,-490.152 195.5,-480.658 195.5,-471.725\" fill=\"none\" stroke=\"black\"/>\n",
       "<polygon fill=\"black\" points=\"199,-471.607 195.5,-461.607 192,-471.607 199,-471.607\" stroke=\"black\"/>\n",
       "</g>\n",
       "<!-- 140695758773048 -->\n",
       "<g class=\"node\" id=\"node3\"><title>140695758773048</title>\n",
       "<polygon fill=\"none\" points=\"64,-332.5 64,-378.5 327,-378.5 327,-332.5 64,-332.5\" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"113\" y=\"-351.8\">lstm_1: LSTM</text>\n",
       "<polyline fill=\"none\" points=\"162,-332.5 162,-378.5 \" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"189.5\" y=\"-363.3\">input:</text>\n",
       "<polyline fill=\"none\" points=\"162,-355.5 217,-355.5 \" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"189.5\" y=\"-340.3\">output:</text>\n",
       "<polyline fill=\"none\" points=\"217,-332.5 217,-378.5 \" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"272\" y=\"-363.3\">(None, 1024, 64)</text>\n",
       "<polyline fill=\"none\" points=\"217,-355.5 327,-355.5 \" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"272\" y=\"-340.3\">(None, 1024, 64)</text>\n",
       "</g>\n",
       "<!-- 140694076641688&#45;&gt;140695758773048 -->\n",
       "<g class=\"edge\" id=\"edge2\"><title>140694076641688-&gt;140695758773048</title>\n",
       "<path d=\"M195.5,-415.366C195.5,-407.152 195.5,-397.658 195.5,-388.725\" fill=\"none\" stroke=\"black\"/>\n",
       "<polygon fill=\"black\" points=\"199,-388.607 195.5,-378.607 192,-388.607 199,-388.607\" stroke=\"black\"/>\n",
       "</g>\n",
       "<!-- 140694076643424 -->\n",
       "<g class=\"node\" id=\"node4\"><title>140694076643424</title>\n",
       "<polygon fill=\"none\" points=\"64,-249.5 64,-295.5 327,-295.5 327,-249.5 64,-249.5\" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"113\" y=\"-268.8\">lstm_2: LSTM</text>\n",
       "<polyline fill=\"none\" points=\"162,-249.5 162,-295.5 \" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"189.5\" y=\"-280.3\">input:</text>\n",
       "<polyline fill=\"none\" points=\"162,-272.5 217,-272.5 \" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"189.5\" y=\"-257.3\">output:</text>\n",
       "<polyline fill=\"none\" points=\"217,-249.5 217,-295.5 \" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"272\" y=\"-280.3\">(None, 1024, 64)</text>\n",
       "<polyline fill=\"none\" points=\"217,-272.5 327,-272.5 \" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"272\" y=\"-257.3\">(None, 64)</text>\n",
       "</g>\n",
       "<!-- 140695758773048&#45;&gt;140694076643424 -->\n",
       "<g class=\"edge\" id=\"edge3\"><title>140695758773048-&gt;140694076643424</title>\n",
       "<path d=\"M195.5,-332.366C195.5,-324.152 195.5,-314.658 195.5,-305.725\" fill=\"none\" stroke=\"black\"/>\n",
       "<polygon fill=\"black\" points=\"199,-305.607 195.5,-295.607 192,-305.607 199,-305.607\" stroke=\"black\"/>\n",
       "</g>\n",
       "<!-- 140692935746560 -->\n",
       "<g class=\"node\" id=\"node5\"><title>140692935746560</title>\n",
       "<polygon fill=\"none\" points=\"0,-166.5 0,-212.5 391,-212.5 391,-166.5 0,-166.5\" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"130\" y=\"-185.8\">batch_normalization_2: BatchNormalization</text>\n",
       "<polyline fill=\"none\" points=\"260,-166.5 260,-212.5 \" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"287.5\" y=\"-197.3\">input:</text>\n",
       "<polyline fill=\"none\" points=\"260,-189.5 315,-189.5 \" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"287.5\" y=\"-174.3\">output:</text>\n",
       "<polyline fill=\"none\" points=\"315,-166.5 315,-212.5 \" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"353\" y=\"-197.3\">(None, 64)</text>\n",
       "<polyline fill=\"none\" points=\"315,-189.5 391,-189.5 \" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"353\" y=\"-174.3\">(None, 64)</text>\n",
       "</g>\n",
       "<!-- 140694076643424&#45;&gt;140692935746560 -->\n",
       "<g class=\"edge\" id=\"edge4\"><title>140694076643424-&gt;140692935746560</title>\n",
       "<path d=\"M195.5,-249.366C195.5,-241.152 195.5,-231.658 195.5,-222.725\" fill=\"none\" stroke=\"black\"/>\n",
       "<polygon fill=\"black\" points=\"199,-222.607 195.5,-212.607 192,-222.607 199,-222.607\" stroke=\"black\"/>\n",
       "</g>\n",
       "<!-- 140695748779536 -->\n",
       "<g class=\"node\" id=\"node6\"><title>140695748779536</title>\n",
       "<polygon fill=\"none\" points=\"79,-83.5 79,-129.5 312,-129.5 312,-83.5 79,-83.5\" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"130\" y=\"-102.8\">dense_4: Dense</text>\n",
       "<polyline fill=\"none\" points=\"181,-83.5 181,-129.5 \" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"208.5\" y=\"-114.3\">input:</text>\n",
       "<polyline fill=\"none\" points=\"181,-106.5 236,-106.5 \" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"208.5\" y=\"-91.3\">output:</text>\n",
       "<polyline fill=\"none\" points=\"236,-83.5 236,-129.5 \" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"274\" y=\"-114.3\">(None, 64)</text>\n",
       "<polyline fill=\"none\" points=\"236,-106.5 312,-106.5 \" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"274\" y=\"-91.3\">(None, 32)</text>\n",
       "</g>\n",
       "<!-- 140692935746560&#45;&gt;140695748779536 -->\n",
       "<g class=\"edge\" id=\"edge5\"><title>140692935746560-&gt;140695748779536</title>\n",
       "<path d=\"M195.5,-166.366C195.5,-158.152 195.5,-148.658 195.5,-139.725\" fill=\"none\" stroke=\"black\"/>\n",
       "<polygon fill=\"black\" points=\"199,-139.607 195.5,-129.607 192,-139.607 199,-139.607\" stroke=\"black\"/>\n",
       "</g>\n",
       "<!-- 140692666492688 -->\n",
       "<g class=\"node\" id=\"node7\"><title>140692666492688</title>\n",
       "<polygon fill=\"none\" points=\"79,-0.5 79,-46.5 312,-46.5 312,-0.5 79,-0.5\" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"130\" y=\"-19.8\">dense_5: Dense</text>\n",
       "<polyline fill=\"none\" points=\"181,-0.5 181,-46.5 \" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"208.5\" y=\"-31.3\">input:</text>\n",
       "<polyline fill=\"none\" points=\"181,-23.5 236,-23.5 \" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"208.5\" y=\"-8.3\">output:</text>\n",
       "<polyline fill=\"none\" points=\"236,-0.5 236,-46.5 \" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"274\" y=\"-31.3\">(None, 32)</text>\n",
       "<polyline fill=\"none\" points=\"236,-23.5 312,-23.5 \" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"274\" y=\"-8.3\">(None, 2)</text>\n",
       "</g>\n",
       "<!-- 140695748779536&#45;&gt;140692666492688 -->\n",
       "<g class=\"edge\" id=\"edge6\"><title>140695748779536-&gt;140692666492688</title>\n",
       "<path d=\"M195.5,-83.3664C195.5,-75.1516 195.5,-65.6579 195.5,-56.7252\" fill=\"none\" stroke=\"black\"/>\n",
       "<polygon fill=\"black\" points=\"199,-56.6068 195.5,-46.6068 192,-56.6069 199,-56.6068\" stroke=\"black\"/>\n",
       "</g>\n",
       "</g>\n",
       "</svg>"
      ],
      "text/plain": [
       "<IPython.core.display.SVG object>"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "deeptune_noaux_model = DeepTuneNoAux()\n",
    "deeptune_noaux_model.init(seed)\n",
    "deeptune_noaux_model.model.summary()\n",
    "\n",
    "img = None\n",
    "if isnotebook():\n",
    "    from keras.utils.vis_utils import model_to_dot\n",
    "    from IPython.display import SVG\n",
    "    img = SVG(model_to_dot(deeptune_noaux_model.model, show_shapes=True).create(prog='dot', format='svg'))\n",
    "img"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Evaluating DeepTune wo. auxiliary inputs ...\n",
      " 70% (14 of 20) |#####################         | Elapsed Time: 0:00:00 ETA:  0:00:00"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th>Correct?</th>\n",
       "      <th>Speedup</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Platform</th>\n",
       "      <th>Benchmark Suite</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th rowspan=\"7\" valign=\"top\">AMD Tahiti 7970</th>\n",
       "      <th>AMD SDK</th>\n",
       "      <td>0.562500</td>\n",
       "      <td>0.904486</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>NPB</th>\n",
       "      <td>0.645161</td>\n",
       "      <td>2.505454</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>NVIDIA SDK</th>\n",
       "      <td>0.666667</td>\n",
       "      <td>3.669208</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Parboil</th>\n",
       "      <td>0.789474</td>\n",
       "      <td>6.275650</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Polybench</th>\n",
       "      <td>0.703704</td>\n",
       "      <td>13.915320</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Rodinia</th>\n",
       "      <td>0.580645</td>\n",
       "      <td>4.445974</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>SHOC</th>\n",
       "      <td>0.833333</td>\n",
       "      <td>1.219326</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"7\" valign=\"top\">NVIDIA GTX 970</th>\n",
       "      <th>AMD SDK</th>\n",
       "      <td>0.562500</td>\n",
       "      <td>0.710539</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>NPB</th>\n",
       "      <td>0.812144</td>\n",
       "      <td>1.378846</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>NVIDIA SDK</th>\n",
       "      <td>0.583333</td>\n",
       "      <td>0.981517</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Parboil</th>\n",
       "      <td>0.789474</td>\n",
       "      <td>1.458785</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Polybench</th>\n",
       "      <td>0.740741</td>\n",
       "      <td>1.151140</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Rodinia</th>\n",
       "      <td>0.580645</td>\n",
       "      <td>1.502166</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>SHOC</th>\n",
       "      <td>0.708333</td>\n",
       "      <td>1.833715</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                 Correct?    Speedup\n",
       "Platform        Benchmark Suite                     \n",
       "AMD Tahiti 7970 AMD SDK          0.562500   0.904486\n",
       "                NPB              0.645161   2.505454\n",
       "                NVIDIA SDK       0.666667   3.669208\n",
       "                Parboil          0.789474   6.275650\n",
       "                Polybench        0.703704  13.915320\n",
       "                Rodinia          0.580645   4.445974\n",
       "                SHOC             0.833333   1.219326\n",
       "NVIDIA GTX 970  AMD SDK          0.562500   0.710539\n",
       "                NPB              0.812144   1.378846\n",
       "                NVIDIA SDK       0.583333   0.981517\n",
       "                Parboil          0.789474   1.458785\n",
       "                Polybench        0.740741   1.151140\n",
       "                Rodinia          0.580645   1.502166\n",
       "                SHOC             0.708333   1.833715"
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "print(\"Evaluating DeepTune wo. auxiliary inputs ...\", file=sys.stderr)\n",
    "deeptune_noaux = evaluate(DeepTuneNoAux())\n",
    "deeptune_noaux.groupby(['Platform', 'Benchmark Suite'])['Platform', 'Correct?', 'Speedup'].mean()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Analysis of results. In the paper, we report that removing auxiliary inputs leads to a **12%** reduction in model accuracy:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training without auxiliary inputs yields a 12% reduction in DeepTune's acccuracy, and a 7% reduction in performance.\n"
     ]
    }
   ],
   "source": [
    "speedup_loss = 1 - (deeptune_noaux['Speedup'].mean() / deeptune['Speedup'].mean())\n",
    "acc_loss = 1 - (deeptune_noaux['Correct?'].mean() / deeptune['Correct?'].mean())\n",
    "\n",
    "print(f\"Training without auxiliary inputs yields a \"\n",
    "      f\"{acc_loss:.0%} reduction in DeepTune's acccuracy, and a \"\n",
    "      f\"{speedup_loss:.0%} reduction in performance.\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 5. Seeding Transfer Learning\n",
    "\n",
    "Now we train a DeepTune model which we will re-use in Case Study B by transferring the learned weights: "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Seeding transfer learning ... done.\n"
     ]
    }
   ],
   "source": [
    "print(f\"Seeding transfer learning ... \", end=\"\")\n",
    "\n",
    "model_path = f\"../data/case-study-b/case-study-a.model\"\n",
    "weights_path = f\"../data/case-study-b/case-study-a-weights.h5\"\n",
    "\n",
    "if not fs.exists(weights_path):\n",
    "    deeptune_model = DeepTune()\n",
    "\n",
    "    if fs.exists(model_path):\n",
    "        deeptune_model = deeptune_model.restore(model_path)\n",
    "    else:\n",
    "        df = pd.read_csv(f\"../data/case-study-a/cgo17-amd.csv\")\n",
    "\n",
    "        sequences = encode_srcs(df[\"src\"].values)\n",
    "\n",
    "        features = grewe_features(df)\n",
    "        aux_in = auxiliary_inputs(df)\n",
    "\n",
    "        y = np.array([1 if x == \"GPU\" else 0 for x in df[\"oracle\"].values])\n",
    "        y_1hot = encode_1hot(y)\n",
    "\n",
    "        deeptune_model.init(seed=seed)\n",
    "        deeptune_model.train(df=df,\n",
    "                             features=features,\n",
    "                             aux_in=aux_in,\n",
    "                             sequences=sequences,\n",
    "                             y=y,\n",
    "                             y_1hot=y_1hot,\n",
    "                             verbose=True)\n",
    "        deeptune_model.save(model_path)\n",
    "\n",
    "    deeptune_model.model.save_weights(weights_path)\n",
    "\n",
    "print(\"done.\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "See '*Case Study B.ipynb*' for experiments using transfer learning.\n",
    "\n",
    "End of experiments."
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "e2e-dl",
   "language": "python",
   "name": "e2e-dl"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}